[{"data":1,"prerenderedAt":6253},["ShallowReactive",2],{"/":3},[4,265,498,923,3228,3420,3852,4037,4200,4371,4708,5010,5102,5425,5474,5541,5597,5775,6140],{"id":5,"title":6,"author":7,"body":8,"created":242,"description":243,"extension":244,"headerImage":245,"image":246,"lastUpdated":242,"meta":248,"navigation":260,"path":261,"published":260,"seo":262,"stem":263,"__hash__":264},"news/news/build-anything-that-needs-a-human.md","Build Anything That Needs a Human with Effect AI","Rochelle",{"type":9,"value":10,"toc":233},"minimark",[11,15,18,21,24,29,32,72,74,78,88,91,124,127,129,133,136,142,148,150,154,157,163,169,175,182,184,188,206,212,223,231],[12,13,14],"p",{},"Most software can handle the routine cases well. The hard part is the moment a product reaches something it can't decide on its own: whether a design is clear, whether an answer is right, whether a piece of content should go out. That moment usually needs a person.",[12,16,17],{},"Effect AI is built so any app can reach that person on demand. A product sends a task, a qualified human completes it, and the answer comes back as data the product can use. The same building blocks work for a new standalone app or for a human step added to a service that already exists.",[12,19,20],{},"Below is how those pieces fit together, a real example of an app built on them, and what else they make possible.",[22,23],"hr",{},[25,26,28],"h2",{"id":27},"the-building-blocks","The building blocks",[12,30,31],{},"Every human step on Effect is built from the same five parts.",[33,34,35,48,54,60,66],"ul",{},[36,37,38,42,43,47],"li",{},[39,40,41],"strong",{},"Templates define what the worker sees."," A template is a single HTML file with placeholders like ",[44,45,46],"code",{},"${image_url}"," where each task's data goes. It can show images, audio, video, text, or forms, and it sends back an answer in whatever shape you choose.",[36,49,50,53],{},[39,51,52],{},"Capabilities define who can see it."," Workers earn capabilities by passing tests, such as English Language, Music Transcription, or Design Review. Tasks reach only workers who hold the right ones.",[36,55,56,59],{},[39,57,58],{},"Validation keeps answers usable."," Templates check that a response is complete before it can be submitted. Jobs can also send the same item to several independent workers, and self-serve pipelines that pass a labeling job to a separate validation job are coming to the API.",[36,61,62,65],{},[39,63,64],{},"The Effect API connects it all."," Your backend, app, or agent creates jobs, checks their progress, and collects results, the same way it calls any other service.",[36,67,68,71],{},[39,69,70],{},"Results come back as structured data."," Each answer returns paired with the input it came from, along with the worker and the time it was submitted. Worker payments settle on-chain in EFFECT on Solana.",[22,73],{},[25,75,77],{"id":76},"a-real-example-vibecheck","A real example: VibeCheck",[12,79,80,87],{},[81,82,86],"a",{"href":83,"rel":84},"https://www.canva.com/apps/AAHAAEuEYQY/vibecheck",[85],"nofollow","VibeCheck"," is Effect's app in the Canva Apps Marketplace. It lets designers test a design with real people before it goes live, through a Clarity Check, a Clickability Check, or a side-by-side comparison of two versions.",[12,89,90],{},"The Canva side is a small interface. Everything else runs on the building blocks above:",[92,93,94,100,106,112,118],"ol",{},[36,95,96,99],{},[39,97,98],{},"The design becomes data."," VibeCheck exports a static image of the design and pairs it with the creator's purpose, audience, and goal.",[36,101,102,105],{},[39,103,104],{},"The data fills a template."," Each check is its own template. Clarity asks for a 1 to 10 rating, Clickability flashes the design for a few seconds and asks for a yes or no, and Compare Versions asks for a pick between two designs.",[36,107,108,111],{},[39,109,110],{},"Capabilities route the task."," Design checks reach only workers who have qualified for design review.",[36,113,114,117],{},[39,115,116],{},"People do the part only people can do."," A worker reacts to the design and writes down why. The template won't submit until the answer is complete.",[36,119,120,123],{},[39,121,122],{},"Results come back to Canva."," Ratings, picks, and written reasoning return as structured data, and VibeCheck turns them into feedback inside the editor.",[12,125,126],{},"None of these steps were built just for Canva. VibeCheck connects existing Effect features to a place where designers already work.",[22,128],{},[25,130,132],{"id":131},"two-ways-to-build","Two ways to build",[12,134,135],{},"The pattern behind VibeCheck isn't tied to Canva. It works anywhere an interface can collect input and call an API.",[12,137,138,141],{},[39,139,140],{},"As a standalone app."," VibeCheck lives inside Canva, but the same idea could be its own web or mobile app, a plugin for another design tool, or a browser extension. The app handles the user experience. Effect handles finding qualified people, showing them the task, and returning their answers.",[12,143,144,147],{},[39,145,146],{},"As a step inside an existing service."," You don't need to build a new product to add human judgment. A support platform can send low-confidence replies to people for rating. A marketplace can have new listings checked before they publish. A content pipeline can route flagged items to human review. Each one is a single API call at the point where the service needs a person, and the rest of the system keeps running as it does today.",[22,149],{},[25,151,153],{"id":152},"what-you-could-build","What you could build",[12,155,156],{},"If a task can be shown on a screen and answered by a person, it can run on Effect. A few patterns come up again and again.",[12,158,159,162],{},[39,160,161],{},"A person in the loop, on demand."," Some steps need human sign-off: approving a payment, confirming a fact, checking that a translation sounds natural, or deciding whether content should be published. Send only those steps to Effect and automate everything else.",[12,164,165,168],{},[39,166,167],{},"People and AI working together."," A model drafts and people correct. AI pre-labels images and workers fix what it got wrong. A chatbot flags its low-confidence answers for human rating, and those ratings become training data. Each side handles the part it does best.",[12,170,171,174],{},[39,172,173],{},"Research and data collection."," Surveys, preference studies, voice recordings, and labeled datasets all fit the same model, with qualified workers and independent answers built in.",[12,176,177,178],{},"For a closer look at where a person belongs in an AI pipeline, see ",[81,179,181],{"href":180},"/news/when-should-a-human-step-in","When Should a Human Step In?",[22,183],{},[25,185,187],{"id":186},"get-involved","Get Involved",[12,189,190,193,194,199,200,205],{},[39,191,192],{},"Developers & Researchers:"," ",[81,195,198],{"href":196,"rel":197},"https://tally.so/r/xXgAyr",[85],"Join API early access"," for an API key and test credits, then build your first template in the ",[81,201,204],{"href":202,"rel":203},"https://api.effect.ai/",[85],"API console",".",[12,207,208,211],{},[39,209,210],{},"Organizations:"," Partner with Effect AI to add human judgment to an existing product or build a new one around it.",[12,213,214,193,217,222],{},[39,215,216],{},"Contributors:",[81,218,221],{"href":219,"rel":220},"https://tally.so/r/0Q1lpN",[85],"Sign up"," for the next alpha phase and earn $EFFECT for real tasks.",[12,224,225,226,205],{},"Have an idea for an app? Share it on ",[81,227,230],{"href":228,"rel":229},"https://discord.com/invite/effectnetwork",[85],"Discord",[22,232],{},{"title":234,"searchDepth":235,"depth":235,"links":236},"",2,[237,238,239,240,241],{"id":27,"depth":235,"text":28},{"id":76,"depth":235,"text":77},{"id":131,"depth":235,"text":132},{"id":152,"depth":235,"text":153},{"id":186,"depth":235,"text":187},"2026-10-21","Effect AI lets any app reach a qualified human on demand. Here are the building blocks, a real example in VibeCheck, and what else you could build.","md",null,{"src":247},"/img/news/news-header-workers-building.png",{"head":249},{"meta":250},[251,254,257],{"name":252,"content":253},"keywords","Effect AI, Effect API, human in the loop, task templates, capabilities, VibeCheck, Canva, microtasks, EFFECT token, Solana, developers",{"name":255,"content":256},"author","Rochelle - Effect.AI",{"name":258,"content":259},"copyright","© 2026 Effect.AI",true,"/news/build-anything-that-needs-a-human",{"title":6,"description":243},"news/build-anything-that-needs-a-human","LwoMTqPQh_sVLWl77AqS2C8tRUSqkYQzSjSara06xTc",{"id":266,"title":267,"author":7,"body":268,"created":242,"description":484,"extension":244,"headerImage":485,"image":486,"lastUpdated":242,"meta":487,"navigation":260,"path":494,"published":260,"seo":495,"stem":496,"__hash__":497},"news/news/vibecheck-on-canva-app-marketplace.md","VibeCheck Is Now on the Canva App Marketplace",{"type":9,"value":269,"toc":478},[270,273,283,286,296,298,302,305,311,317,323,326,329,331,335,338,364,367,373,375,379,382,432,434,436,439,449,461,469,476],[12,271,272],{},"Every designer knows the moment right before a design goes live. It looks right to you. It looked right to the team. But will the people it was made for understand it, notice the button, or prefer it over the other version?",[12,274,275,277,278,282],{},[39,276,86],{},", Effect AI's new app, is now available in the ",[81,279,281],{"href":83,"rel":280},[85],"Canva Apps Marketplace",". It sends a design to real people on the Effect AI network and brings their feedback back into Canva, so creators can test a design before it ships without leaving the editor.",[12,284,285],{},"Below is what VibeCheck does, who answers the questions, and how to start using it.",[12,287,288],{},[289,290],"img",{"alt":291,"className":292,"src":295},"The VibeCheck listing in the Canva Apps Marketplace",[293,294],"screenshot","left","/img/news/vibecheck/listing-rounded.png",[22,297],{},[25,299,301],{"id":300},"what-vibecheck-does","What VibeCheck Does",[12,303,304],{},"VibeCheck offers three checks. Each one asks reviewers a different question, because a design can grab attention without being easy to understand, and the reverse is also true.",[12,306,307,310],{},[39,308,309],{},"Clickability Check: would people stop scrolling or click?"," The design appears on screen for a few seconds, the way it would in a busy feed, then disappears. Reviewers answer yes or no and explain what caught their eye or what made the design blend in.",[12,312,313,316],{},[39,314,315],{},"Clarity Check: how easy is this design to understand?"," Reviewers rate the design from 1 (not clear at all) to 10 (very clear). They consider whether the main message comes through in about five seconds, whether the layout leads the eye to the most important information, and whether anything distracts from it.",[12,318,319,322],{},[39,320,321],{},"Compare Versions: which version works better?"," Reviewers see two designs side by side and pick the one that better fits its purpose and audience. They judge overall effectiveness rather than personal taste, and they note what made the difference, such as the headline, contrast, layout, imagery, readability, or call to action.",[12,324,325],{},"Each check can include the design's purpose, its target audience, and its main goal. Reviewers see that context next to the design and use it in their answers, since a bold sale banner and a quiet event invitation shouldn't be judged by the same standard.",[12,327,328],{},"VibeCheck sends reviewers a static image of the design, not the editable file. The app only needs permission to read the design, and nothing in it is changed.",[22,330],{},[25,332,334],{"id":333},"who-gives-the-feedback","Who Gives the Feedback",[12,336,337],{},"The answers come from people on the Effect AI network, the same network that has completed hundreds of thousands of data and labeling tasks during the Effect Alpha. Reviewers work from guidance written specifically for these three checks, and a few rules shape what creators get back:",[33,339,340,346,352,358],{},[36,341,342,345],{},[39,343,344],{},"Qualified reviewers."," Design tasks go only to workers who have passed the Design Review capability test.",[36,347,348,351],{},[39,349,350],{},"Reasoning on every task."," A score or a pick on its own isn't enough. Reviewers explain their decision, and that explanation goes directly to the creator.",[36,353,354,357],{},[39,355,356],{},"A professional standard."," Feedback has to be specific, clearly written, and correctly spelled. \"Looks good\" doesn't meet the bar. A note that the headline blends into the background and hides the offer does.",[36,359,360,363],{},[39,361,362],{},"Real first reactions."," Reviewers are asked how they would actually respond to the design, not how they would redesign it.",[12,365,366],{},"Each result is tied to the task, the worker, and the time it was submitted, and worker payments settle in EFFECT on Solana.",[12,368,369,370,372],{},"This is the human-in-the-loop idea from the recent post ",[81,371,181],{"href":180},", applied to design. Taste, tone, and first impressions have no answer key. People are the ground truth.",[22,374],{},[25,376,378],{"id":377},"getting-started","Getting Started",[12,380,381],{},"VibeCheck is free to install on any Canva plan. To try it:",[92,383,384,387,401,412,422],{},[36,385,386],{},"Sign up or log in to Canva.",[36,388,389,390,205,392,395],{},"Open the Apps Marketplace and search for ",[39,391,86],{},[393,394],"br",{},[289,396],{"alt":397,"className":398,"src":400},"Searching for VibeCheck in the Canva Apps Marketplace",[293,294,399],"small","/img/news/vibecheck/search.png",[36,402,403,404,406],{},"Open the design you want to test and wait for VibeCheck to load in the editor. Then pick a check.",[393,405],{},[289,407],{"alt":408,"className":409,"src":411},"VibeCheck open in the Canva editor, with Clarity Check, Clickability Check, and Compare Versions to choose from",[293,294,410],"medium","/img/news/vibecheck/checks.png",[36,413,414,415,417],{},"Add the design's purpose, audience, and goal, choose how many testers you want, and send it out.",[393,416],{},[289,418],{"alt":419,"className":420,"src":421},"The Clarity Check form in VibeCheck, asking what the design is for, who the target audience is, the main goal, and the number of testers",[293,294,410],"/img/news/vibecheck/clarity-form.png",[36,423,424,425,427],{},"Results, including each reviewer's written reasoning, return to the app as reviewers complete them.",[393,426],{},[289,428],{"alt":429,"className":430,"src":431},"Compare Versions results in VibeCheck: Version A wins with 13 of 20 testers, with each tester's written comment below",[293,294,399],"/img/news/vibecheck/results.png",[22,433],{},[25,435,187],{"id":186},[12,437,438],{},"VibeCheck is the first Effect AI app built for people who make things rather than people who write code. It puts the Effect network where designers already work.",[12,440,441,193,444,448],{},[39,442,443],{},"Designers and marketers:",[81,445,447],{"href":83,"rel":446},[85],"Install VibeCheck"," from the Canva Apps Marketplace and run a test on your next design.",[12,450,451,193,453,456,457,460],{},[39,452,216],{},[81,454,221],{"href":219,"rel":455},[85]," for the ",[39,458,459],{},"next alpha phase"," and start earning $EFFECT for real tasks, including design reviews.",[12,462,463,193,465,468],{},[39,464,192],{},[81,466,198],{"href":196,"rel":467},[85]," to build your own human-in-the-loop workflows on Effect.",[12,470,471,472,205],{},"Questions or feedback on the app? ",[81,473,475],{"href":228,"rel":474},[85],"Join the conversation on Discord",[22,477],{},{"title":234,"searchDepth":235,"depth":235,"links":479},[480,481,482,483],{"id":300,"depth":235,"text":301},{"id":333,"depth":235,"text":334},{"id":377,"depth":235,"text":378},{"id":186,"depth":235,"text":187},"VibeCheck sends your Canva design to real people on the Effect AI network and brings their written feedback back into the editor, so you can test a design before it ships.","/img/news/news-header-task-templates-2.png",{"src":295},{"head":488},{"meta":489},[490,492,493],{"name":252,"content":491},"VibeCheck, Canva, Canva Apps Marketplace, design feedback, design testing, human feedback, A/B testing, clarity check, clickability, human in the loop, Effect AI, EFFECT token, Solana",{"name":255,"content":256},{"name":258,"content":259},"/news/vibecheck-on-canva-app-marketplace",{"title":267,"description":484},"news/vibecheck-on-canva-app-marketplace","rfyfiDJ4PU5SiwHocya7m1VWhJJRqfJJ3tgFP73hgSQ",{"id":499,"title":500,"author":501,"body":502,"created":908,"description":909,"extension":244,"headerImage":910,"image":911,"lastUpdated":908,"meta":912,"navigation":260,"path":180,"published":260,"seo":920,"stem":921,"__hash__":922},"news/news/when-should-a-human-step-in.md","When Should a Human Step In? 7 Signals for Human-in-the-Loop AI","Miguel",{"type":9,"value":503,"toc":891},[504,507,513,516,518,522,525,531,536,539,551,555,558,565,569,572,579,583,586,593,597,600,607,611,614,621,625,628,635,644,646,650,653,753,756,758,762,765,771,777,783,794,800,802,806,812,818,824,830,832,836,856,859,861,863,866,876,884,889],[12,505,506],{},"\"Will AI replace humans?\" is the wrong question. So is \"Does AI still need us?\" Neither one helps you ship a better system.",[12,508,509,510],{},"Here's the truth: every AI product already runs on people. People wrote the training data. People labeled the edge cases, rated the responses, and built the benchmarks the model was graded on. So the real question is sharper: ",[39,511,512],{},"where does a human need to be in the loop, and where are they just in the way?",[12,514,515],{},"Below is a seven-point test that answers it, a map of where people fit across the AI lifecycle, and how to bring them in on Effect.",[22,517],{},[25,519,521],{"id":520},"the-human-in-the-loop-test","The Human-in-the-Loop Test",[12,523,524],{},"The rule is simple. If even one of these signals fits a step in your pipeline, put a person there. If none fit, automate it.",[12,526,527],{},[289,528],{"alt":529,"src":530},"The Human-in-the-Loop Test: seven signals that a human should step in","/img/news/step-in-test.png",[532,533,535],"h3",{"id":534},"_1-there-is-no-ground-truth-yet","1. There is no ground truth yet",[12,537,538],{},"A model can only learn labels someone has already defined. A new product line, a new defect type, a new category of content: they all start from zero. Someone has to create the first trustworthy examples, and those examples become the yardstick for every label after them. Get them wrong, and the error spreads into everything built on top.",[540,541,543],"div",{"style":542},"background:#f1f1f4;border-radius:6px;padding:0.1em 1.4em;margin:1.5em 0",[12,544,545,550],{},[546,547,549],"span",{"style":548},"display:block;width:fit-content;background:#E2FF03;color:#000;border-radius:2px;padding:2px 8px;margin-bottom:0.6em;font-size:0.65em;font-weight:600;letter-spacing:0.08em;text-transform:uppercase","Try it on Effect","\nPost a CSV job with the Bounding Box Labeler template to build your first labeled set. Then run Bounding Box Validation so a second worker checks every box.",[532,552,554],{"id":553},"_2-the-answer-is-subjective","2. The answer is subjective",[12,556,557],{},"Which reply is more helpful? Is this joke funny? Is this design clear? There's no answer key for taste, tone, or preference. Human judgment is the ground truth, which is exactly why RLHF preference data comes from people. A model can copy what people liked yesterday. It can't predict what your audience will love tomorrow.",[540,559,560],{"style":542},[12,561,562,564],{},[546,563,549],{"style":548},"\nA constant job sends the same comparison to many workers. Turn on \"one task per worker\" and every vote comes from a different person.",[532,566,568],{"id":567},"_3-the-model-isnt-sure","3. The model isn't sure",[12,570,571],{},"Low confidence scores and models that disagree are the loudest alarms you have. They point to the exact cases where the model is guessing. Route only those to people and every human minute counts. Skip that routing and you're stuck with two bad options: review everything and pay for it, or review nothing and ship the guesses.",[540,573,574],{"style":542},[12,575,576,578],{},[546,577,549],{"style":548},"\nYour backend or agent calls the Effect API for flagged items only and collects the answers when they're ready. The easy 90% never costs you a human task.",[532,580,582],{"id":581},"_4-a-mistake-is-expensive-or-permanent","4. A mistake is expensive or permanent",[12,584,585],{},"Sending a payment. Publishing content. Removing a post. Giving medical or legal information. When an error can't be undone, a confident model isn't enough. Someone should sign off. A quick human review costs far less than one wrong payment or one post taken down by mistake.",[540,587,588],{"style":542},[12,589,590,592],{},[546,591,549],{"style":548},"\nBuild an approval step as a custom template. Every result records the task, the worker, and the time, so you always know who approved what.",[532,594,596],{"id":595},"_5-the-world-has-changed","5. The world has changed",[12,598,599],{},"Models freeze the day training ends. The world doesn't. New slang, new products, and new events turn into quiet errors the model can't detect on its own. A person spots a new brand name or meme instantly. That makes a small human sample the cheapest early-warning system you can run.",[540,601,602],{"style":542},[12,603,604,606],{},[546,605,549],{"style":548},"\nRun a small, recurring sample of live outputs past human reviewers. The estimate endpoint shows the exact cost before anything is charged, so a weekly spot-check stays a predictable line item.",[532,608,610],{"id":609},"_6-it-needs-local-cultural-or-physical-presence","6. It needs local, cultural, or physical presence",[12,612,613],{},"Regional dialects, under-represented languages, local knowledge, and real human voices can't be faked convincingly. Neither can anything that requires actually being somewhere. Someone from Lisbon and someone from São Paulo say the same sentence differently. Train a voice model on only one, and it stumbles on the other.",[540,615,616],{"style":542},[12,617,618,620],{},[546,619,549],{"style":548},"\nOur Common Voice templates have workers write and validate sentences for open voice datasets, and our speech templates have them record and verify audio. Capabilities like Microphone Access and English Language make sure tasks only reach workers who can do them.",[532,622,624],{"id":623},"_7-you-are-grading-the-ai-itself","7. You are grading the AI itself",[12,626,627],{},"Evaluation is where humans matter most, and where they get skipped most. Using one model to judge another is useful, but without human calibration it's grading its own homework. The judge often shares the blind spots of the model it's grading, so the same mistakes sail past both.",[540,629,630],{"style":542},[12,631,632,634],{},[546,633,549],{"style":548},"\nHave one group of workers write hard questions and another verify the answers. Then have people rate a sample of outputs and compare their scores with your automated judge. When they disagree, trust the people and fix the judge.",[12,636,637,638,643],{},"One more reason to keep people involved: research shows that models trained on ",[81,639,642],{"href":640,"rel":641},"https://www.nature.com/articles/s41586-024-07566-y",[85],"recursively generated data degrade over time",". Fresh human data keeps them grounded in reality.",[22,645],{},[25,647,649],{"id":648},"where-people-fit-across-the-ai-lifecycle","Where people fit across the AI lifecycle",[12,651,652],{},"Human involvement isn't a fixed dial. It rises and falls with each stage:",[654,655,656,672],"table",{},[657,658,659],"thead",{},[660,661,662,666,669],"tr",{},[663,664,665],"th",{},"Stage",[663,667,668],{},"Human involvement",[663,670,671],{},"What people do",[673,674,675,689,701,714,727,740],"tbody",{},[660,676,677,683,686],{},[678,679,680],"td",{},[39,681,682],{},"Data creation",[678,684,685],{},"Heavy",[678,687,688],{},"Write, record, transcribe, and collect the raw material.",[660,690,691,696,698],{},[678,692,693],{},[39,694,695],{},"Evaluation",[678,697,685],{},[678,699,700],{},"Write benchmarks and keep automated judges honest.",[660,702,703,708,711],{},[678,704,705],{},[39,706,707],{},"Labeling",[678,709,710],{},"Medium",[678,712,713],{},"Fix what AI pre-labels get wrong.",[660,715,716,721,724],{},[678,717,718],{},[39,719,720],{},"Training",[678,722,723],{},"Targeted",[678,725,726],{},"Supply preference judgments and corrections.",[660,728,729,734,737],{},[678,730,731],{},[39,732,733],{},"Deployment",[678,735,736],{},"Light but critical",[678,738,739],{},"Approve the actions that can't be undone.",[660,741,742,747,750],{},[678,743,744],{},[39,745,746],{},"Monitoring",[678,748,749],{},"Light and regular",[678,751,752],{},"Spot-check for drift.",[12,754,755],{},"Teams that understand this curve put human effort where it changes outcomes, not everywhere and not nowhere.",[22,757],{},[25,759,761],{"id":760},"nail-the-handoff","Nail the Handoff",[12,763,764],{},"Bringing a human in is only half the job. The other half is designing the handoff so their work is accurate and fair.",[12,766,767,770],{},[39,768,769],{},"Pick the right people."," Capabilities control who sees a task, and workers earn them by passing a test, like Design Review for design feedback or Music Transcription for lyric work.",[12,772,773,776],{},[39,774,775],{},"Never let people check their own work."," When we chain a labeling job into a validation job, the validation task never goes back to the original worker. We already run these pipelines on our own datasets, and self-serve pipelines through the API are coming soon.",[12,778,779,782],{},[39,780,781],{},"Ask more than one person."," Independent answers to the same task reveal consensus and expose outliers.",[12,784,785,788,789,793],{},[39,786,787],{},"Shape the task to the question."," Every template is a small web page, so a task can look like whatever your problem needs. Our ",[81,790,792],{"href":791},"/news/build-your-own-task-template","template guide"," shows you how to build one.",[12,795,796,799],{},[39,797,798],{},"Respect the worker's time."," Every task comes with clear instructions and its own time limit, and workers are paid in EFFECT directly on Solana.",[22,801],{},[25,803,805],{"id":804},"see-it-in-action","See It in Action",[12,807,808,811],{},[39,809,810],{},"A support bot"," handles most tickets on its own. Replies below a confidence threshold go to Effect for human rating, and those ratings become next month's fine-tuning data.",[12,813,814,817],{},[39,815,816],{},"A design team"," in Canva runs a clickability check and an A/B comparison before launch, without ever leaving the editor. Only workers who passed the Design Review test see the tasks.",[12,819,820,823],{},[39,821,822],{},"A music platform"," lets AI draft lyric timestamps. Qualified workers correct them, and a separate group validates the corrections.",[12,825,826,829],{},[39,827,828],{},"A research agent"," hits a claim it can't verify. It posts a one-question task through the API and keeps going once a person confirms it.",[22,831],{},[25,833,835],{"id":834},"why-effect","Why Effect",[33,837,838,844,850],{},[36,839,840,843],{},[39,841,842],{},"On demand and programmable."," No vendor contracts. No minimum volumes. Your code, or your agent, calls people the same way it calls any other service.",[36,845,846,849],{},[39,847,848],{},"Quality built in."," Qualified workers, independent validators, and redundant answers come standard. You don't have to assemble them yourself.",[36,851,852,855],{},[39,853,854],{},"A clear record."," Every result comes back with the task, the worker, and the submission time, and payments settle on-chain.",[12,857,858],{},"The playbook: automate what the model handles confidently, escalate what it doesn't, and leave the questions only humans can answer to humans. Effect makes that last part easy.",[22,860],{},[25,862,187],{"id":186},[12,864,865],{},"Ready to help build the next phase of Effect AI?",[12,867,868,193,870,456,873,875],{},[39,869,216],{},[81,871,221],{"href":219,"rel":872},[85],[39,874,459],{}," and start earning $EFFECT for real tasks.",[12,877,878,193,880,883],{},[39,879,192],{},[81,881,198],{"href":196,"rel":882},[85]," for test credits, or team up with us to build and validate high-quality datasets for your models and research.",[12,885,886,888],{},[39,887,210],{}," Partner with Effect AI on responsible, mission-aligned data initiatives, from AI development to social impact.",[22,890],{},{"title":234,"searchDepth":235,"depth":235,"links":892},[893,903,904,905,906,907],{"id":520,"depth":235,"text":521,"children":894},[895,897,898,899,900,901,902],{"id":534,"depth":896,"text":535},3,{"id":553,"depth":896,"text":554},{"id":567,"depth":896,"text":568},{"id":581,"depth":896,"text":582},{"id":595,"depth":896,"text":596},{"id":609,"depth":896,"text":610},{"id":623,"depth":896,"text":624},{"id":648,"depth":235,"text":649},{"id":760,"depth":235,"text":761},{"id":804,"depth":235,"text":805},{"id":834,"depth":235,"text":835},{"id":186,"depth":235,"text":187},"2026-10-13","Seven signals that show exactly where a human belongs in your AI pipeline, from data creation to evaluation, and how to plug them in on Effect AI.","/img/news/news-header-ai-collaboration.png",{"src":910},{"head":913},{"meta":914},[915,917,919],{"name":252,"content":916},"human in the loop, HITL, AI training data, data labeling, RLHF, preference data, AI evaluation, model collapse, AI agents, Effect AI, Effect API, EFFECT token, Solana, data validation, AI oversight",{"name":255,"content":918},"Miguel - Effect.AI",{"name":258,"content":259},{"title":500,"description":909},"news/when-should-a-human-step-in","txG92TLr7NZMH0rgX0LDfxvLnTBUkk4ABklrASxj9ZA",{"id":924,"title":925,"author":501,"body":926,"created":3214,"description":3215,"extension":244,"headerImage":3216,"image":3217,"lastUpdated":3214,"meta":3218,"navigation":260,"path":791,"published":260,"seo":3225,"stem":3226,"__hash__":3227},"news/news/build-your-own-task-template.md","Developer Guide: Build Your Own Task Template on the Effect API",{"type":9,"value":927,"toc":3200},[928,935,938,948,950,954,957,992,994,998,1001,1007,1010,1014,1017,1086,1103,1107,1110,1113,1636,1647,1651,1654,1707,1718,1721,1926,1929,1933,1936,1939,1945,2005,2016,2026,2030,2033,2787,2789,2793,2799,2814,2821,2839,2846,2855,2862,2870,2873,2880,2887,2889,2893,2908,2939,2946,2952,3001,3008,3016,3018,3022,3029,3159,3168,3174,3176,3178,3181,3190,3195,3197],[12,929,930,931,934],{},"Every task on Effect is a small web page. A worker opens it, answers one question, and presses submit. That page is a ",[39,932,933],{},"template",", and with the Effect API you can write your own.",[12,936,937],{},"This guide covers what a template is, walks through a complete example, and shows how to test it and run it as a job. Full docs are on the way. Until then, start here.",[12,939,940,943,944,947],{},[39,941,942],{},"You'll need:"," an API key from the ",[81,945,204],{"href":202,"rel":946},[85]," and some basic HTML.",[22,949],{},[25,951,953],{"id":952},"how-it-works","How it works",[12,955,956],{},"A template is a single HTML file. It can use any HTML, CSS and JavaScript you like, including libraries from a CDN.",[92,958,959,968,974,980,986],{},[36,960,961,964,965,967],{},[39,962,963],{},"You write the page once",", with placeholders like ",[44,966,46],{}," wherever each task's data should go.",[36,969,970,973],{},[39,971,972],{},"You upload a CSV."," Each column matches a placeholder, and each row becomes a task.",[36,975,976,979],{},[39,977,978],{},"Workers see the page"," with that row's values filled in, and answer.",[36,981,982,985],{},[39,983,984],{},"Your page sends back the answer"," in whatever shape you choose.",[36,987,988,991],{},[39,989,990],{},"You collect the results",", each one paired with the row it came from.",[22,993],{},[25,995,997],{"id":996},"the-example","The example",[12,999,1000],{},"We'll build a template that asks workers to classify an image. This is what a worker sees:",[12,1002,1003],{},[289,1004],{"alt":1005,"src":1006},"The image-subject template as a worker sees it: a photo of a black puppy, four subject options with Animal selected, a blurry-or-broken checkbox and a Submit button","/img/news/template-image-subject.png",[12,1008,1009],{},"Every template has the same three parts, plus instructions for the worker. Here they are in this one.",[532,1011,1013],{"id":1012},"_1-the-fields","1. The fields",[12,1015,1016],{},"Fields are the data that changes from task to task. This template has two: the image to show, and an id so you can match answers back to your data.",[1018,1019,1023],"pre",{"className":1020,"code":1021,"language":1022,"meta":234,"style":234},"language-js shiki shiki-themes github-dark","const imageUrl = \"${image_url}\";\nconst imageId = \"${image_id}\";\ndocument.getElementById(\"photo\").src = imageUrl;\n","js",[44,1024,1025,1048,1062],{"__ignoreMap":234},[546,1026,1029,1033,1037,1040,1044],{"class":1027,"line":1028},"line",1,[546,1030,1032],{"class":1031},"snl16","const",[546,1034,1036],{"class":1035},"sDLfK"," imageUrl",[546,1038,1039],{"class":1031}," =",[546,1041,1043],{"class":1042},"sU2Wk"," \"${image_url}\"",[546,1045,1047],{"class":1046},"s95oV",";\n",[546,1049,1050,1052,1055,1057,1060],{"class":1027,"line":235},[546,1051,1032],{"class":1031},[546,1053,1054],{"class":1035}," imageId",[546,1056,1039],{"class":1031},[546,1058,1059],{"class":1042}," \"${image_id}\"",[546,1061,1047],{"class":1046},[546,1063,1064,1067,1071,1074,1077,1080,1083],{"class":1027,"line":896},[546,1065,1066],{"class":1046},"document.",[546,1068,1070],{"class":1069},"svObZ","getElementById",[546,1072,1073],{"class":1046},"(",[546,1075,1076],{"class":1042},"\"photo\"",[546,1078,1079],{"class":1046},").src ",[546,1081,1082],{"class":1031},"=",[546,1084,1085],{"class":1046}," imageUrl;\n",[12,1087,1088,1089,1092,1093,1095,1096,1099,1100,205],{},"Each ",[44,1090,1091],{},"${...}"," becomes a column in your CSV, so the headers must match the placeholders exactly. The console lists a template's fields, and a job with a missing column is rejected. Put placeholders inside double-quoted strings in your script, as above, and use JavaScript to put them on the page. That keeps quotes and symbols in your data from breaking the page. Avoid using ",[44,1094,1091],{}," for other JavaScript expressions: the template system treats them as placeholders too. For example, write ",[44,1097,1098],{},"`Score: ` + score"," instead of ",[44,1101,1102],{},"`Score: ${score}`",[532,1104,1106],{"id":1105},"_2-the-task","2. The task",[12,1108,1109],{},"This part is ordinary HTML: the question, the image and the answer options. Style it however you like. Keep it focused on one decision, since workers move through many tasks quickly.",[12,1111,1112],{},"Here's the HTML for this example:",[1018,1114,1118],{"className":1115,"code":1116,"language":1117,"meta":234,"style":234},"language-html shiki shiki-themes github-dark","\u003Cstyle>\n  body { font-family: system-ui, sans-serif; max-width: 640px; margin: 2rem auto; padding: 0 1rem; }\n  img { max-width: 100%; border-radius: 8px; }\n  label { display: block; margin: .4rem 0; }\n  button { margin-top: 1rem; }\n  #error { color: #c00; display: none; }\n\u003C/style>\n\n\u003Ch2>What is the main subject of this image?\u003C/h2>\n\u003Cimg id=\"photo\" alt=\"Image to classify\" />\n\n\u003Cdiv class=\"options\">\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"person\" /> Person\u003C/label>\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"animal\" /> Animal\u003C/label>\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"vehicle\" /> Vehicle\u003C/label>\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"other\" /> Something else\u003C/label>\n\u003C/div>\n\u003Clabel>\u003Cinput type=\"checkbox\" id=\"unclear\" /> The image is blurry or broken\u003C/label>\n\n\u003Cp id=\"error\">Pick a subject, or mark the image as unclear.\u003C/p>\n\u003Cbutton id=\"submit\">Submit\u003C/button>\n","html",[44,1119,1120,1132,1203,1234,1266,1286,1313,1323,1329,1343,1368,1373,1390,1436,1473,1510,1547,1556,1588,1593,1614],{"__ignoreMap":234},[546,1121,1122,1125,1129],{"class":1027,"line":1028},[546,1123,1124],{"class":1046},"\u003C",[546,1126,1128],{"class":1127},"s4JwU","style",[546,1130,1131],{"class":1046},">\n",[546,1133,1134,1137,1140,1143,1146,1149,1152,1155,1158,1161,1163,1166,1169,1171,1174,1176,1179,1182,1185,1187,1190,1192,1195,1198,1200],{"class":1027,"line":235},[546,1135,1136],{"class":1127},"  body",[546,1138,1139],{"class":1046}," { ",[546,1141,1142],{"class":1035},"font-family",[546,1144,1145],{"class":1046},": ",[546,1147,1148],{"class":1035},"system-ui",[546,1150,1151],{"class":1046},", ",[546,1153,1154],{"class":1035},"sans-serif",[546,1156,1157],{"class":1046},"; ",[546,1159,1160],{"class":1035},"max-width",[546,1162,1145],{"class":1046},[546,1164,1165],{"class":1035},"640",[546,1167,1168],{"class":1031},"px",[546,1170,1157],{"class":1046},[546,1172,1173],{"class":1035},"margin",[546,1175,1145],{"class":1046},[546,1177,1178],{"class":1035},"2",[546,1180,1181],{"class":1031},"rem",[546,1183,1184],{"class":1035}," auto",[546,1186,1157],{"class":1046},[546,1188,1189],{"class":1035},"padding",[546,1191,1145],{"class":1046},[546,1193,1194],{"class":1035},"0",[546,1196,1197],{"class":1035}," 1",[546,1199,1181],{"class":1031},[546,1201,1202],{"class":1046},"; }\n",[546,1204,1205,1208,1210,1212,1214,1217,1220,1222,1225,1227,1230,1232],{"class":1027,"line":896},[546,1206,1207],{"class":1127},"  img",[546,1209,1139],{"class":1046},[546,1211,1160],{"class":1035},[546,1213,1145],{"class":1046},[546,1215,1216],{"class":1035},"100",[546,1218,1219],{"class":1031},"%",[546,1221,1157],{"class":1046},[546,1223,1224],{"class":1035},"border-radius",[546,1226,1145],{"class":1046},[546,1228,1229],{"class":1035},"8",[546,1231,1168],{"class":1031},[546,1233,1202],{"class":1046},[546,1235,1237,1240,1242,1245,1247,1250,1252,1254,1256,1259,1261,1264],{"class":1027,"line":1236},4,[546,1238,1239],{"class":1127},"  label",[546,1241,1139],{"class":1046},[546,1243,1244],{"class":1035},"display",[546,1246,1145],{"class":1046},[546,1248,1249],{"class":1035},"block",[546,1251,1157],{"class":1046},[546,1253,1173],{"class":1035},[546,1255,1145],{"class":1046},[546,1257,1258],{"class":1035},".4",[546,1260,1181],{"class":1031},[546,1262,1263],{"class":1035}," 0",[546,1265,1202],{"class":1046},[546,1267,1269,1272,1274,1277,1279,1282,1284],{"class":1027,"line":1268},5,[546,1270,1271],{"class":1127},"  button",[546,1273,1139],{"class":1046},[546,1275,1276],{"class":1035},"margin-top",[546,1278,1145],{"class":1046},[546,1280,1281],{"class":1035},"1",[546,1283,1181],{"class":1031},[546,1285,1202],{"class":1046},[546,1287,1289,1292,1294,1297,1299,1302,1304,1306,1308,1311],{"class":1027,"line":1288},6,[546,1290,1291],{"class":1069},"  #error",[546,1293,1139],{"class":1046},[546,1295,1296],{"class":1035},"color",[546,1298,1145],{"class":1046},[546,1300,1301],{"class":1035},"#c00",[546,1303,1157],{"class":1046},[546,1305,1244],{"class":1035},[546,1307,1145],{"class":1046},[546,1309,1310],{"class":1035},"none",[546,1312,1202],{"class":1046},[546,1314,1316,1319,1321],{"class":1027,"line":1315},7,[546,1317,1318],{"class":1046},"\u003C/",[546,1320,1128],{"class":1127},[546,1322,1131],{"class":1046},[546,1324,1326],{"class":1027,"line":1325},8,[546,1327,1328],{"emptyLinePlaceholder":260},"\n",[546,1330,1332,1334,1336,1339,1341],{"class":1027,"line":1331},9,[546,1333,1124],{"class":1046},[546,1335,25],{"class":1127},[546,1337,1338],{"class":1046},">What is the main subject of this image?\u003C/",[546,1340,25],{"class":1127},[546,1342,1131],{"class":1046},[546,1344,1346,1348,1350,1353,1355,1357,1360,1362,1365],{"class":1027,"line":1345},10,[546,1347,1124],{"class":1046},[546,1349,289],{"class":1127},[546,1351,1352],{"class":1069}," id",[546,1354,1082],{"class":1046},[546,1356,1076],{"class":1042},[546,1358,1359],{"class":1069}," alt",[546,1361,1082],{"class":1046},[546,1363,1364],{"class":1042},"\"Image to classify\"",[546,1366,1367],{"class":1046}," />\n",[546,1369,1371],{"class":1027,"line":1370},11,[546,1372,1328],{"emptyLinePlaceholder":260},[546,1374,1376,1378,1380,1383,1385,1388],{"class":1027,"line":1375},12,[546,1377,1124],{"class":1046},[546,1379,540],{"class":1127},[546,1381,1382],{"class":1069}," class",[546,1384,1082],{"class":1046},[546,1386,1387],{"class":1042},"\"options\"",[546,1389,1131],{"class":1046},[546,1391,1393,1396,1399,1402,1405,1408,1410,1413,1416,1418,1421,1424,1426,1429,1432,1434],{"class":1027,"line":1392},13,[546,1394,1395],{"class":1046},"  \u003C",[546,1397,1398],{"class":1127},"label",[546,1400,1401],{"class":1046},">\u003C",[546,1403,1404],{"class":1127},"input",[546,1406,1407],{"class":1069}," type",[546,1409,1082],{"class":1046},[546,1411,1412],{"class":1042},"\"radio\"",[546,1414,1415],{"class":1069}," name",[546,1417,1082],{"class":1046},[546,1419,1420],{"class":1042},"\"subject\"",[546,1422,1423],{"class":1069}," value",[546,1425,1082],{"class":1046},[546,1427,1428],{"class":1042},"\"person\"",[546,1430,1431],{"class":1046}," /> Person\u003C/",[546,1433,1398],{"class":1127},[546,1435,1131],{"class":1046},[546,1437,1439,1441,1443,1445,1447,1449,1451,1453,1455,1457,1459,1461,1463,1466,1469,1471],{"class":1027,"line":1438},14,[546,1440,1395],{"class":1046},[546,1442,1398],{"class":1127},[546,1444,1401],{"class":1046},[546,1446,1404],{"class":1127},[546,1448,1407],{"class":1069},[546,1450,1082],{"class":1046},[546,1452,1412],{"class":1042},[546,1454,1415],{"class":1069},[546,1456,1082],{"class":1046},[546,1458,1420],{"class":1042},[546,1460,1423],{"class":1069},[546,1462,1082],{"class":1046},[546,1464,1465],{"class":1042},"\"animal\"",[546,1467,1468],{"class":1046}," /> Animal\u003C/",[546,1470,1398],{"class":1127},[546,1472,1131],{"class":1046},[546,1474,1476,1478,1480,1482,1484,1486,1488,1490,1492,1494,1496,1498,1500,1503,1506,1508],{"class":1027,"line":1475},15,[546,1477,1395],{"class":1046},[546,1479,1398],{"class":1127},[546,1481,1401],{"class":1046},[546,1483,1404],{"class":1127},[546,1485,1407],{"class":1069},[546,1487,1082],{"class":1046},[546,1489,1412],{"class":1042},[546,1491,1415],{"class":1069},[546,1493,1082],{"class":1046},[546,1495,1420],{"class":1042},[546,1497,1423],{"class":1069},[546,1499,1082],{"class":1046},[546,1501,1502],{"class":1042},"\"vehicle\"",[546,1504,1505],{"class":1046}," /> Vehicle\u003C/",[546,1507,1398],{"class":1127},[546,1509,1131],{"class":1046},[546,1511,1513,1515,1517,1519,1521,1523,1525,1527,1529,1531,1533,1535,1537,1540,1543,1545],{"class":1027,"line":1512},16,[546,1514,1395],{"class":1046},[546,1516,1398],{"class":1127},[546,1518,1401],{"class":1046},[546,1520,1404],{"class":1127},[546,1522,1407],{"class":1069},[546,1524,1082],{"class":1046},[546,1526,1412],{"class":1042},[546,1528,1415],{"class":1069},[546,1530,1082],{"class":1046},[546,1532,1420],{"class":1042},[546,1534,1423],{"class":1069},[546,1536,1082],{"class":1046},[546,1538,1539],{"class":1042},"\"other\"",[546,1541,1542],{"class":1046}," /> Something else\u003C/",[546,1544,1398],{"class":1127},[546,1546,1131],{"class":1046},[546,1548,1550,1552,1554],{"class":1027,"line":1549},17,[546,1551,1318],{"class":1046},[546,1553,540],{"class":1127},[546,1555,1131],{"class":1046},[546,1557,1559,1561,1563,1565,1567,1569,1571,1574,1576,1578,1581,1584,1586],{"class":1027,"line":1558},18,[546,1560,1124],{"class":1046},[546,1562,1398],{"class":1127},[546,1564,1401],{"class":1046},[546,1566,1404],{"class":1127},[546,1568,1407],{"class":1069},[546,1570,1082],{"class":1046},[546,1572,1573],{"class":1042},"\"checkbox\"",[546,1575,1352],{"class":1069},[546,1577,1082],{"class":1046},[546,1579,1580],{"class":1042},"\"unclear\"",[546,1582,1583],{"class":1046}," /> The image is blurry or broken\u003C/",[546,1585,1398],{"class":1127},[546,1587,1131],{"class":1046},[546,1589,1591],{"class":1027,"line":1590},19,[546,1592,1328],{"emptyLinePlaceholder":260},[546,1594,1596,1598,1600,1602,1604,1607,1610,1612],{"class":1027,"line":1595},20,[546,1597,1124],{"class":1046},[546,1599,12],{"class":1127},[546,1601,1352],{"class":1069},[546,1603,1082],{"class":1046},[546,1605,1606],{"class":1042},"\"error\"",[546,1608,1609],{"class":1046},">Pick a subject, or mark the image as unclear.\u003C/",[546,1611,12],{"class":1127},[546,1613,1131],{"class":1046},[546,1615,1617,1619,1622,1624,1626,1629,1632,1634],{"class":1027,"line":1616},21,[546,1618,1124],{"class":1046},[546,1620,1621],{"class":1127},"button",[546,1623,1352],{"class":1069},[546,1625,1082],{"class":1046},[546,1627,1628],{"class":1042},"\"submit\"",[546,1630,1631],{"class":1046},">Submit\u003C/",[546,1633,1621],{"class":1127},[546,1635,1131],{"class":1046},[12,1637,1638,1639,1642,1643,1646],{},"The ",[44,1640,1641],{},"\u003Cimg>"," starts out empty, and the script from part 1 fills in its ",[44,1644,1645],{},"src",". The error message stays hidden until the check in part 3 shows it.",[532,1648,1650],{"id":1649},"_3-the-answer","3. The answer",[12,1652,1653],{},"When the worker is done, the page sends their answer back:",[1018,1655,1657],{"className":1020,"code":1656,"language":1022,"meta":234,"style":234},"window.parent.postMessage({\n  task: \"submit\",\n  values: { imageId: imageId, subject: \"animal\", unclear: false }\n}, \"*\");\n",[44,1658,1659,1670,1680,1696],{"__ignoreMap":234},[546,1660,1661,1664,1667],{"class":1027,"line":1028},[546,1662,1663],{"class":1046},"window.parent.",[546,1665,1666],{"class":1069},"postMessage",[546,1668,1669],{"class":1046},"({\n",[546,1671,1672,1675,1677],{"class":1027,"line":235},[546,1673,1674],{"class":1046},"  task: ",[546,1676,1628],{"class":1042},[546,1678,1679],{"class":1046},",\n",[546,1681,1682,1685,1687,1690,1693],{"class":1027,"line":896},[546,1683,1684],{"class":1046},"  values: { imageId: imageId, subject: ",[546,1686,1465],{"class":1042},[546,1688,1689],{"class":1046},", unclear: ",[546,1691,1692],{"class":1035},"false",[546,1694,1695],{"class":1046}," }\n",[546,1697,1698,1701,1704],{"class":1027,"line":1236},[546,1699,1700],{"class":1046},"}, ",[546,1702,1703],{"class":1042},"\"*\"",[546,1705,1706],{"class":1046},");\n",[12,1708,1709,1710,1713,1714,1717],{},"You choose what goes in ",[44,1711,1712],{},"values",", and your results come back in that same shape. Include an id such as ",[44,1715,1716],{},"imageId"," so each answer can be matched to its input and understood on its own.",[12,1719,1720],{},"Always add client-side validation checks to your template. Since workers cannot edit their submissions after sending, attaching validation logic to your template's Submit button prevents empty or missing responses.  In this example, the Submit button runs a check first:",[1018,1722,1724],{"className":1020,"code":1723,"language":1022,"meta":234,"style":234},"document.getElementById(\"submit\").addEventListener(\"click\", () => {\n  const choice = document.querySelector('input[name=\"subject\"]:checked');\n  const unclear = document.getElementById(\"unclear\").checked;\n\n  // Nothing picked yet: show an error and stop, so nothing is sent\n  if (!choice && !unclear) {\n    document.getElementById(\"error\").style.display = \"block\";\n    return;\n  }\n\n  // Only reached once the worker has picked a subject or ticked the box\n  window.parent.postMessage({\n    task: \"submit\",\n    values: { imageId: imageId, subject: choice ? choice.value : null, unclear: unclear }\n  }, \"*\");\n});\n",[44,1725,1726,1756,1779,1799,1803,1809,1832,1853,1860,1865,1869,1874,1883,1892,1912,1921],{"__ignoreMap":234},[546,1727,1728,1730,1732,1734,1736,1739,1742,1744,1747,1750,1753],{"class":1027,"line":1028},[546,1729,1066],{"class":1046},[546,1731,1070],{"class":1069},[546,1733,1073],{"class":1046},[546,1735,1628],{"class":1042},[546,1737,1738],{"class":1046},").",[546,1740,1741],{"class":1069},"addEventListener",[546,1743,1073],{"class":1046},[546,1745,1746],{"class":1042},"\"click\"",[546,1748,1749],{"class":1046},", () ",[546,1751,1752],{"class":1031},"=>",[546,1754,1755],{"class":1046}," {\n",[546,1757,1758,1761,1764,1766,1769,1772,1774,1777],{"class":1027,"line":235},[546,1759,1760],{"class":1031},"  const",[546,1762,1763],{"class":1035}," choice",[546,1765,1039],{"class":1031},[546,1767,1768],{"class":1046}," document.",[546,1770,1771],{"class":1069},"querySelector",[546,1773,1073],{"class":1046},[546,1775,1776],{"class":1042},"'input[name=\"subject\"]:checked'",[546,1778,1706],{"class":1046},[546,1780,1781,1783,1786,1788,1790,1792,1794,1796],{"class":1027,"line":896},[546,1782,1760],{"class":1031},[546,1784,1785],{"class":1035}," unclear",[546,1787,1039],{"class":1031},[546,1789,1768],{"class":1046},[546,1791,1070],{"class":1069},[546,1793,1073],{"class":1046},[546,1795,1580],{"class":1042},[546,1797,1798],{"class":1046},").checked;\n",[546,1800,1801],{"class":1027,"line":1236},[546,1802,1328],{"emptyLinePlaceholder":260},[546,1804,1805],{"class":1027,"line":1268},[546,1806,1808],{"class":1807},"sAwPA","  // Nothing picked yet: show an error and stop, so nothing is sent\n",[546,1810,1811,1814,1817,1820,1823,1826,1829],{"class":1027,"line":1288},[546,1812,1813],{"class":1031},"  if",[546,1815,1816],{"class":1046}," (",[546,1818,1819],{"class":1031},"!",[546,1821,1822],{"class":1046},"choice ",[546,1824,1825],{"class":1031},"&&",[546,1827,1828],{"class":1031}," !",[546,1830,1831],{"class":1046},"unclear) {\n",[546,1833,1834,1837,1839,1841,1843,1846,1848,1851],{"class":1027,"line":1315},[546,1835,1836],{"class":1046},"    document.",[546,1838,1070],{"class":1069},[546,1840,1073],{"class":1046},[546,1842,1606],{"class":1042},[546,1844,1845],{"class":1046},").style.display ",[546,1847,1082],{"class":1031},[546,1849,1850],{"class":1042}," \"block\"",[546,1852,1047],{"class":1046},[546,1854,1855,1858],{"class":1027,"line":1325},[546,1856,1857],{"class":1031},"    return",[546,1859,1047],{"class":1046},[546,1861,1862],{"class":1027,"line":1331},[546,1863,1864],{"class":1046},"  }\n",[546,1866,1867],{"class":1027,"line":1345},[546,1868,1328],{"emptyLinePlaceholder":260},[546,1870,1871],{"class":1027,"line":1370},[546,1872,1873],{"class":1807},"  // Only reached once the worker has picked a subject or ticked the box\n",[546,1875,1876,1879,1881],{"class":1027,"line":1375},[546,1877,1878],{"class":1046},"  window.parent.",[546,1880,1666],{"class":1069},[546,1882,1669],{"class":1046},[546,1884,1885,1888,1890],{"class":1027,"line":1392},[546,1886,1887],{"class":1046},"    task: ",[546,1889,1628],{"class":1042},[546,1891,1679],{"class":1046},[546,1893,1894,1897,1900,1903,1906,1909],{"class":1027,"line":1438},[546,1895,1896],{"class":1046},"    values: { imageId: imageId, subject: choice ",[546,1898,1899],{"class":1031},"?",[546,1901,1902],{"class":1046}," choice.value ",[546,1904,1905],{"class":1031},":",[546,1907,1908],{"class":1035}," null",[546,1910,1911],{"class":1046},", unclear: unclear }\n",[546,1913,1914,1917,1919],{"class":1027,"line":1475},[546,1915,1916],{"class":1046},"  }, ",[546,1918,1703],{"class":1042},[546,1920,1706],{"class":1046},[546,1922,1923],{"class":1027,"line":1512},[546,1924,1925],{"class":1046},"});\n",[12,1927,1928],{},"If the worker hasn't picked a subject or ticked the blurry-or-broken box, they see \"Pick a subject, or mark the image as unclear.\" and the task stays open.",[532,1930,1932],{"id":1931},"_4-the-instructions","4. The instructions",[12,1934,1935],{},"Instructions tell workers what a good answer looks like. Workers open them from an info button beside the task, and clear instructions are one of the easiest ways to get better answers. Treat them as part of the template, not an extra. You can use headings, lists, links and images.",[12,1937,1938],{},"There are two ways to add them. Pick one, not both.",[12,1940,1941,1944],{},[39,1942,1943],{},"Option 1: bake them into the template (recommended)."," Add this to your script, with the instructions written as HTML right in the call. It sends them to the worker app when the task loads:",[1018,1946,1948],{"className":1020,"code":1947,"language":1022,"meta":234,"style":234},"window.top.postMessage({\n  type: \"task-instructions\",\n  instructions: `\n    \u003Ch3>How to answer\u003C/h3>\n    \u003Cp>Pick what the photo is mostly about. If two subjects look equally important, pick the larger one.\u003C/p>\n    \u003Cp>Tick \u003Cstrong>blurry or broken\u003C/strong> only if you can't tell what the image shows.\u003C/p>\n  `\n}, \"*\");\n",[44,1949,1950,1959,1969,1977,1982,1987,1992,1997],{"__ignoreMap":234},[546,1951,1952,1955,1957],{"class":1027,"line":1028},[546,1953,1954],{"class":1046},"window.top.",[546,1956,1666],{"class":1069},[546,1958,1669],{"class":1046},[546,1960,1961,1964,1967],{"class":1027,"line":235},[546,1962,1963],{"class":1046},"  type: ",[546,1965,1966],{"class":1042},"\"task-instructions\"",[546,1968,1679],{"class":1046},[546,1970,1971,1974],{"class":1027,"line":896},[546,1972,1973],{"class":1046},"  instructions: ",[546,1975,1976],{"class":1042},"`\n",[546,1978,1979],{"class":1027,"line":1236},[546,1980,1981],{"class":1042},"    \u003Ch3>How to answer\u003C/h3>\n",[546,1983,1984],{"class":1027,"line":1268},[546,1985,1986],{"class":1042},"    \u003Cp>Pick what the photo is mostly about. If two subjects look equally important, pick the larger one.\u003C/p>\n",[546,1988,1989],{"class":1027,"line":1288},[546,1990,1991],{"class":1042},"    \u003Cp>Tick \u003Cstrong>blurry or broken\u003C/strong> only if you can't tell what the image shows.\u003C/p>\n",[546,1993,1994],{"class":1027,"line":1315},[546,1995,1996],{"class":1042},"  `\n",[546,1998,1999,2001,2003],{"class":1027,"line":1325},[546,2000,1700],{"class":1046},[546,2002,1703],{"class":1042},[546,2004,1706],{"class":1046},[12,2006,2007,2008,2011,2012,2015],{},"Note that instructions go to ",[44,2009,2010],{},"window.top",", while the answer goes to ",[44,2013,2014],{},"window.parent",". With this option, everything lives in one file. The task and its instructions are edited, reviewed and registered together, whether you use the console or the API.",[12,2017,2018,2021,2022,2025],{},[39,2019,2020],{},"Option 2: use the API console."," When you create the template in the console, paste the instruction HTML into the ",[39,2023,2024],{},"Worker instructions"," box. The console adds it to your template when you submit. It's quick, but the instructions then live in the console instead of your file, so registering the same file over the API later leaves them out. The console won't accept a template that already sends its own instructions, so if you chose option 1, leave the box empty.",[532,2027,2029],{"id":2028},"the-full-template","The full template",[12,2031,2032],{},"Here's all of it together. Copy it as a starting point.",[1018,2034,2036],{"className":1115,"code":2035,"language":1117,"meta":234,"style":234},"\u003Cstyle>\n  body { font-family: system-ui, sans-serif; max-width: 640px; margin: 2rem auto; padding: 0 1rem; }\n  img { max-width: 100%; border-radius: 8px; }\n  label { display: block; margin: .4rem 0; }\n  button { margin-top: 1rem; }\n  #error { color: #c00; display: none; }\n\u003C/style>\n\n\u003Ch2>What is the main subject of this image?\u003C/h2>\n\u003Cimg id=\"photo\" alt=\"Image to classify\" />\n\n\u003Cdiv class=\"options\">\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"person\" /> Person\u003C/label>\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"animal\" /> Animal\u003C/label>\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"vehicle\" /> Vehicle\u003C/label>\n  \u003Clabel>\u003Cinput type=\"radio\" name=\"subject\" value=\"other\" /> Something else\u003C/label>\n\u003C/div>\n\u003Clabel>\u003Cinput type=\"checkbox\" id=\"unclear\" /> The image is blurry or broken\u003C/label>\n\n\u003Cp id=\"error\">Pick a subject, or mark the image as unclear.\u003C/p>\n\u003Cbutton id=\"submit\">Submit\u003C/button>\n\n\u003Cscript>\n  // 1. Fields\n  const imageUrl = \"${image_url}\";\n  const imageId = \"${image_id}\";\n  document.getElementById(\"photo\").src = imageUrl;\n\n  // 2. Instructions: send them to the worker app on load\n  window.top.postMessage({\n    type: \"task-instructions\",\n    instructions: `\n      \u003Ch3>How to answer\u003C/h3>\n      \u003Cp>Pick what the photo is mostly about. If two subjects look equally important, pick the larger one.\u003C/p>\n      \u003Cp>Tick \u003Cstrong>blurry or broken\u003C/strong> only if you can't tell what the image shows.\u003C/p>\n    `\n  }, \"*\");\n\n  // 3. Answer: check the input, then submit\n  document.getElementById(\"submit\").addEventListener(\"click\", () => {\n    const choice = document.querySelector('input[name=\"subject\"]:checked');\n    const unclear = document.getElementById(\"unclear\").checked;\n    if (!choice && !unclear) {\n      document.getElementById(\"error\").style.display = \"block\";\n      return;\n    }\n    window.parent.postMessage({\n      task: \"submit\",\n      values: { imageId: imageId, subject: choice ? choice.value : null, unclear: unclear }\n    }, \"*\");\n  });\n\u003C/script>\n",[44,2037,2038,2046,2098,2124,2150,2166,2188,2196,2200,2212,2232,2236,2250,2284,2318,2352,2386,2394,2422,2426,2444,2462,2467,2477,2483,2496,2509,2527,2532,2538,2548,2558,2566,2572,2578,2584,2590,2599,2604,2610,2635,2655,2674,2692,2712,2720,2726,2736,2746,2762,2772,2778],{"__ignoreMap":234},[546,2039,2040,2042,2044],{"class":1027,"line":1028},[546,2041,1124],{"class":1046},[546,2043,1128],{"class":1127},[546,2045,1131],{"class":1046},[546,2047,2048,2050,2052,2054,2056,2058,2060,2062,2064,2066,2068,2070,2072,2074,2076,2078,2080,2082,2084,2086,2088,2090,2092,2094,2096],{"class":1027,"line":235},[546,2049,1136],{"class":1127},[546,2051,1139],{"class":1046},[546,2053,1142],{"class":1035},[546,2055,1145],{"class":1046},[546,2057,1148],{"class":1035},[546,2059,1151],{"class":1046},[546,2061,1154],{"class":1035},[546,2063,1157],{"class":1046},[546,2065,1160],{"class":1035},[546,2067,1145],{"class":1046},[546,2069,1165],{"class":1035},[546,2071,1168],{"class":1031},[546,2073,1157],{"class":1046},[546,2075,1173],{"class":1035},[546,2077,1145],{"class":1046},[546,2079,1178],{"class":1035},[546,2081,1181],{"class":1031},[546,2083,1184],{"class":1035},[546,2085,1157],{"class":1046},[546,2087,1189],{"class":1035},[546,2089,1145],{"class":1046},[546,2091,1194],{"class":1035},[546,2093,1197],{"class":1035},[546,2095,1181],{"class":1031},[546,2097,1202],{"class":1046},[546,2099,2100,2102,2104,2106,2108,2110,2112,2114,2116,2118,2120,2122],{"class":1027,"line":896},[546,2101,1207],{"class":1127},[546,2103,1139],{"class":1046},[546,2105,1160],{"class":1035},[546,2107,1145],{"class":1046},[546,2109,1216],{"class":1035},[546,2111,1219],{"class":1031},[546,2113,1157],{"class":1046},[546,2115,1224],{"class":1035},[546,2117,1145],{"class":1046},[546,2119,1229],{"class":1035},[546,2121,1168],{"class":1031},[546,2123,1202],{"class":1046},[546,2125,2126,2128,2130,2132,2134,2136,2138,2140,2142,2144,2146,2148],{"class":1027,"line":1236},[546,2127,1239],{"class":1127},[546,2129,1139],{"class":1046},[546,2131,1244],{"class":1035},[546,2133,1145],{"class":1046},[546,2135,1249],{"class":1035},[546,2137,1157],{"class":1046},[546,2139,1173],{"class":1035},[546,2141,1145],{"class":1046},[546,2143,1258],{"class":1035},[546,2145,1181],{"class":1031},[546,2147,1263],{"class":1035},[546,2149,1202],{"class":1046},[546,2151,2152,2154,2156,2158,2160,2162,2164],{"class":1027,"line":1268},[546,2153,1271],{"class":1127},[546,2155,1139],{"class":1046},[546,2157,1276],{"class":1035},[546,2159,1145],{"class":1046},[546,2161,1281],{"class":1035},[546,2163,1181],{"class":1031},[546,2165,1202],{"class":1046},[546,2167,2168,2170,2172,2174,2176,2178,2180,2182,2184,2186],{"class":1027,"line":1288},[546,2169,1291],{"class":1069},[546,2171,1139],{"class":1046},[546,2173,1296],{"class":1035},[546,2175,1145],{"class":1046},[546,2177,1301],{"class":1035},[546,2179,1157],{"class":1046},[546,2181,1244],{"class":1035},[546,2183,1145],{"class":1046},[546,2185,1310],{"class":1035},[546,2187,1202],{"class":1046},[546,2189,2190,2192,2194],{"class":1027,"line":1315},[546,2191,1318],{"class":1046},[546,2193,1128],{"class":1127},[546,2195,1131],{"class":1046},[546,2197,2198],{"class":1027,"line":1325},[546,2199,1328],{"emptyLinePlaceholder":260},[546,2201,2202,2204,2206,2208,2210],{"class":1027,"line":1331},[546,2203,1124],{"class":1046},[546,2205,25],{"class":1127},[546,2207,1338],{"class":1046},[546,2209,25],{"class":1127},[546,2211,1131],{"class":1046},[546,2213,2214,2216,2218,2220,2222,2224,2226,2228,2230],{"class":1027,"line":1345},[546,2215,1124],{"class":1046},[546,2217,289],{"class":1127},[546,2219,1352],{"class":1069},[546,2221,1082],{"class":1046},[546,2223,1076],{"class":1042},[546,2225,1359],{"class":1069},[546,2227,1082],{"class":1046},[546,2229,1364],{"class":1042},[546,2231,1367],{"class":1046},[546,2233,2234],{"class":1027,"line":1370},[546,2235,1328],{"emptyLinePlaceholder":260},[546,2237,2238,2240,2242,2244,2246,2248],{"class":1027,"line":1375},[546,2239,1124],{"class":1046},[546,2241,540],{"class":1127},[546,2243,1382],{"class":1069},[546,2245,1082],{"class":1046},[546,2247,1387],{"class":1042},[546,2249,1131],{"class":1046},[546,2251,2252,2254,2256,2258,2260,2262,2264,2266,2268,2270,2272,2274,2276,2278,2280,2282],{"class":1027,"line":1392},[546,2253,1395],{"class":1046},[546,2255,1398],{"class":1127},[546,2257,1401],{"class":1046},[546,2259,1404],{"class":1127},[546,2261,1407],{"class":1069},[546,2263,1082],{"class":1046},[546,2265,1412],{"class":1042},[546,2267,1415],{"class":1069},[546,2269,1082],{"class":1046},[546,2271,1420],{"class":1042},[546,2273,1423],{"class":1069},[546,2275,1082],{"class":1046},[546,2277,1428],{"class":1042},[546,2279,1431],{"class":1046},[546,2281,1398],{"class":1127},[546,2283,1131],{"class":1046},[546,2285,2286,2288,2290,2292,2294,2296,2298,2300,2302,2304,2306,2308,2310,2312,2314,2316],{"class":1027,"line":1438},[546,2287,1395],{"class":1046},[546,2289,1398],{"class":1127},[546,2291,1401],{"class":1046},[546,2293,1404],{"class":1127},[546,2295,1407],{"class":1069},[546,2297,1082],{"class":1046},[546,2299,1412],{"class":1042},[546,2301,1415],{"class":1069},[546,2303,1082],{"class":1046},[546,2305,1420],{"class":1042},[546,2307,1423],{"class":1069},[546,2309,1082],{"class":1046},[546,2311,1465],{"class":1042},[546,2313,1468],{"class":1046},[546,2315,1398],{"class":1127},[546,2317,1131],{"class":1046},[546,2319,2320,2322,2324,2326,2328,2330,2332,2334,2336,2338,2340,2342,2344,2346,2348,2350],{"class":1027,"line":1475},[546,2321,1395],{"class":1046},[546,2323,1398],{"class":1127},[546,2325,1401],{"class":1046},[546,2327,1404],{"class":1127},[546,2329,1407],{"class":1069},[546,2331,1082],{"class":1046},[546,2333,1412],{"class":1042},[546,2335,1415],{"class":1069},[546,2337,1082],{"class":1046},[546,2339,1420],{"class":1042},[546,2341,1423],{"class":1069},[546,2343,1082],{"class":1046},[546,2345,1502],{"class":1042},[546,2347,1505],{"class":1046},[546,2349,1398],{"class":1127},[546,2351,1131],{"class":1046},[546,2353,2354,2356,2358,2360,2362,2364,2366,2368,2370,2372,2374,2376,2378,2380,2382,2384],{"class":1027,"line":1512},[546,2355,1395],{"class":1046},[546,2357,1398],{"class":1127},[546,2359,1401],{"class":1046},[546,2361,1404],{"class":1127},[546,2363,1407],{"class":1069},[546,2365,1082],{"class":1046},[546,2367,1412],{"class":1042},[546,2369,1415],{"class":1069},[546,2371,1082],{"class":1046},[546,2373,1420],{"class":1042},[546,2375,1423],{"class":1069},[546,2377,1082],{"class":1046},[546,2379,1539],{"class":1042},[546,2381,1542],{"class":1046},[546,2383,1398],{"class":1127},[546,2385,1131],{"class":1046},[546,2387,2388,2390,2392],{"class":1027,"line":1549},[546,2389,1318],{"class":1046},[546,2391,540],{"class":1127},[546,2393,1131],{"class":1046},[546,2395,2396,2398,2400,2402,2404,2406,2408,2410,2412,2414,2416,2418,2420],{"class":1027,"line":1558},[546,2397,1124],{"class":1046},[546,2399,1398],{"class":1127},[546,2401,1401],{"class":1046},[546,2403,1404],{"class":1127},[546,2405,1407],{"class":1069},[546,2407,1082],{"class":1046},[546,2409,1573],{"class":1042},[546,2411,1352],{"class":1069},[546,2413,1082],{"class":1046},[546,2415,1580],{"class":1042},[546,2417,1583],{"class":1046},[546,2419,1398],{"class":1127},[546,2421,1131],{"class":1046},[546,2423,2424],{"class":1027,"line":1590},[546,2425,1328],{"emptyLinePlaceholder":260},[546,2427,2428,2430,2432,2434,2436,2438,2440,2442],{"class":1027,"line":1595},[546,2429,1124],{"class":1046},[546,2431,12],{"class":1127},[546,2433,1352],{"class":1069},[546,2435,1082],{"class":1046},[546,2437,1606],{"class":1042},[546,2439,1609],{"class":1046},[546,2441,12],{"class":1127},[546,2443,1131],{"class":1046},[546,2445,2446,2448,2450,2452,2454,2456,2458,2460],{"class":1027,"line":1616},[546,2447,1124],{"class":1046},[546,2449,1621],{"class":1127},[546,2451,1352],{"class":1069},[546,2453,1082],{"class":1046},[546,2455,1628],{"class":1042},[546,2457,1631],{"class":1046},[546,2459,1621],{"class":1127},[546,2461,1131],{"class":1046},[546,2463,2465],{"class":1027,"line":2464},22,[546,2466,1328],{"emptyLinePlaceholder":260},[546,2468,2470,2472,2475],{"class":1027,"line":2469},23,[546,2471,1124],{"class":1046},[546,2473,2474],{"class":1127},"script",[546,2476,1131],{"class":1046},[546,2478,2480],{"class":1027,"line":2479},24,[546,2481,2482],{"class":1807},"  // 1. Fields\n",[546,2484,2486,2488,2490,2492,2494],{"class":1027,"line":2485},25,[546,2487,1760],{"class":1031},[546,2489,1036],{"class":1035},[546,2491,1039],{"class":1031},[546,2493,1043],{"class":1042},[546,2495,1047],{"class":1046},[546,2497,2499,2501,2503,2505,2507],{"class":1027,"line":2498},26,[546,2500,1760],{"class":1031},[546,2502,1054],{"class":1035},[546,2504,1039],{"class":1031},[546,2506,1059],{"class":1042},[546,2508,1047],{"class":1046},[546,2510,2512,2515,2517,2519,2521,2523,2525],{"class":1027,"line":2511},27,[546,2513,2514],{"class":1046},"  document.",[546,2516,1070],{"class":1069},[546,2518,1073],{"class":1046},[546,2520,1076],{"class":1042},[546,2522,1079],{"class":1046},[546,2524,1082],{"class":1031},[546,2526,1085],{"class":1046},[546,2528,2530],{"class":1027,"line":2529},28,[546,2531,1328],{"emptyLinePlaceholder":260},[546,2533,2535],{"class":1027,"line":2534},29,[546,2536,2537],{"class":1807},"  // 2. Instructions: send them to the worker app on load\n",[546,2539,2541,2544,2546],{"class":1027,"line":2540},30,[546,2542,2543],{"class":1046},"  window.top.",[546,2545,1666],{"class":1069},[546,2547,1669],{"class":1046},[546,2549,2551,2554,2556],{"class":1027,"line":2550},31,[546,2552,2553],{"class":1046},"    type: ",[546,2555,1966],{"class":1042},[546,2557,1679],{"class":1046},[546,2559,2561,2564],{"class":1027,"line":2560},32,[546,2562,2563],{"class":1046},"    instructions: ",[546,2565,1976],{"class":1042},[546,2567,2569],{"class":1027,"line":2568},33,[546,2570,2571],{"class":1042},"      \u003Ch3>How to answer\u003C/h3>\n",[546,2573,2575],{"class":1027,"line":2574},34,[546,2576,2577],{"class":1042},"      \u003Cp>Pick what the photo is mostly about. If two subjects look equally important, pick the larger one.\u003C/p>\n",[546,2579,2581],{"class":1027,"line":2580},35,[546,2582,2583],{"class":1042},"      \u003Cp>Tick \u003Cstrong>blurry or broken\u003C/strong> only if you can't tell what the image shows.\u003C/p>\n",[546,2585,2587],{"class":1027,"line":2586},36,[546,2588,2589],{"class":1042},"    `\n",[546,2591,2593,2595,2597],{"class":1027,"line":2592},37,[546,2594,1916],{"class":1046},[546,2596,1703],{"class":1042},[546,2598,1706],{"class":1046},[546,2600,2602],{"class":1027,"line":2601},38,[546,2603,1328],{"emptyLinePlaceholder":260},[546,2605,2607],{"class":1027,"line":2606},39,[546,2608,2609],{"class":1807},"  // 3. Answer: check the input, then submit\n",[546,2611,2613,2615,2617,2619,2621,2623,2625,2627,2629,2631,2633],{"class":1027,"line":2612},40,[546,2614,2514],{"class":1046},[546,2616,1070],{"class":1069},[546,2618,1073],{"class":1046},[546,2620,1628],{"class":1042},[546,2622,1738],{"class":1046},[546,2624,1741],{"class":1069},[546,2626,1073],{"class":1046},[546,2628,1746],{"class":1042},[546,2630,1749],{"class":1046},[546,2632,1752],{"class":1031},[546,2634,1755],{"class":1046},[546,2636,2638,2641,2643,2645,2647,2649,2651,2653],{"class":1027,"line":2637},41,[546,2639,2640],{"class":1031},"    const",[546,2642,1763],{"class":1035},[546,2644,1039],{"class":1031},[546,2646,1768],{"class":1046},[546,2648,1771],{"class":1069},[546,2650,1073],{"class":1046},[546,2652,1776],{"class":1042},[546,2654,1706],{"class":1046},[546,2656,2658,2660,2662,2664,2666,2668,2670,2672],{"class":1027,"line":2657},42,[546,2659,2640],{"class":1031},[546,2661,1785],{"class":1035},[546,2663,1039],{"class":1031},[546,2665,1768],{"class":1046},[546,2667,1070],{"class":1069},[546,2669,1073],{"class":1046},[546,2671,1580],{"class":1042},[546,2673,1798],{"class":1046},[546,2675,2677,2680,2682,2684,2686,2688,2690],{"class":1027,"line":2676},43,[546,2678,2679],{"class":1031},"    if",[546,2681,1816],{"class":1046},[546,2683,1819],{"class":1031},[546,2685,1822],{"class":1046},[546,2687,1825],{"class":1031},[546,2689,1828],{"class":1031},[546,2691,1831],{"class":1046},[546,2693,2695,2698,2700,2702,2704,2706,2708,2710],{"class":1027,"line":2694},44,[546,2696,2697],{"class":1046},"      document.",[546,2699,1070],{"class":1069},[546,2701,1073],{"class":1046},[546,2703,1606],{"class":1042},[546,2705,1845],{"class":1046},[546,2707,1082],{"class":1031},[546,2709,1850],{"class":1042},[546,2711,1047],{"class":1046},[546,2713,2715,2718],{"class":1027,"line":2714},45,[546,2716,2717],{"class":1031},"      return",[546,2719,1047],{"class":1046},[546,2721,2723],{"class":1027,"line":2722},46,[546,2724,2725],{"class":1046},"    }\n",[546,2727,2729,2732,2734],{"class":1027,"line":2728},47,[546,2730,2731],{"class":1046},"    window.parent.",[546,2733,1666],{"class":1069},[546,2735,1669],{"class":1046},[546,2737,2739,2742,2744],{"class":1027,"line":2738},48,[546,2740,2741],{"class":1046},"      task: ",[546,2743,1628],{"class":1042},[546,2745,1679],{"class":1046},[546,2747,2749,2752,2754,2756,2758,2760],{"class":1027,"line":2748},49,[546,2750,2751],{"class":1046},"      values: { imageId: imageId, subject: choice ",[546,2753,1899],{"class":1031},[546,2755,1902],{"class":1046},[546,2757,1905],{"class":1031},[546,2759,1908],{"class":1035},[546,2761,1911],{"class":1046},[546,2763,2765,2768,2770],{"class":1027,"line":2764},50,[546,2766,2767],{"class":1046},"    }, ",[546,2769,1703],{"class":1042},[546,2771,1706],{"class":1046},[546,2773,2775],{"class":1027,"line":2774},51,[546,2776,2777],{"class":1046},"  });\n",[546,2779,2781,2783,2785],{"class":1027,"line":2780},52,[546,2782,1318],{"class":1046},[546,2784,2474],{"class":1127},[546,2786,1131],{"class":1046},[22,2788],{},[25,2790,2792],{"id":2791},"try-it-in-the-console","Try it in the console",[12,2794,1638,2795,2798],{},[81,2796,204],{"href":202,"rel":2797},[85]," is the quickest way to test a template. Sign up there to get an API key, then:",[92,2800,2801,2811],{},[36,2802,2803,2804,2807,2808,205],{},"Open the ",[39,2805,2806],{},"Templates"," tab and press ",[39,2809,2810],{},"+ Create a template",[36,2812,2813],{},"Give it a name and paste in the HTML.",[12,2815,2816],{},[289,2817],{"alt":2818,"src":2819,"className":2820},"The console's Submit custom template form, with the template name filled in and the template HTML pasted into the HTML code box","/img/news/console-create-template.png",[293],[92,2822,2823,2829],{"start":896},[36,2824,2825,2826,2828],{},"If your instructions aren't baked into the HTML, paste them into ",[39,2827,2024],{}," (option 2 above).",[36,2830,2831,2832,2835,2836,205],{},"Press ",[39,2833,2834],{},"Preview",". Swap the sample values for real ones, such as an actual image URL, and press ",[39,2837,2838],{},"Re-render",[12,2840,2841],{},[289,2842],{"alt":2843,"src":2844,"className":2845},"The console's template preview, with sample values for image_url and image_id filled in above the rendered task","/img/news/console-preview-template.png",[293],[92,2847,2848],{"start":1268},[36,2849,2850,2851,2854],{},"Open the instructions button in the preview header to check how they read. Then answer the task and press ",[39,2852,2853],{},"Submit",". The console shows exactly the result you'd get back from a real job.",[12,2856,2857],{},[289,2858],{"alt":2859,"src":2860,"className":2861},"The console preview after pressing Submit: a confirmation that nothing was posted or charged, the Answer as JSON, and the Result row exactly as the results API returns it","/img/news/console-preview-result.png",[293],[92,2863,2864],{"start":1288},[36,2865,2866,2867,205],{},"When you're happy with it, press ",[39,2868,2869],{},"Submit template",[12,2871,2872],{},"Your template is ready to use right away. Until our team approves it, the console marks it as unapproved:",[12,2874,2875],{},[289,2876],{"alt":2877,"src":2878,"className":2879},"The yellow caution banner the console shows on an unapproved template","/img/news/console-unapproved-banner.png",[293],[12,2881,2882,2883,2886],{},"Workers see a similar caution note on its tasks. Tick ",[39,2884,2885],{},"Request team approval"," to have us review it. Any later edit counts as a new template and needs approval again, so finish testing first.",[22,2888],{},[25,2890,2892],{"id":2891},"run-a-job","Run a job",[12,2894,2895,2896,2899,2900,2903,2904,2907],{},"On the ",[39,2897,2898],{},"Create Job"," tab, pick your template in the job configuration and set a reward per task. The ",[39,2901,2902],{},"CSV data input"," section then shows the columns that template requires. Use ",[39,2905,2906],{},"Download CSV template"," there to get a CSV with those required columns already in its header row, then add your task data and paste the CSV:",[1018,2909,2913],{"className":2910,"code":2911,"language":2912,"meta":234,"style":234},"language-csv shiki shiki-themes github-dark","image_id,image_url\nimg-001,https://example.com/cat.jpg\nimg-002,https://example.com/car.jpg\n","csv",[44,2914,2915,2923,2931],{"__ignoreMap":234},[546,2916,2917,2920],{"class":1027,"line":1028},[546,2918,2919],{"class":1046},"image_id,",[546,2921,2922],{"class":1031},"image_url\n",[546,2924,2925,2928],{"class":1027,"line":235},[546,2926,2927],{"class":1046},"img-001,",[546,2929,2930],{"class":1031},"https://example.com/cat.jpg\n",[546,2932,2933,2936],{"class":1027,"line":896},[546,2934,2935],{"class":1046},"img-002,",[546,2937,2938],{"class":1031},"https://example.com/car.jpg\n",[12,2940,2941],{},[289,2942],{"alt":2943,"src":2944,"className":2945},"The console's Create Job tab: job type, job configuration with the template, reward and job name, CSV data input listing the template's required columns, and the cost estimate with the Create job button","/img/news/console-create-job.png",[293],[12,2947,2948,2949,1905],{},"Then choose a mode under ",[39,2950,2951],{},"Job type",[654,2953,2954,2966],{},[657,2955,2956],{},[660,2957,2958,2960,2963],{},[663,2959],{},[663,2961,2962],{},"CSV mode",[663,2964,2965],{},"Survey mode",[673,2967,2968,2979,2990],{},[660,2969,2970,2973,2976],{},[678,2971,2972],{},"Each row gets",[678,2974,2975],{},"1 answer",[678,2977,2978],{},"Answers from N different workers",[660,2980,2981,2984,2987],{},[678,2982,2983],{},"Cost",[678,2985,2986],{},"reward × rows",[678,2988,2989],{},"reward × rows × N",[660,2991,2992,2995,2998],{},[678,2993,2994],{},"Best for",[678,2996,2997],{},"Labelling, tagging, transcription: one right answer per item",[678,2999,3000],{},"Opinions and preferences, or several answers per item to take a majority vote",[12,3002,3003,3004,3007],{},"For surveys, tick ",[39,3005,3006],{},"Restrict each worker to one task"," so that each row's answers really come from different people. Leave it off in CSV mode, where it limits the whole job to 25 rows.",[12,3009,3010,3011,3015],{},"The cost is charged when the job starts. If you cancel, you get back whatever hasn't been completed. New accounts start with no credits, so ask us on ",[81,3012,230],{"href":3013,"rel":3014},"https://discord.gg/effectai",[85]," for a top-up.",[22,3017],{},[25,3019,3021],{"id":3020},"get-your-results","Get your results",[12,3023,3024,3025,3028],{},"Results show up on the ",[39,3026,3027],{},"Jobs"," tab as workers finish, and you can download them as a CSV. Each result pairs the row you sent with the answer your page submitted:",[1018,3030,3034],{"className":3031,"code":3032,"language":3033,"meta":234,"style":234},"language-json shiki shiki-themes github-dark","{\n  \"input\": { \"image_id\": \"img-002\", \"image_url\": \"https://example.com/car.jpg\" },\n  \"result\": \"{\\\"task\\\":\\\"submit\\\",\\\"values\\\":{\\\"imageId\\\":\\\"img-002\\\",\\\"subject\\\":\\\"vehicle\\\",\\\"unclear\\\":false}}\"\n}\n","json",[44,3035,3036,3041,3070,3154],{"__ignoreMap":234},[546,3037,3038],{"class":1027,"line":1028},[546,3039,3040],{"class":1046},"{\n",[546,3042,3043,3046,3049,3052,3054,3057,3059,3062,3064,3067],{"class":1027,"line":235},[546,3044,3045],{"class":1035},"  \"input\"",[546,3047,3048],{"class":1046},": { ",[546,3050,3051],{"class":1035},"\"image_id\"",[546,3053,1145],{"class":1046},[546,3055,3056],{"class":1042},"\"img-002\"",[546,3058,1151],{"class":1046},[546,3060,3061],{"class":1035},"\"image_url\"",[546,3063,1145],{"class":1046},[546,3065,3066],{"class":1042},"\"https://example.com/car.jpg\"",[546,3068,3069],{"class":1046}," },\n",[546,3071,3072,3075,3077,3080,3083,3086,3088,3090,3092,3095,3097,3100,3102,3104,3106,3109,3111,3113,3115,3117,3119,3122,3124,3126,3128,3131,3133,3135,3137,3140,3142,3144,3146,3149,3151],{"class":1027,"line":896},[546,3073,3074],{"class":1035},"  \"result\"",[546,3076,1145],{"class":1046},[546,3078,3079],{"class":1042},"\"{",[546,3081,3082],{"class":1035},"\\\"",[546,3084,3085],{"class":1042},"task",[546,3087,3082],{"class":1035},[546,3089,1905],{"class":1042},[546,3091,3082],{"class":1035},[546,3093,3094],{"class":1042},"submit",[546,3096,3082],{"class":1035},[546,3098,3099],{"class":1042},",",[546,3101,3082],{"class":1035},[546,3103,1712],{"class":1042},[546,3105,3082],{"class":1035},[546,3107,3108],{"class":1042},":{",[546,3110,3082],{"class":1035},[546,3112,1716],{"class":1042},[546,3114,3082],{"class":1035},[546,3116,1905],{"class":1042},[546,3118,3082],{"class":1035},[546,3120,3121],{"class":1042},"img-002",[546,3123,3082],{"class":1035},[546,3125,3099],{"class":1042},[546,3127,3082],{"class":1035},[546,3129,3130],{"class":1042},"subject",[546,3132,3082],{"class":1035},[546,3134,1905],{"class":1042},[546,3136,3082],{"class":1035},[546,3138,3139],{"class":1042},"vehicle",[546,3141,3082],{"class":1035},[546,3143,3099],{"class":1042},[546,3145,3082],{"class":1035},[546,3147,3148],{"class":1042},"unclear",[546,3150,3082],{"class":1035},[546,3152,3153],{"class":1042},":false}}\"\n",[546,3155,3156],{"class":1027,"line":1236},[546,3157,3158],{"class":1046},"}\n",[12,3160,3161,3162,3165,3166,205],{},"The answer is inside ",[44,3163,3164],{},"result",", under ",[44,3167,1712],{},[12,3169,3170,3171,1738],{},"You can check this shape before posting a job by submitting a task in the console preview (step 5 of ",[81,3172,2792],{"href":3173},"#try-it-in-the-console",[22,3175],{},[25,3177,187],{"id":186},[12,3179,3180],{},"Want to join the next phase of Effect AI?",[12,3182,3183,193,3185,3189],{},[39,3184,192],{},[81,3186,3188],{"href":196,"rel":3187},[85],"Join the API early access"," to get test credits, or collaborate with us to build or validate high-quality datasets for your AI models and research initiatives.",[12,3191,3192,3194],{},[39,3193,210],{}," Partner with Effect AI on responsible, mission-aligned data initiatives across a wide range of domains, from AI development to social impact.",[22,3196],{},[1128,3198,3199],{},"html pre.shiki code .snl16, html code.shiki .snl16{--shiki-default:#F97583}html pre.shiki code .sDLfK, html code.shiki .sDLfK{--shiki-default:#79B8FF}html pre.shiki code .sU2Wk, html code.shiki .sU2Wk{--shiki-default:#9ECBFF}html pre.shiki code .s95oV, html code.shiki .s95oV{--shiki-default:#E1E4E8}html pre.shiki code .svObZ, html code.shiki .svObZ{--shiki-default:#B392F0}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html pre.shiki code .s4JwU, html code.shiki .s4JwU{--shiki-default:#85E89D}html pre.shiki code .sAwPA, html code.shiki .sAwPA{--shiki-default:#6A737D}",{"title":234,"searchDepth":235,"depth":235,"links":3201},[3202,3203,3210,3211,3212,3213],{"id":952,"depth":235,"text":953},{"id":996,"depth":235,"text":997,"children":3204},[3205,3206,3207,3208,3209],{"id":1012,"depth":896,"text":1013},{"id":1105,"depth":896,"text":1106},{"id":1649,"depth":896,"text":1650},{"id":1931,"depth":896,"text":1932},{"id":2028,"depth":896,"text":2029},{"id":2791,"depth":235,"text":2792},{"id":2891,"depth":235,"text":2892},{"id":3020,"depth":235,"text":3021},{"id":186,"depth":235,"text":187},"2026-10-06","How Effect AI task templates work, a worked example you can copy, and how to test it in the API console and run it as a job.","/img/news/news-header-task-templates.png",{"src":3216},{"head":3219},{"meta":3220},[3221,3223,3224],{"name":252,"content":3222},"Effect AI, Effect API, API console, task templates, HTML templates, microtasks, human in the loop, data labeling, developers, guide",{"name":255,"content":918},{"name":258,"content":259},{"title":925,"description":3215},"news/build-your-own-task-template","O8_Szg_SF-sQ2CbXJFYD79aL0fRIqd4cwBUII_FZi8s",{"id":3229,"title":3230,"author":7,"body":3231,"created":3405,"description":3406,"extension":244,"headerImage":245,"image":3407,"lastUpdated":3405,"meta":3409,"navigation":260,"path":3416,"published":260,"seo":3417,"stem":3418,"__hash__":3419},"news/news/building-connected-effect.md","Building a More Connected Effect",{"type":9,"value":3232,"toc":3395},[3233,3236,3239,3242,3246,3249,3252,3256,3259,3262,3266,3272,3275,3278,3281,3284,3288,3294,3297,3300,3304,3310,3313,3316,3321,3324,3329,3332,3338,3341,3345,3348,3351,3355,3358,3361,3364,3367,3369,3371,3373,3384,3389,3393],[12,3234,3235],{},"During Effect AI's July 15 Community Call, updates were shared on the Canva app, Effect API, Effect CLI, AI workers, community participation, and Alpha 6.",[12,3237,3238],{},"Each project is being developed as another way for people and technology to connect with Effect. Access is being opened to developers through the API and CLI, Effect workflows are being brought into familiar apps, and new ways of combining human and AI work are being tested.",[12,3240,3241],{},"Together, these projects are being shaped into a more connected Effect ecosystem.",[25,3243,3245],{"id":3244},"bringing-effect-workflows-into-canva","Bringing Effect Workflows Into Canva",[12,3247,3248],{},"An early version of the Effect Canva app was demonstrated during the call. Through the app, human feedback could be requested without a Canva project having to be moved into a separate review process. Designs could be submitted for clarity checks, clickability testing, comparisons, or general audience feedback, and structured responses could be returned through Effect.",[12,3250,3251],{},"Canva is being used as an early example of how Effect can be added to tools people already use. Instead of Effect being treated as a separate platform for posting and completing tasks, its feedback and validation tools could be built directly into other apps.",[25,3253,3255],{"id":3254},"the-effect-api-was-shown-in-action","The Effect API Was Shown in Action",[12,3257,3258],{},"A live walkthrough of the Effect API and console was shown. Through the console, API keys can be managed, task templates can be viewed, jobs can be posted, and completed results can be reviewed or exported. Several templates were shown, and existing templates can be used, or custom workflows can be created for a particular project or dataset.",[12,3260,3261],{},"Developers interested in testing the API or building Effect-powered workflows can sign up for API access here.",[25,3263,3265],{"id":3264},"testing-ai-workers-through-real-workflows","Testing AI Workers Through Real Workflows",[12,3267,3268],{},[289,3269],{"alt":3270,"src":3271},"AI Workers overview slide","/img/news/ai-workers.png",[12,3273,3274],{},"AI workers were another major topic during the call.",[12,3276,3277],{},"Within Effect, an AI worker is treated as a distinct type of worker rather than being presented as a human participant. The goal is to make its role clear, including the model or system behind its work, so requesters can understand where an output came from.",[12,3279,3280],{},"AI workers could be used to prepare tasks, organize information, route work, transcribe content, or complete an initial processing step. The results could then be reviewed, corrected, or validated by a human worker.",[12,3282,3283],{},"Automation is being used where it makes sense, but human oversight will still be needed wherever quality, context, and trust matter.",[25,3285,3287],{"id":3286},"direct-infrastructure-access-with-the-effect-cli","Direct Infrastructure Access With the Effect CLI",[12,3289,3290],{},[289,3291],{"alt":3292,"src":3293},"Effect CLI overview slide","/img/news/effect-cli.png",[12,3295,3296],{},"The Effect CLI was also demonstrated. Through the CLI, Effect can be accessed directly from the terminal instead of everything being managed through a visual interface. Account details, wallet balances, node status, network connections, and latency information was shown during the demonstration.",[12,3298,3299],{},"Through the API, Effect can be added to an application as a service. Through the CLI, more direct control over the network and its infrastructure can be provided. This access can be especially useful for testing, automation, node operation, and larger agent-based workflows.",[25,3301,3303],{"id":3302},"rethinking-the-effect-dao-name","Rethinking the Effect DAO Name",[12,3305,3306],{},[289,3307],{"alt":3308,"src":3309},"Renaming the Effect DAO slide","/img/news/dao-renaming.png",[12,3311,3312],{},"The name \"Effect DAO\" has been used for years, but a fairly narrow impression can be created by it. The word DAO is often associated with token voting, formal governance, and finished proposals. In reality, community participation usually begins much earlier. Ideas are shared, questions are asked, and proposals are improved through feedback before a vote is ever held.",[12,3314,3315],{},"We are considering a new name that better communicates that broader purpose. Two options were presented during the call: The Port and The Dock.",[12,3317,3318],{},[39,3319,3320],{},"The Port",[12,3322,3323],{},"The Port was presented as a central gathering place where discussions could be joined, proposals could be shared, and the network's direction could be shaped.",[12,3325,3326],{},[39,3327,3328],{},"The Dock",[12,3330,3331],{},"The Dock was presented as a more focused working space where ideas could be brought in, discussed, and prepared for their next step.",[12,3333,3334],{},[289,3335],{"alt":3336,"src":3337},"The Port vs The Dock comparison","/img/news/port-vs-dock.png",[12,3339,3340],{},"No final name was selected during the call. Both options were shared so a wider conversation could be started about what the space should represent.",[25,3342,3344],{"id":3343},"alpha-6-is-planned-for-august","Alpha 6 Is Planned for August",[12,3346,3347],{},"One of the most anticipated updates was also shared: Alpha 6 is planned to begin in August.",[12,3349,3350],{},"A gradual rollout is expected rather than one large opening. Alpha testers will be brought into the updated platform, testing will be carried out under real conditions, and tasks will be released in controlled batches. Existing Alpha testers will continue to be included, and new testers will be onboarded.",[25,3352,3354],{"id":3353},"one-connected-ecosystem","One Connected Ecosystem",[12,3356,3357],{},"Across the Canva app, API, CLI, AI workers, DAO, and Alpha 6, the same larger direction is being followed.",[12,3359,3360],{},"More ways are being created for developers, applications, AI systems, human workers, and community members to connect through Effect. Automation is being used where it helps, while human judgement and validation remain central.",[12,3362,3363],{},"The July community call showed how these pieces are beginning to come together. Through Alpha 6 and continued testing, more of that vision will be put into practice.",[12,3365,3366],{},"Developers who want to connect their own applications, AI agents, or workflows to the network can sign up for Effect API access.",[22,3368],{},[25,3370,187],{"id":186},[12,3372,3180],{},[12,3374,3375,193,3377,3381,3382,875],{},[39,3376,216],{},[81,3378,3380],{"href":219,"rel":3379},[85],"Sign Up"," to join the ",[39,3383,459],{},[12,3385,3386,3388],{},[39,3387,192],{}," Collaborate with us to build or validate high-quality datasets for your AI models and research initiatives.",[12,3390,3391,3194],{},[39,3392,210],{},[22,3394],{},{"title":234,"searchDepth":235,"depth":235,"links":3396},[3397,3398,3399,3400,3401,3402,3403,3404],{"id":3244,"depth":235,"text":3245},{"id":3254,"depth":235,"text":3255},{"id":3264,"depth":235,"text":3265},{"id":3286,"depth":235,"text":3287},{"id":3302,"depth":235,"text":3303},{"id":3343,"depth":235,"text":3344},{"id":3353,"depth":235,"text":3354},{"id":186,"depth":235,"text":187},"2026-07-23","Highlights from the July 2026 Community Call and the tools shaping what comes next, including the Canva app, Effect API, Effect CLI, AI workers, the Effect DAO rename, and Alpha 6.",{"src":3408},"/img/news/supercharging-ai.jpg",{"head":3410},{"meta":3411},[3412,3414,3415],{"name":252,"content":3413},"AI, community call, Canva, API, CLI, AI workers, Effect DAO, Alpha 6, human AI collaboration, decentralized",{"name":255,"content":256},{"name":258,"content":259},"/news/building-connected-effect",{"title":3230,"description":3406},"news/building-connected-effect","IImiuJxVxuz4GRHLghDQ9qFvM-lFGERhKBMGOCqLrD8",{"id":3421,"title":3422,"author":501,"body":3423,"created":3836,"description":3837,"extension":244,"headerImage":245,"image":3838,"lastUpdated":3836,"meta":3840,"navigation":260,"path":3848,"published":260,"seo":3849,"stem":3850,"__hash__":3851},"news/news/phase-5-wrapped.md","Phase 5 Wrapped: Growth, Challenges, and the Road to Phase 6",{"type":9,"value":3424,"toc":3825},[3425,3429,3432,3435,3438,3440,3583,3585,3589,3592,3597,3600,3603,3607,3610,3613,3615,3619,3622,3639,3641,3645,3648,3662,3665,3674,3676,3680,3683,3687,3690,3693,3697,3700,3704,3707,3710,3713,3717,3720,3727,3729,3733,3736,3753,3755,3759,3762,3779,3782,3784,3788,3791,3794,3797,3799,3801,3803,3814,3819,3823],[25,3426,3428],{"id":3427},"reflecting-on-phase-5-of-the-effect-alpha","Reflecting on Phase 5 of the Effect Alpha",[12,3430,3431],{},"Phase 5 was the longest alpha phase we have run to date, and that was by design. Rather than cycling through phases quickly, we wanted to stress test the longevity of our systems, observe how the platform behaved over months of continuous operation, and push our infrastructure in ways that shorter phases simply cannot reveal. What we learned in that time was invaluable.",[12,3433,3434],{},"This phase brought real growth in worker participation, the introduction of two brand new task types, important infrastructure challenges that forced us to build better solutions, and a sharper focus on data quality than any phase before it. It also pushed us to be more honest about what needs to improve before Effect AI is ready to scale further.",[12,3436,3437],{},"Below is a full look at what we accomplished, what we faced, and what it all means for the future of the Effect platform.",[22,3439],{},[3441,3442,3443,3575],"center",{},[654,3444,3445,3466],{},[657,3446,3447],{},[660,3448,3449,3451,3454,3457,3460,3463],{},[663,3450],{},[663,3452,3453],{"align":3441},"Phase 1",[663,3455,3456],{"align":3441},"Phase 2",[663,3458,3459],{"align":3441},"Phase 3",[663,3461,3462],{"align":3441},"Phase 4",[663,3464,3465],{"align":3441},"Phase 5",[673,3467,3468,3490,3511,3531,3553],{},[660,3469,3470,3475,3478,3481,3484,3487],{},[678,3471,3472],{},[39,3473,3474],{},"Workers Onboarded",[678,3476,3477],{"align":3441},"20",[678,3479,3480],{"align":3441},"35",[678,3482,3483],{"align":3441},"48",[678,3485,3486],{"align":3441},"77",[678,3488,3489],{"align":3441},"100+",[660,3491,3492,3497,3500,3503,3505,3508],{},[678,3493,3494],{},[39,3495,3496],{},"Task Types (Total)",[678,3498,3499],{"align":3441},"3",[678,3501,3502],{"align":3441},"6",[678,3504,1229],{"align":3441},[678,3506,3507],{"align":3441},"10",[678,3509,3510],{"align":3441},"12",[660,3512,3513,3518,3520,3523,3525,3528],{},[678,3514,3515],{},[39,3516,3517],{},"Datasets Worked On",[678,3519,1281],{"align":3441},[678,3521,3522],{"align":3441},"4",[678,3524,1178],{"align":3441},[678,3526,3527],{"align":3441},"5+",[678,3529,3530],{"align":3441},"6+",[660,3532,3533,3538,3541,3544,3547,3550],{},[678,3534,3535],{},[39,3536,3537],{},"Tasks Completed",[678,3539,3540],{"align":3441},"5,200+",[678,3542,3543],{"align":3441},"42,000+",[678,3545,3546],{"align":3441},"57,000+",[678,3548,3549],{"align":3441},"90,000+",[678,3551,3552],{"align":3441},"81,000+",[660,3554,3555,3560,3563,3566,3569,3572],{},[678,3556,3557],{},[39,3558,3559],{},"EFFECT Paid Out",[678,3561,3562],{"align":3441},"15,000+ EFFECT",[678,3564,3565],{"align":3441},"80,000+ EFFECT",[678,3567,3568],{"align":3441},"185,000+ EFFECT",[678,3570,3571],{"align":3441},"250,000+ EFFECT",[678,3573,3574],{"align":3441},"730,000+ EFFECT",[540,3576,3578,3579],{"align":3441,"style":3577},"margin-top:16px; padding:12px; background:#E8F5E9; border-radius:8px;","\n  ",[39,3580,3582],{"style":3581},"font-size:20px; color:#2E7D32;","Total EFFECT Paid Out Across All Phases: 1,260,000+",[22,3584],{},[25,3586,3588],{"id":3587},"phase-5-new-task-types-and-expanded-datasets","Phase 5: New Task Types and Expanded Datasets",[12,3590,3591],{},"One of the headline achievements of Phase 5 was the introduction of two entirely new task types that significantly expanded what contributors can do on the platform.",[3593,3594,3596],"h4",{"id":3595},"image-annotation-with-bounding-boxes","Image Annotation with Bounding Boxes",[12,3598,3599],{},"Image annotation arrived in Phase 5 as a major addition to our template library. Workers draw precise bounding boxes around objects in images, label them accordingly, and the system captures both the spatial data and the labels. This task type opens the door to computer vision datasets and quality control workflows, areas that represent a significant and growing portion of AI training demand.",[12,3601,3602],{},"We also began development and testing of the corresponding image annotation validation task, giving the platform an end-to-end workflow for this data type. Validation tasks were actively tested toward the end of the phase, with worker feedback helping us refine the template, particularly on mobile devices.",[3593,3604,3606],{"id":3605},"novela-dataset","Novela Dataset",[12,3608,3609],{},"We launched the Novela dataset during Phase 5 as a complementary addition to our existing Common Voice work. Novela tasks are built around creating recordings of human-read audio paired with sentences drawn from soap opera and novela scripts. This work directly supports the training of text-to-speech models and helps AI systems learn the natural rhythms, emotion, and variation of human conversational speech.",[12,3611,3612],{},"Unlike sentence creation tasks, Novela tasks do not require contributors to write anything themselves, which makes them more accessible and significantly reduces the risk of AI-generated submissions compromising the dataset. The Novela workflow includes both recording and validation steps, giving us a complete and quality-controlled pipeline.",[22,3614],{},[25,3616,3618],{"id":3617},"platform-and-worker-experience-improvements","Platform and Worker Experience Improvements",[12,3620,3621],{},"Phase 5 saw a steady stream of improvements to the worker experience, many of them driven directly by feedback from alpha testers.",[33,3623,3624,3627,3630,3633,3636],{},[36,3625,3626],{},"A TaskPosterStatus component was added, giving workers real-time visibility into which tasks are active, queued, or coming up next. This removed a lot of the guesswork around when and what to work on.",[36,3628,3629],{},"Task qualification requirements are now displayed directly in the Task Poster Status area, so workers know exactly what capabilities they need before they start.",[36,3631,3632],{},"A scheduling system was introduced behind the scenes, allowing task requesters to queue tasks for specific times. Workers now see a predictable pipeline rather than tasks appearing without notice.",[36,3634,3635],{},"Skipping or reporting a task now prompts workers to provide brief feedback on what went wrong, giving the team much more useful signal for debugging and improvement.",[36,3637,3638],{},"A regular task posting schedule was introduced toward the end of the phase, with tasks going live on Mondays and Thursdays. Predictability matters to contributors, and this was a direct response to community feedback.",[22,3640],{},[25,3642,3644],{"id":3643},"new-capability-tests-and-the-alpha-support-portal","New Capability Tests and the Alpha Support Portal",[12,3646,3647],{},"Phase 5 marked a significant shift in how we assess and match workers to tasks. The original language test that had been in place since the start of the Alpha was retired in favor of four new, purpose-built capability assessments.",[33,3649,3650,3653,3656,3659],{},[36,3651,3652],{},"English Language 2.0",[36,3654,3655],{},"Common Voice Contributor",[36,3657,3658],{},"Common Voice Validator",[36,3660,3661],{},"Music Transcription (with Music Transcription Validation available depending on results)",[12,3663,3664],{},"These tests are timed and attempt-limited, reflecting the fact that the data being created is destined for real-world applications. Unlike earlier assessments, there are no unlimited retries. This raises the bar for participation and helps ensure that the workers completing tasks are genuinely prepared for them.",[12,3666,3667,3668,3673],{},"Alongside the new tests, we launched the ",[81,3669,3672],{"href":3670,"rel":3671},"https://sites.google.com/effect.ai/guides",[85],"Alpha Testing Support Portal",", a dedicated resource hub where contributors can find task guides, capability documentation, and platform instructions all in one place. The portal has already proven useful in reducing support questions and helping workers understand what is expected of them before they begin.",[22,3675],{},[25,3677,3679],{"id":3678},"challenges-we-faced-and-how-we-responded","Challenges We Faced and How We Responded",[12,3681,3682],{},"Phase 5 was our most transparent phase yet when it came to confronting platform issues head on. Several significant challenges emerged, and we want to be open about what happened and how we addressed them.",[3593,3684,3686],{"id":3685},"payment-pool-and-security-vulnerability","Payment Pool and Security Vulnerability",[12,3688,3689],{},"Midway through the phase, we identified a critical issue with the payments and payment pool system that required us to temporarily pause tasks while we investigated. We discovered and patched a security vulnerability in the payment pipeline. Tasks resumed once the fix was confirmed, and all worker balances were preserved.",[12,3691,3692],{},"This incident surfaced a deeper architectural issue: the Manager Node was becoming a bottleneck, especially under high simultaneous claim loads. Our response was to begin developing AI payment provers, a solution that offloads payment verification from the Manager Node and distributes it as tasks that workers can run directly on their own systems. This is a meaningful step toward a more resilient and decentralized architecture.",[3593,3694,3696],{"id":3695},"manager-node-instability","Manager Node Instability",[12,3698,3699],{},"Related to the above, the Manager Node experienced more frequent downtime than in previous phases. This was caused by a combination of factors including attempts by bad actors to circumvent the payment pipeline and the sheer volume of simultaneous claims. We implemented safeguards and continued monitoring throughout the phase, and the work done here will directly inform the infrastructure improvements planned for Phase 6.",[3593,3701,3703],{"id":3702},"data-quality-and-llm-abuse","Data Quality and LLM Abuse",[12,3705,3706],{},"Phase 5 brought our most serious data quality challenge to date. During a review of Common Voice sentence creation submissions, we identified clear patterns of workers using large language models to generate sentences rather than writing them from their own knowledge and experience.",[12,3708,3709],{},"This is a fundamental violation of what these tasks are for. The entire purpose of human-generated sentence datasets is to capture authentic, natural human language. AI-generated sentences undermine that and reduce the value of the data for the organizations and researchers who depend on it.",[12,3711,3712],{},"Our response was firm. We paused affected tasks, conducted thorough reviews, revoked capabilities, and issued bans for confirmed violations. We also introduced the Novela dataset as a task type that does not involve sentence writing at all, which naturally reduces the opportunity for this kind of abuse. Moving forward, we are building additional safeguards to detect and prevent LLM usage in tasks where human authorship is required.",[3593,3714,3716],{"id":3715},"multi-device-account-fragmentation","Multi-Device Account Fragmentation",[12,3718,3719],{},"We saw an increase in reports from workers experiencing what appeared to be account resets. The root cause was our lack of cross-device and cross-browser support. Workers switching between devices were effectively creating separate worker accounts with separate balances, leading to confusion and, in some cases, unclaimed rewards.",[12,3721,3722,3723,3726],{},"We communicated this clearly and added a notice directly to the worker homepage. The longer-term fix, ",[39,3724,3725],{},"state syncing",", is already in development and will be one of the headline features of Phase 6.",[22,3728],{},[25,3730,3732],{"id":3731},"what-we-learned-from-phase-5","What We Learned From Phase 5",[12,3734,3735],{},"Across the full run of Phase 5, a number of clear themes emerged that will shape how we build and operate going forward.",[33,3737,3738,3741,3744,3747,3750],{},[36,3739,3740],{},"Running a longer phase was the right call. The issues we uncovered around infrastructure stability, payment bottlenecks, and data quality patterns would not have been visible in a shorter cycle.",[36,3742,3743],{},"Data quality enforcement has to be proactive, not reactive. The LLM abuse we encountered was a wake-up call. We are investing in better detection, clearer instructions, and task designs that make abuse harder.",[36,3745,3746],{},"Predictability improves contribution quality. The introduction of a regular posting schedule and better task visibility immediately had a positive effect on worker engagement and output consistency.",[36,3748,3749],{},"Our architecture needs to distribute load better. The Manager Node bottleneck is a known limitation that we are actively solving. AI payment provers and state syncing are both steps in that direction.",[36,3751,3752],{},"The community is resilient and committed. Despite the challenges of this phase, the majority of our workers showed up consistently, followed the rules, and helped us push the platform forward.",[22,3754],{},[25,3756,3758],{"id":3757},"looking-ahead-to-phase-6","Looking Ahead To Phase 6",[12,3760,3761],{},"Phase 6 will build directly on what Phase 5 revealed. Here is what you can expect:",[33,3763,3764,3767,3770,3773,3776],{},[36,3765,3766],{},"State syncing across devices and browsers, so workers can pick up where they left off without losing progress, capabilities, or rewards",[36,3768,3769],{},"Manager Node task posting fixes and broader infrastructure improvements to reduce downtime and bottlenecks",[36,3771,3772],{},"More exciting task types to expand the range of work available on the platform",[36,3774,3775],{},"API access (tentative), which would allow developers and organizations to interface with the Effect network programmatically",[36,3777,3778],{},"New capabilities and tests to match the task types coming in Phase 6, maintaining our commitment to quality and appropriate worker-task matching",[12,3780,3781],{},"The next phase will continue the steady, iterative approach that has defined the Alpha so far. Improving what works, fixing what does not, and expanding capability step by step.",[22,3783],{},[25,3785,3787],{"id":3786},"closing-thoughts","Closing Thoughts",[12,3789,3790],{},"Phase 5 pushed Effect forward in ways that were sometimes uncomfortable but always necessary. We uncovered real vulnerabilities, confronted difficult community situations head on, and came out of it with a clearer picture of what this platform needs to be.",[12,3792,3793],{},"Every worker who showed up, followed the instructions, and contributed honest work helped build something real. The datasets created during this phase are not test data. They are going into live applications. That is worth taking seriously, and most of our community does.",[12,3795,3796],{},"Thank you for being part of this. Phase 6 is coming, and we are more prepared for it than we have ever been.",[22,3798],{},[25,3800,187],{"id":186},[12,3802,3180],{},[12,3804,3805,193,3807,3381,3811,3813],{},[39,3806,216],{},[81,3808,3380],{"href":3809,"rel":3810},"https://forms.gle/ovZXKQBvemXVn5Gy5",[85],[39,3812,459],{}," and start earning $EFFECT for real tasks",[12,3815,3816,3818],{},[39,3817,192],{}," Collaborate with us to build or validate high-quality datasets for your AI models and research initiatives",[12,3820,3821,3194],{},[39,3822,210],{},[22,3824],{},{"title":234,"searchDepth":235,"depth":235,"links":3826},[3827,3828,3829,3830,3831,3832,3833,3834,3835],{"id":3427,"depth":235,"text":3428},{"id":3587,"depth":235,"text":3588},{"id":3617,"depth":235,"text":3618},{"id":3643,"depth":235,"text":3644},{"id":3678,"depth":235,"text":3679},{"id":3731,"depth":235,"text":3732},{"id":3757,"depth":235,"text":3758},{"id":3786,"depth":235,"text":3787},{"id":186,"depth":235,"text":187},"2026-04-24","A deep look at the milestones, challenges, and progress made during Phase 5 of the Effect Alpha.",{"src":3839},"/img/news/BlogBG-08.png",{"head":3841},{"meta":3842},[3843,3845,3847],{"name":252,"content":3844},"AI, alpha, testing, solana, effect, data, human AI collaboration, decentralized, effect alpha, datasets, image annotation, novela, capability tests, data quality",{"name":255,"content":3846},"Effect AI",{"name":258,"content":259},"/news/phase-5-wrapped",{"title":3422,"description":3837},"news/phase-5-wrapped","oOQ0tlSQHWvZv-pLqSxShzCOaxwvE72KwCrfJVxl5mk",{"id":3853,"title":3854,"author":3846,"body":3855,"created":4022,"description":4023,"extension":244,"headerImage":245,"image":4024,"lastUpdated":4022,"meta":4026,"navigation":260,"path":4033,"published":260,"seo":4034,"stem":4035,"__hash__":4036},"news/news/march-platform-updates.md","March Platform Updates: Worker Experience, New Templates, and Infrastructure Advances",{"type":9,"value":3856,"toc":4014},[3857,3860,3864,3871,3878,3881,3883,3887,3894,3897,3938,3941,3948,3950,3954,3961,3963,3967,3974,3977,3979,3983,3986,3989,3991,3993,3995,4004,4008,4012],[12,3858,3859],{},"March has been a strong month for Effect. We shipped several worker-facing improvements, expanded our task template library, launched a new public dataset, and made critical progress on infrastructure that will enable smoother scaling. Here's what landed.",[25,3861,3863],{"id":3862},"worker-experience-gets-better","Worker Experience Gets Better",[12,3865,3866,3867,3870],{},"We added a ",[39,3868,3869],{},"TaskPosterStatus component"," that gives workers live visibility into task availability across the platform. This removes the guesswork: contributors can now see which tasks are active, queued, or coming soon. Real-time task status helps workers plan their time and reduces friction in the discovery process.",[12,3872,3873,3874,3877],{},"This is paired with a new ",[39,3875,3876],{},"scheduling system"," behind the scenes that allows task requesters to queue tasks for specific times. Workers benefit immediately: they see the upcoming pipeline and can prepare for high-value work. Task requesters get better control over when their work goes live.",[12,3879,3880],{},"Together, these changes signal a shared goal: make the platform predictable for everyone involved.",[22,3882],{},[25,3884,3886],{"id":3885},"new-templates-and-datasets","New Templates and Datasets",[12,3888,3889,3890,3893],{},"Our task template library now includes a ",[39,3891,3892],{},"new image annotation tool with bounding boxes",": a powerful addition for computer vision datasets and quality control workflows. Workers draw precise bounding boxes around objects in images, label them, and the system captures both the spatial data and the labels. The tool integrates seamlessly with our existing validation architecture.",[12,3895,3896],{},"Here's our complete template roster:",[33,3898,3899,3905,3911,3917,3926,3932],{},[36,3900,3901,3904],{},[39,3902,3903],{},"Sentence Generation",": Free-form text creation",[36,3906,3907,3910],{},[39,3908,3909],{},"Sentence Recording",": Audio capture with quality validation",[36,3912,3913,3916],{},[39,3914,3915],{},"Image Classification and Metadata Creation",": Multi-label classification",[36,3918,3919,193,3921,3925],{},[39,3920,3596],{},[3922,3923,3924],"em",{},"(new)",": Precise spatial annotation",[36,3927,3928,3931],{},[39,3929,3930],{},"Music Lyric Transcription",": Audio to text",[36,3933,3934,3937],{},[39,3935,3936],{},"Full Song Lyric Transcription Validation",": Multi-stage verification",[12,3939,3940],{},"All templates include built-in validation steps to ensure data quality.",[12,3942,3943,3944,3947],{},"We also introduced the ",[39,3945,3946],{},"Novela dataset",", a significant addition that expands the range of work available to contributors. Novela is built around creating recordings of human-read audio paired with sentences, helping train text-to-speech models and LLMs on natural human conversation speech. This work is foundational for building AI systems that sound and respond like real people. Contributors working on Novela are directly supporting the next generation of conversational AI.",[22,3949],{},[25,3951,3953],{"id":3952},"website-gets-a-dedicated-blog-space","Website Gets a Dedicated Blog Space",[12,3955,3956,3957,3960],{},"Our news and updates are growing. For a while, we featured blog posts in a carousel on our homepage, but that format wasn't giving them the space and visibility they deserved. We've now built a ",[39,3958,3959],{},"dedicated blogs page"," for the Effect AI website with a featured section highlighting the latest post. This makes it easier for the community to discover, read, and follow platform updates, learnings, and announcements. It also signals to the outside world that we're invested in sharing what we're building and learning along the way.",[22,3962],{},[25,3964,3966],{"id":3965},"state-syncing-coming-soon","State Syncing (Coming Soon)",[12,3968,3969,3970,3973],{},"Behind the scenes, we're laying groundwork for seamless multi-device support. Our upcoming ",[39,3971,3972],{},"state syncing feature"," will eliminate the fragmentation that happens when workers use different devices or browsers. Right now, switching from a phone to a desktop means starting fresh as a different worker account. With state syncing, your worker capabilities, task progress, rewards, qualifications, and all other metrics will sync automatically across platforms. Start a task on your phone during a commute, pick it up on your laptop at home, and resume on a tablet later, all without losing context or progress.",[12,3975,3976],{},"This is a foundational piece for a truly permissionless network: contributors should be able to pick up work from anywhere, on any device, without friction. The feature is currently in internal testing and expected to deploy within Q2.",[22,3978],{},[25,3980,3982],{"id":3981},"what-this-means","What This Means",[12,3984,3985],{},"March's updates reflect a consistent theme: lowering friction for both workers and task requesters while building the infrastructure that makes Effect scalable and reliable.",[12,3987,3988],{},"Workers get better visibility, more task variety, and smoother experiences across devices. Task requesters get more control and confidence in their data pipelines. The platform gets stronger infrastructure to support growth.",[22,3990],{},[25,3992,187],{"id":186},[12,3994,3180],{},[12,3996,3997,193,3999,3381,4002,875],{},[39,3998,216],{},[81,4000,3380],{"href":3809,"rel":4001},[85],[39,4003,459],{},[12,4005,4006,3388],{},[39,4007,192],{},[12,4009,4010,3194],{},[39,4011,210],{},[22,4013],{},{"title":234,"searchDepth":235,"depth":235,"links":4015},[4016,4017,4018,4019,4020,4021],{"id":3862,"depth":235,"text":3863},{"id":3885,"depth":235,"text":3886},{"id":3952,"depth":235,"text":3953},{"id":3965,"depth":235,"text":3966},{"id":3981,"depth":235,"text":3982},{"id":186,"depth":235,"text":187},"2026-04-01","This month brought worker UI improvements, new task templates, the Novela dataset, a dedicated blogs page, and key infrastructure enhancements to support scale.",{"src":4025},"/img/news/BlogBG-07.jpg",{"head":4027},{"meta":4028},[4029,4031,4032],{"name":252,"content":4030},"updates, worker experience, task templates, datasets, platform improvements, scheduling, state syncing",{"name":255,"content":3846},{"name":258,"content":259},"/news/march-platform-updates",{"title":3854,"description":4023},"news/march-platform-updates","ESUAnnVrJ1G70DKTp_zCYCL3fvX2EG07_a4ZS9ONxRI",{"id":4038,"title":4039,"author":501,"body":4040,"created":4185,"description":4186,"extension":244,"headerImage":245,"image":4187,"lastUpdated":4185,"meta":4189,"navigation":260,"path":4196,"published":260,"seo":4197,"stem":4198,"__hash__":4199},"news/news/machines-still-need-us.md","The Machines Still Need Us: Why Effect AI Is More Critical Than Ever in the Age of LLMs and AI Agents",{"type":9,"value":4041,"toc":4176},[4042,4046,4049,4052,4055,4058,4060,4064,4067,4070,4073,4076,4078,4082,4085,4088,4091,4094,4096,4100,4103,4106,4109,4112,4114,4118,4121,4124,4127,4130,4133,4135,4139,4142,4145,4148,4151,4153,4155,4157,4166,4170,4174],[25,4043,4045],{"id":4044},"the-promise-vs-the-reality","The Promise vs. The Reality",[12,4047,4048],{},"Everyone is talking about AI agents. Autonomous systems that browse the web, write code, manage your calendar, close your deals, and maybe one day run your company. The hype is deafening, and some of it is genuinely deserved. Large language models have leaped from novelty to necessity in under three years. AI agents are being woven into the fabric of how businesses operate, how developers build, and how decisions get made.",[12,4050,4051],{},"But here's what doesn't make the headlines: none of it works without humans.",[12,4053,4054],{},"Not because the models aren't impressive. They are. But because intelligence, real, reliable, contextually grounded intelligence, doesn't come from compute alone. It comes from data. And data comes from people. Specifically, it comes from millions of people around the world annotating, labeling, validating, moderating, and fine-tuning the outputs of these models every single day.",[12,4056,4057],{},"That human layer is the foundation everything else is built on. And right now, it's largely invisible, often undervalued, and almost entirely controlled by centralized intermediaries. Effect AI was built to change that.",[22,4059],{},[25,4061,4063],{"id":4062},"the-llm-economy-is-built-on-a-foundation-of-human-intelligence","The LLM Economy Is Built on a Foundation of Human Intelligence",[12,4065,4066],{},"Every frontier model you've ever used, ChatGPT, Claude, Gemini, or any of the agents being built on top of them, exists because humans taught it how to think. Not through code alone, but through millions of decisions made by real people: labeling images, ranking responses, annotating text, flagging errors, and shaping the feedback loops that turn raw compute into something useful.",[12,4068,4069],{},"This is not a historical footnote. It is an ongoing reality. The AI industry's appetite for human intelligence doesn't shrink as models get smarter. It grows. More capable models require more nuanced training data, more precise feedback, and more sophisticated human judgment to keep them aligned with how the real world actually works.",[12,4071,4072],{},"AI agents make this even more pronounced. Unlike a model that answers a single question and stops, an agent takes a sequence of actions autonomously to complete a goal. Every new capability unlocked by agentic AI introduces new edge cases, new failure points, and new demands for the kind of contextual human understanding that no model can fully replicate on its own. Human intelligence isn't a stepping stone on the way to fully autonomous AI. For the foreseeable future, it's a permanent and essential layer of the stack.",[12,4074,4075],{},"The question the industry hasn't fully answered yet is how to source that intelligence fairly, transparently, and at scale. That's the problem Effect AI was built to solve.",[22,4077],{},[25,4079,4081],{"id":4080},"enter-effect-ai-decentralized-intelligence-infrastructure","Enter Effect AI: Decentralized Intelligence Infrastructure",[12,4083,4084],{},"Effect AI is a decentralized protocol built around one core idea: that human intelligence is infrastructure, and infrastructure should be open, transparent, and owned by the people who participate in it.",[12,4086,4087],{},"The protocol is built around three types of nodes that work together to move tasks through the network seamlessly. Worker nodes are the people on the ground, anyone in the world who connects to the network and completes AI tasks in exchange for EFFECT tokens. Manager nodes keep things organized, handling the flow of work between those who need tasks done and the workers available to do them. Provider nodes are the businesses and developers on the other side, the ones publishing tasks, defining what needs to be done, and staking EFFECT tokens as payment.",[12,4089,4090],{},"What makes this powerful is that no single company sits in the middle controlling any of it. Every task, every submission, and every payment flows through the protocol transparently. The rules are the same for everyone, they're written into the network itself, and anyone can participate without needing approval from a gatekeeper.",[12,4092,4093],{},"The result is an AI data network that is faster, fairer, and more accountable than anything the centralized model can offer.",[22,4095],{},[25,4097,4099],{"id":4098},"how-it-works-putting-human-intelligence-on-chain","How It Works: Putting Human Intelligence On-Chain",[12,4101,4102],{},"When a company needs human intelligence to power their AI, the process is straightforward. They publish a campaign, stake EFFECT tokens as payment, and define exactly what needs to be done. From there, they can open the task up to the entire global workforce, or get more precise by selecting workers with specific capabilities. They can even create their own qualification tests for workers to complete before accessing the task, making sure the right people are doing the right work.",[12,4104,4105],{},"Once a task is live, workers anywhere in the world can pick it up through their browser, no approval process, no gatekeepers. They complete the work, and depending on the campaign's requirements, it can move through multiple steps and validation rounds before it's finalized. All of this is recorded and settled on-chain with payment automatically released to worker wallets once approved. The smart contract is the guarantee.",[12,4107,4108],{},"For workers, this opens up real economic opportunity. In regions where traditional banking infrastructure is thin and platforms like Amazon Mechanical Turk restrict participation, Effect AI offers a permissionless on-ramp to global AI work. Every rule, every payout, and every rejection is trackable on-chain, and workers can even earn passive rewards simply for keeping their node online.",[12,4110,4111],{},"For businesses and developers, it delivers something centralized platforms fundamentally cannot: trustless, transparent, auditable data pipelines where every task, submission, and validation is recorded on-chain.",[22,4113],{},[25,4115,4117],{"id":4116},"the-role-effect-ai-plays-in-the-ai-agent-economy","The Role Effect AI Plays in the AI Agent Economy",[12,4119,4120],{},"Here's where things get genuinely exciting for builders in the Web3 and DePIN space.",[12,4122,4123],{},"AI agents are not a finished product. They are living systems that degrade without maintenance, drift without correction, and fail without oversight. Every agentic system deployed in production today needs a continuous supply of human intelligence to stay accurate and useful. That means human feedback to keep models aligned with real world intent, expert review of the edge cases that automated checks simply can't catch, and specialized data for the specific verticals those agents operate in. As agents become more capable and take on more complex tasks, these requirements don't get smaller. They scale with the ambition of the systems being built.",[12,4125,4126],{},"This isn't a one-time expense. It's a continuous operational requirement. And as the number of deployed AI agents grows from thousands to millions, the demand for this infrastructure doesn't shrink. It compounds.",[12,4128,4129],{},"Effect AI is the DePIN layer for human intelligence in the AI stack. Just as Render Network decentralizes GPU compute and Helium decentralizes wireless infrastructure, Effect AI decentralizes the human-in-the-loop work that every AI system requires. We are not competing with AI. We are the infrastructure that makes AI production-ready.",[12,4131,4132],{},"Think of it this way: the AI agent economy needs three things to function, namely compute, models, and human intelligence. The first two are rapidly being commoditized and decentralized. Effect AI is building the decentralized marketplace for the third.",[22,4134],{},[25,4136,4138],{"id":4137},"the-bottom-line","The Bottom Line",[12,4140,4141],{},"The LLM era didn't make human intelligence less valuable. It made it more valuable and more in demand than ever before. The AI agent economy is being built on top of continuous human feedback loops, and the world is beginning to recognize that the human layer underneath AI deserves to be treated with the same openness, fairness, and transparency that Web3 was built to deliver.",[12,4143,4144],{},"Effect AI is the Web3 answer to a Web2 problem that sits at the very heart of the AI revolution.",[12,4146,4147],{},"If you're a builder in the DePIN space, a DeFi native watching the AI agent economy emerge, or someone who believes that decentralized infrastructure always wins in the long run, Effect AI deserves your attention.",[12,4149,4150],{},"The machines still need us. Effect AI makes sure we get paid for it.",[22,4152],{},[25,4154,187],{"id":186},[12,4156,3180],{},[12,4158,4159,193,4161,3381,4164,875],{},[39,4160,216],{},[81,4162,3380],{"href":3809,"rel":4163},[85],[39,4165,459],{},[12,4167,4168,3388],{},[39,4169,192],{},[12,4171,4172,3194],{},[39,4173,210],{},[22,4175],{},{"title":234,"searchDepth":235,"depth":235,"links":4177},[4178,4179,4180,4181,4182,4183,4184],{"id":4044,"depth":235,"text":4045},{"id":4062,"depth":235,"text":4063},{"id":4080,"depth":235,"text":4081},{"id":4098,"depth":235,"text":4099},{"id":4116,"depth":235,"text":4117},{"id":4137,"depth":235,"text":4138},{"id":186,"depth":235,"text":187},"2026-03-17","As AI agents become the norm, the human intelligence layer underneath them has never been more essential. Here's why Effect AI sits at the center of it all.",{"src":4188},"/img/news/BlogBG-06.jpg",{"head":4190},{"meta":4191},[4192,4194,4195],{"name":252,"content":4193},"AI agents, LLMs, human intelligence, decentralized, effect AI, EFFECT token, DePIN, data labeling, annotation, Solana, Web3, AI infrastructure, human-in-the-loop, worker nodes, AI data",{"name":255,"content":918},{"name":258,"content":259},"/news/machines-still-need-us",{"title":4039,"description":4186},"news/machines-still-need-us","BYleJCb-UWIG-jvW-7zKsW_WzQz3X8tRCAUJby8Itmk",{"id":4201,"title":4202,"author":501,"body":4203,"created":4356,"description":4357,"extension":244,"headerImage":245,"image":4358,"lastUpdated":4356,"meta":4360,"navigation":260,"path":4367,"published":260,"seo":4368,"stem":4369,"__hash__":4370},"news/news/effectai-in-2025.md","Effect AI in 2025: A Year of Momentum, Testing, and Community-Driven Progress",{"type":9,"value":4204,"toc":4346},[4205,4208,4211,4215,4218,4221,4225,4228,4231,4245,4248,4251,4255,4258,4261,4265,4268,4271,4274,4278,4281,4284,4288,4291,4294,4308,4312,4315,4318,4321,4323,4325,4327,4336,4340,4344],[12,4206,4207],{},"2025 was a pivotal year for Effect. It marked the transition from long-term groundwork into active, hands-on testing with a growing community of contributors. Major infrastructure milestones landed, the ecosystem was simplified through a unified token, and the platform began proving itself through real usage rather than speculation.",[12,4209,4210],{},"Most importantly, 2025 was about learning. The Alpha is still in full effect, and everything built this year reflects real data, real feedback, and real human work happening on the network.",[25,4212,4214],{"id":4213},"a-unified-token-and-a-cleaner-foundation","A unified token and a cleaner foundation",[12,4216,4217],{},"At the start of the year, Effect completed its migration to a single EFFECT token on Solana. This snapshot-based transition consolidated the ecosystem and created a cleaner foundation for the platform moving forward.",[12,4219,4220],{},"The migration enabled faster iteration, lower transaction costs, and a clearer incentive structure for participants. It also set the stage for scaling activity during Alpha testing without exposing users to unnecessary complexity.",[25,4222,4224],{"id":4223},"alpha-testing-at-real-scale","Alpha testing at real scale",[12,4226,4227],{},"Throughout 2025, Effect ran multiple Alpha phases and steadily expanded access. These were not closed demos or synthetic tests. They were live environments where contributors completed real microtasks and interacted with the platform as intended.",[12,4229,4230],{},"Over the course of the year:",[33,4232,4233,4236,4239,4242],{},[36,4234,4235],{},"Hundreds of thousands of tasks were completed during Alpha testing",[36,4237,4238],{},"Close to a hundred active testers participated across multiple task types",[36,4240,4241],{},"Individual contributors completed thousands of tasks, helping stress-test throughput and quality systems",[36,4243,4244],{},"Task execution, validation, and payment flows were exercised repeatedly under real usage patterns",[12,4246,4247],{},"These metrics gave the team concrete insight into what works, what breaks, and where the platform needs refinement. Feedback from Alpha directly influenced improvements to task instructions, dashboards, onboarding, and worker experience.",[12,4249,4250],{},"Importantly, the Alpha is still ongoing. New task types, improved workflows, and additional testing waves are continuing as the platform evolves.",[25,4252,4254],{"id":4253},"quality-fairness-and-worker-experience","Quality, fairness, and worker experience",[12,4256,4257],{},"A major focus during Alpha has been ensuring that contributors are treated fairly and that high-quality work is rewarded. Throughout 2025, Effect tested and refined approaches to quality control, reviewer feedback, and incentive alignment.",[12,4259,4260],{},"Experiments around clearer task guidance, better validation flows, and early ideas such as base pay for availability helped surface valuable lessons. These learnings continue to shape how Effect balances scalability with worker trust and task accuracy.",[25,4262,4264],{"id":4263},"real-world-datasets-and-applied-use-cases","Real-world datasets and applied use cases",[12,4266,4267],{},"Rather than focusing on hypothetical use cases, Effect spent 2025 validating the platform through practical dataset work. Contributors supported public datasets such as Mozilla Common Voice, helping expand language coverage and improve training data quality.",[12,4269,4270],{},"In parallel, the team explored additional applied use cases internally to understand how the platform performs across different domains. These experiments helped validate that Effect can support nuanced, human-in-the-loop workflows where AI models alone are not enough.",[12,4272,4273],{},"Together, this work reinforced Effect’s core thesis: humans remain essential in AI pipelines, especially for validation, transcription, and contextual judgment.",[25,4275,4277],{"id":4276},"lower-friction-and-broader-access","Lower friction and broader access",[12,4279,4280],{},"Another recurring theme in 2025 was accessibility. Social login options were introduced to reduce onboarding friction, making it easier for non-crypto-native users to participate. Task flows were simplified, and the platform architecture was optimized to keep interactions responsive and predictable, even as testing activity increased.",[12,4282,4283],{},"These improvements helped increase tester retention and reduced the time it takes for new contributors to become productive during Alpha.",[25,4285,4287],{"id":4286},"openness-and-momentum-heading-into-2026","Openness and momentum heading into 2026",[12,4289,4290],{},"As the platform matured, Effect reaffirmed its commitment to openness. Plans to open source core components, publish updated documentation, and share clearer roadmaps reflect a long-term goal of building a protocol shaped by its community.",[12,4292,4293],{},"By the end of 2025, Effect had:",[33,4295,4296,4299,4302,4305],{},[36,4297,4298],{},"A unified token and streamlined infrastructure",[36,4300,4301],{},"An active Alpha with real usage metrics",[36,4303,4304],{},"Proven task flows tested by hundreds of contributors",[36,4306,4307],{},"Early dataset contributions demonstrating real-world value",[25,4309,4311],{"id":4310},"looking-ahead","Looking ahead",[12,4313,4314],{},"Effect enters the next phase with momentum and clarity. The Alpha remains active, with continued testing, refinement, and expansion ahead. The focus moving forward is simple: stabilize the core experience, onboard more contributors, expand task diversity, and keep learning from real usage.",[12,4316,4317],{},"If you are interested in contributing, experimenting, or following along as decentralized human intelligence takes shape, now is the time to get involved. Join the community, participate in the Alpha, and help shape the future of Effect.",[12,4319,4320],{},"The work is ongoing, and the best is still ahead.",[22,4322],{},[25,4324,187],{"id":186},[12,4326,3180],{},[12,4328,4329,193,4331,3381,4334,875],{},[39,4330,216],{},[81,4332,3380],{"href":3809,"rel":4333},[85],[39,4335,459],{},[12,4337,4338,3388],{},[39,4339,192],{},[12,4341,4342,3194],{},[39,4343,210],{},[22,4345],{},{"title":234,"searchDepth":235,"depth":235,"links":4347},[4348,4349,4350,4351,4352,4353,4354,4355],{"id":4213,"depth":235,"text":4214},{"id":4223,"depth":235,"text":4224},{"id":4253,"depth":235,"text":4254},{"id":4263,"depth":235,"text":4264},{"id":4276,"depth":235,"text":4277},{"id":4286,"depth":235,"text":4287},{"id":4310,"depth":235,"text":4311},{"id":186,"depth":235,"text":187},"2026-01-16","Throughout 2025, Effect moved from groundwork to real-world testing. This recap highlights the milestones, learnings, and community effort that defined the year.",{"src":4359},"/img/news/BlogBG-2025-Recap.jpg",{"head":4361},{"meta":4362},[4363,4365,4366],{"name":252,"content":4364},"2025, Recap, progress, AI, agents, machine, learning, natural, language, interfaces, alpha, testing, solana, effect",{"name":255,"content":918},{"name":258,"content":259},"/news/effectai-in-2025",{"title":4202,"description":4357},"news/effectai-in-2025","77umNqAdnMjCzyZFmdpt5zkZSxc31FUN-Xz9SWj0j3Y",{"id":4372,"title":4373,"author":501,"body":4374,"created":4693,"description":4694,"extension":244,"headerImage":245,"image":4695,"lastUpdated":4693,"meta":4696,"navigation":260,"path":4704,"published":260,"seo":4705,"stem":4706,"__hash__":4707},"news/news/raising-the-bar.md","Raising the Bar: What We Achieved in Alpha Phases 3 and 4",{"type":9,"value":4375,"toc":4683},[4376,4378,4382,4385,4388,4391,4393,4491,4493,4497,4500,4503,4523,4526,4528,4532,4535,4538,4541,4543,4547,4550,4553,4573,4576,4578,4582,4585,4602,4605,4607,4611,4614,4617,4643,4645,4647,4649,4652,4655,4658,4660,4662,4664,4673,4677,4681],[22,4377],{},[25,4379,4381],{"id":4380},"reflecting-on-phases-3-and-4-of-the-effect-alpha","Reflecting on Phases 3 and 4 of the Effect Alpha",[12,4383,4384],{},"As we continue building the future of decentralized human AI collaboration, Phases 3 and 4 of the Effect Alpha mark two of the most important chapters in our journey so far. These phases carried us from early infrastructure and workflow validation into large scale task execution, real dataset creation, and significant improvements to the overall platform experience.",[12,4386,4387],{},"Effect has always been driven by one goal. To create an open and accessible ecosystem where people and AI work together to power the next generation of intelligent systems. With each phase, we become more confident in that vision and more capable of delivering on it. These two phases represented a major shift in the Alpha. Workers took on more complex tasks, new datasets moved through the system, and the foundation of the future Effect network became more visible.",[12,4389,4390],{},"Below is a look back at what we achieved in Phases 3 and 4, what improved, and what these milestones mean for the future of the Effect platform.",[22,4392],{},[3441,4394,4395,4486],{},[654,4396,4397,4411],{},[657,4398,4399],{},[660,4400,4401,4403,4405,4407,4409],{},[663,4402],{},[663,4404,3453],{"align":3441},[663,4406,3456],{"align":3441},[663,4408,3459],{"align":3441},[663,4410,3462],{"align":3441},[673,4412,4413,4427,4442,4458,4472],{},[660,4414,4415,4419,4421,4423,4425],{},[678,4416,4417],{},[39,4418,3474],{},[678,4420,3477],{"align":3441},[678,4422,3480],{"align":3441},[678,4424,3483],{"align":3441},[678,4426,3486],{"align":3441},[660,4428,4429,4434,4436,4438,4440],{},[678,4430,4431],{},[39,4432,4433],{},"Task Types Introduced",[678,4435,3499],{"align":3441},[678,4437,3502],{"align":3441},[678,4439,1229],{"align":3441},[678,4441,3507],{"align":3441},[660,4443,4444,4449,4451,4453,4455],{},[678,4445,4446],{},[39,4447,4448],{},"Datasets Created or Verified",[678,4450,1281],{"align":3441},[678,4452,3522],{"align":3441},[678,4454,1178],{"align":3441},[678,4456,4457],{"align":3441},"5+ (incl. Mozilla Common Voice)",[660,4459,4460,4464,4466,4468,4470],{},[678,4461,4462],{},[39,4463,3537],{},[678,4465,3540],{"align":3441},[678,4467,3543],{"align":3441},[678,4469,3546],{"align":3441},[678,4471,3549],{"align":3441},[660,4473,4474,4478,4480,4482,4484],{},[678,4475,4476],{},[39,4477,3559],{},[678,4479,3562],{"align":3441},[678,4481,3565],{"align":3441},[678,4483,3568],{"align":3441},[678,4485,3571],{"align":3441},[540,4487,3578,4488],{"align":3441,"style":3577},[39,4489,4490],{"style":3581},"Total EFFECT Paid Out: 530,000+",[22,4492],{},[25,4494,4496],{"id":4495},"phase-3-building-the-core-foundation","Phase 3: Building The Core Foundation",[12,4498,4499],{},"Phase 3 was all about stress testing the foundation. During this period, we focused on improving the stability and performance of the core systems that power Effect. The goal was simple. Ensure the manager, worker clients, and task flows operate smoothly and consistently under increased load.",[12,4501,4502],{},"Key achievements in Phase 3:",[33,4504,4505,4508,4511,4514,4517,4520],{},[36,4506,4507],{},"Strengthened the task submission, validation, and reward pipeline",[36,4509,4510],{},"Expanded capability tests to match contributors to the right tasks",[36,4512,4513],{},"Established repeatable processes for posting and completing tasks at scale",[36,4515,4516],{},"Built the groundwork for external dataset collaborations and large scale task batches",[36,4518,4519],{},"First introduction of penalties (manually administered at this stage)",[36,4521,4522],{},"Improved platform reliability and reduced task failure rates",[12,4524,4525],{},"Phase 3 gave us the confidence to push the platform further. By the end of it, we had a stable worker base, consistent task throughput, and an improving validation process. This set the stage for Phase 4.",[22,4527],{},[25,4529,4531],{"id":4530},"phase-4-scale-datasets-and-real-workloads","Phase 4: Scale, Datasets, and Real Workloads",[12,4533,4534],{},"Phase 4 was one of the most active and productive periods in the Alpha so far. We expanded worker participation, generated real datasets, improved worker quality controls, and introduced early workflows for new task types. It represented a meaningful shift from testing to tangible output.",[12,4536,4537],{},"This was also the phase where we began creating real datasets for Mozilla Common Voice. One of these datasets, containing more than 15,000 generated English sentences, has already been submitted successfully. This collaboration continues to expand, and we have already prepared additional task types for future phases.",[12,4539,4540],{},"We also began early internal testing of our music lyric transcription workflow. This workflow is not yet part of an official collaboration, but we are exploring it as a potential use case. The initial results have been promising and we aim to have a full task flow, including validation tasks, prepared before the start of Phase 5.",[22,4542],{},[25,4544,4546],{"id":4545},"improvements-to-worker-quality-accuracy-and-platform-protection","Improvements To Worker Quality, Accuracy, and Platform Protection",[12,4548,4549],{},"As the number of tasks increased, so did our focus on accuracy and data quality. Phase 4 provided valuable insight into how workers behave at scale, what types of mistakes occur most often, and how to mitigate them.",[12,4551,4552],{},"Current protections and upcoming improvements include:",[33,4554,4555,4558,4561,4564,4567,4570],{},[36,4556,4557],{},"Automated quality checks inside task templates",[36,4559,4560],{},"Validation rules that prevent incorrect results from being submitted",[36,4562,4563],{},"Manual worker performance reviews while the reputation system is still in development",[36,4565,4566],{},"Capability tests that will soon become longer and more thorough",[36,4568,4569],{},"Better instructions and clearer expectations for every task type",[36,4571,4572],{},"Matching workers to tasks that fit their skill levels",[12,4574,4575],{},"This multilayered system has already improved quality significantly, and Phase 5 will continue strengthening these protections.",[22,4577],{},[25,4579,4581],{"id":4580},"what-we-learned-from-phases-3-and-4","What We Learned From Phases 3 and 4",[12,4583,4584],{},"Across both phases, several themes emerged.",[33,4586,4587,4590,4593,4596,4599],{},[36,4588,4589],{},"The platform can now handle real production scale workloads",[36,4591,4592],{},"Worker quality can be improved through better automation, clearer templates, and more precise capability tests",[36,4594,4595],{},"The community is engaged, consistent, and able to generate high quality datasets",[36,4597,4598],{},"The infrastructure is strong but still has room to grow, especially around speed and scalability",[36,4600,4601],{},"Dataset creation and validation for external organizations like Mozilla Common Voice is a major focus",[12,4603,4604],{},"With these lessons, we are better prepared to refine the user experience, expand our task types, and introduce more advanced features in Phase 5.",[22,4606],{},[25,4608,4610],{"id":4609},"looking-ahead-to-phase-5","Looking Ahead To Phase 5",[12,4612,4613],{},"Phase 5 will build directly on what we accomplished in the previous phases. This next stage will introduce important upgrades across stability, functionality, and ecosystem growth.",[12,4615,4616],{},"Here is what you can expect:",[33,4618,4619,4622,4625,4628,4631,4634,4637,4640],{},[36,4620,4621],{},"Expanded Mozilla task types, including Spontaneous Speech",[36,4623,4624],{},"Updated task instructions to support higher accuracy",[36,4626,4627],{},"Longer capability tests to better evaluate worker skills",[36,4629,4630],{},"Full rollout of the lyric transcription validation workflow",[36,4632,4633],{},"A more polished and streamlined worker UX",[36,4635,4636],{},"Continued improvements to manager stability and connection reliability",[36,4638,4639],{},"Early groundwork for applications, staking, and AI Workers",[36,4641,4642],{},"A stronger base for future task providers and developer integrations",[12,4644,3781],{},[22,4646],{},[25,4648,3787],{"id":3786},[12,4650,4651],{},"Phases 3 and 4 pushed Effect forward in a remarkable way. We moved from foundational systems and early experimentation to real dataset creation, validated workflows, and tens of thousands of completed tasks powered by a global community.",[12,4653,4654],{},"Every contribution from our workers, developers, and supporters has helped shape the future of the Effect platform. We are grateful for the continued participation and excitement around what we are building together.",[12,4656,4657],{},"Phase 5 is the next major step in this journey. We look forward to sharing even more progress as we move closer to a fully decentralized ecosystem for human AI collaboration.",[22,4659],{},[25,4661,187],{"id":186},[12,4663,3180],{},[12,4665,4666,193,4668,3381,4671,3813],{},[39,4667,216],{},[81,4669,3380],{"href":3809,"rel":4670},[85],[39,4672,459],{},[12,4674,4675,3818],{},[39,4676,192],{},[12,4678,4679,3194],{},[39,4680,210],{},[22,4682],{},{"title":234,"searchDepth":235,"depth":235,"links":4684},[4685,4686,4687,4688,4689,4690,4691,4692],{"id":4380,"depth":235,"text":4381},{"id":4495,"depth":235,"text":4496},{"id":4530,"depth":235,"text":4531},{"id":4545,"depth":235,"text":4546},{"id":4580,"depth":235,"text":4581},{"id":4609,"depth":235,"text":4610},{"id":3786,"depth":235,"text":3787},{"id":186,"depth":235,"text":187},"2025-11-14","A deep look at the milestones, progress, and insights gained during Phases 3 and 4 of the Effect Alpha.",{"src":3839},{"head":4697},{"meta":4698},[4699,4701,4702],{"name":252,"content":4700},"AI, agents, machine, learning, natural, language, interfaces, alpha, testing, solana, effect, data, human AI collaboration, decentralized, effect alpha, datasets",{"name":255,"content":918},{"name":258,"content":4703},"© 2025 Effect.AI","/news/raising-the-bar",{"title":4373,"description":4694},"news/raising-the-bar","U4RSvaWrU4O3P1kkcn0pXC6Sc8CTwoSTnQgxn-8VNRc",{"id":4709,"title":4710,"author":501,"body":4711,"created":4995,"description":4996,"extension":244,"headerImage":245,"image":4997,"lastUpdated":4995,"meta":4999,"navigation":260,"path":5006,"published":260,"seo":5007,"stem":5008,"__hash__":5009},"news/news/reflecting-on-the-alpha.md","Reflecting on the Alpha: Building the Future of Effect AI Together",{"type":9,"value":4712,"toc":4987},[4713,4716,4719,4721,4725,4728,4731,4745,4748,4750,4754,4760,4765,4768,4782,4849,4851,4855,4858,4866,4876,4883,4891,4894,4896,4900,4903,4923,4930,4932,4936,4939,4953,4955,4957,4959,4969,4973,4977,4979],[12,4714,4715],{},"Over the past few months, we’ve been testing and improving the Effect AI platform through a series of Alpha phases, and today, we want to share what’s happened so far! These first two phases brought together a diverse group of early contributors who helped us shape the foundation of a new kind of decentralized data platform.",[12,4717,4718],{},"From validating real-world data to exploring new task flows, the Alpha proved that it’s possible to align ethical, inclusive data work with the efficiency and affordability that modern AI projects demand",[22,4720],{},[25,4722,4724],{"id":4723},"why-were-running-the-alpha","Why We’re Running the Alpha",[12,4726,4727],{},"The Alpha is our way of testing in the real world, not just the platform itself, but the workflows, reward systems, and worker experience that will define Effect AI going forward. Each phase allows us to experiment, learn, and make meaningful improvements based on direct feedback.",[12,4729,4730],{},"Our goals are simple:",[33,4732,4733,4736,4739,4742],{},[36,4734,4735],{},"Build powerful, easy-to-use tools that simplify data validation, annotation, and a wide range of AI training workflows.",[36,4737,4738],{},"Make rewards hassle free and accessible via $EFFECT on Solana",[36,4740,4741],{},"Create real & impactful work opportunities for workers worldwide",[36,4743,4744],{},"Support open, responsible, and community-led AI development",[12,4746,4747],{},"This is just the beginning, and more Alpha phases are planned as we continue to test and scale what we’ve built.",[22,4749],{},[25,4751,4753],{"id":4752},"what-happened-in-phase-1-2","What Happened in Phase 1 & 2",[12,4755,4756,4757,4759],{},"In ",[39,4758,3453],{},", we launched a private pilot with core community members. We focused on testing core flows like task loading, task approval, and $EFFECT claims. This first round gave us the confidence to expand.",[12,4761,4756,4762,4764],{},[39,4763,3456],{},", we onboarded more testers, introduced improvements based on early feedback, and refined the backend to handle more volume and real-time payouts. We also enhanced platform performance and began collecting more complex data types.",[12,4766,4767],{},"Some of the tasks our community completed included:",[33,4769,4770,4773,4776,4779],{},[36,4771,4772],{},"Image verification",[36,4774,4775],{},"Annotation validation",[36,4777,4778],{},"Perception-based evaluation",[36,4780,4781],{},"Sentence validation and text quality review",[3441,4783,4784],{},[654,4785,4786,4796],{},[657,4787,4788],{},[660,4789,4790,4792,4794],{},[663,4791],{},[663,4793,3453],{"align":3441},[663,4795,3456],{"align":3441},[673,4797,4798,4808,4818,4829,4839],{},[660,4799,4800,4804,4806],{},[678,4801,4802],{},[39,4803,3474],{},[678,4805,3477],{"align":3441},[678,4807,3480],{"align":3441},[660,4809,4810,4814,4816],{},[678,4811,4812],{},[39,4813,4433],{},[678,4815,3499],{"align":3441},[678,4817,3502],{"align":3441},[660,4819,4820,4825,4827],{},[678,4821,4822],{},[39,4823,4824],{},"Datasets Verified",[678,4826,1281],{"align":3441},[678,4828,3522],{"align":3441},[660,4830,4831,4835,4837],{},[678,4832,4833],{},[39,4834,3537],{},[678,4836,3540],{"align":3441},[678,4838,3543],{"align":3441},[660,4840,4841,4845,4847],{},[678,4842,4843],{},[39,4844,3559],{},[678,4846,3562],{"align":3441},[678,4848,3565],{"align":3441},[22,4850],{},[25,4852,4854],{"id":4853},"voices-from-the-community","Voices from the Community",[12,4856,4857],{},"Our testers didn’t just complete tasks, they helped build the platform with their insight and ideas. Here’s some of what they shared:",[4859,4860,4861],"blockquote",{},[12,4862,4863],{},[3922,4864,4865],{},"“I remember back in the day when we used EOS for tasks, one of the biggest headaches was always the wallet and claims process. In contrast, my experience claiming Effect on the Solana network was significantly smoother, easily 10x better.”\n— Worker 1",[4859,4867,4868],{},[12,4869,4870],{},[3922,4871,4872,4873,4875],{},"“The user interface is generally smooth and intuitive, but I feel that some features could be streamlined.”",[393,4874],{},"\n— Worker 2",[4859,4877,4878],{},[12,4879,4880],{},[3922,4881,4882],{},"“I really like the new Effect worker website! Here everything works just fine and fast. Claiming payments is easy, easier then with the EOS version! The tasks are nice and I am looking forward to what is coming next!\nI also like the support we get here. And all the updates and news.”\n— Worker 3",[4859,4884,4885],{},[12,4886,4887,4890],{},[3922,4888,4889],{},"“The process worked fine, even though it sometimes was lagging/ not loading tasks. But I think that's just some glitches that can easily be solved.\nWhat I would like is some kind of log where you can see where the rewards came from; doing a task, being online, etc.","*\n— Worker 4*",[12,4892,4893],{},"Feedback like this directly shaped our improvements throughout the Alpha, and reinforced the importance of building with the community, not just for it.",[22,4895],{},[25,4897,4899],{"id":4898},"what-we-learned","What We Learned",[12,4901,4902],{},"The Alpha taught us a lot. Some of the biggest takeaways include:",[33,4904,4905,4911,4917],{},[36,4906,4907,4910],{},[39,4908,4909],{},"Simple UX matters:"," smoother, faster interactions help contributors stay focused and productive",[36,4912,4913,4916],{},[39,4914,4915],{},"Solana integration works:"," our payment system is faster and more accessible than past versions",[36,4918,4919,4922],{},[39,4920,4921],{},"Community is everything:"," real feedback made the platform better than any internal testing ever could",[12,4924,4925,4926,4929],{},"We also confirmed that there’s strong global interest in a platform like this, especially one that provides ",[39,4927,4928],{},"real value for time spent",", and contributes to meaningful AI development.",[22,4931],{},[25,4933,4935],{"id":4934},"up-next-scaling-the-alpha-shaping-the-future","Up Next: Scaling the Alpha, Shaping the Future",[12,4937,4938],{},"With two Alpha phases complete, we’re now preparing for the next stage of testing, building on everything we’ve learned so far.\nIn the next phase, we will:",[33,4940,4941,4944,4947,4950],{},[36,4942,4943],{},"Expand access to more contributors",[36,4945,4946],{},"Launch additional task types to expand the range of trainable AI models",[36,4948,4949],{},"Improve worker rewards and performance tracking",[36,4951,4952],{},"Optimize task flows, loading speeds, and interface responsiveness to improve worker efficiency.",[22,4954],{},[25,4956,187],{"id":186},[12,4958,3180],{},[12,4960,4961,193,4963,3381,4967,3813],{},[39,4962,216],{},[81,4964,3380],{"href":4965,"rel":4966},"https://docs.google.com/forms/d/e/1FAIpQLSfLh2U4_wJEmqjO0cZGtag_SFhBD2i-9MMvh789kmsLLXcSVQ/viewform",[85],[39,4968,459],{},[12,4970,4971,3818],{},[39,4972,192],{},[12,4974,4975,3194],{},[39,4976,210],{},[22,4978],{},[12,4980,4981,4982,205],{},"Explore further details about our innovative approach by visiting our documentation at ",[81,4983,4986],{"href":4984,"rel":4985},"http://docs.effect.ai",[85],"docs.effect.ai",{"title":234,"searchDepth":235,"depth":235,"links":4988},[4989,4990,4991,4992,4993,4994],{"id":4723,"depth":235,"text":4724},{"id":4752,"depth":235,"text":4753},{"id":4853,"depth":235,"text":4854},{"id":4898,"depth":235,"text":4899},{"id":4934,"depth":235,"text":4935},{"id":186,"depth":235,"text":187},"2025-06-25","Over the past few months, we’ve been testing and improving the Effect AI platform through a series of Alpha phases, and today, we want to share what’s happened so far!",{"src":4998},"/img/news/effectnli.jpg",{"head":5000},{"meta":5001},[5002,5004,5005],{"name":252,"content":5003},"AI, agents, machine, learning, natural, language, interfaces, alpha, testing, solana, effect",{"name":255,"content":918},{"name":258,"content":4703},"/news/reflecting-on-the-alpha",{"title":4710,"description":4996},"news/reflecting-on-the-alpha","QjZ3vphIoT6cOGntbpOO03BNr0HM_VXKV3XhLUfNsTY",{"id":5011,"title":5012,"author":7,"body":5013,"created":5088,"description":5017,"extension":244,"headerImage":245,"image":5089,"lastUpdated":5088,"meta":5090,"navigation":260,"path":5098,"published":260,"seo":5099,"stem":5100,"__hash__":5101},"news/news/the-future-of-ai-depends-on-better-prompts-and-you-can-help-build-it.md","The Future of AI Depends on Better Prompts and You Can Help Build It",{"type":9,"value":5014,"toc":5086},[5015,5018,5021,5026,5029,5032,5035,5038,5041,5044,5048,5051,5058,5064,5067,5071,5074,5077,5080],[12,5016,5017],{},"Artificial intelligence is only as good as the prompts that guide it. With the right input, AI models become smarter, more reliable, and more efficient. That’s why at Effect AI, we’re creating a space where prompt engineers and data scientists can collaborate to push AI development forward.",[12,5019,5020],{},"Our platform connects skilled prompt engineers with AI researchers who need high-quality, fine-tuned inputs. Whether you’re a tech-savvy linguist, a machine learning expert, or simply someone eager to break into the AI industry, this is your opportunity to shape the future of artificial intelligence.",[5022,5023,5025],"h1",{"id":5024},"why-prompt-engineering-matters","Why Prompt Engineering Matters",[12,5027,5028],{},"Crafting effective prompts is an art and a science. Even small changes in wording can drastically impact AI behavior, influencing accuracy, bias, and contextual understanding. Skilled prompt engineers understand natural language processing (NLP), model architecture, and human communication to refine AI-generated outputs.",[12,5030,5031],{},"At Effect AI, our platform can support cutting-edge techniques like:",[12,5033,5034],{},"🔹 Prompt tuning – refining AI responses for better performance",[12,5036,5037],{},"🔹 Chain-of-thought prompting – helping models \"think\" step by step",[12,5039,5040],{},"🔹 Few-shot learning – training AI with minimal examples",[12,5042,5043],{},"Research demonstrates that even small adjustments to prompts can significantly affect AI model behavior. When these inputs are fine-tuned, AI can become more intelligent, ethical, and adaptable. Through techniques like prompt engineering, AI systems can achieve greater accuracy and consistency, leading to improved real-world applications.",[5022,5045,5047],{"id":5046},"a-marketplace-for-ai-innovation","A Marketplace for AI Innovation",[12,5049,5050],{},"AI is evolving fast, and prompt engineering is at the heart of it. Effect AI’s platform is a marketplace where AI builders and prompt engineers collaborate in real time.",[12,5052,5053,5054,5057],{},"🔹 ",[39,5055,5056],{},"For prompt engineers:"," This is your chance to gain hands-on experience, refine AI behavior, and build a career in a rapidly growing field.",[12,5059,5053,5060,5063],{},[39,5061,5062],{},"For data scientists:"," Gain access to expertly crafted prompts that improve model precision and efficiency.",[12,5065,5066],{},"Whether you’re an experienced prompt engineer or just starting out, our platform provides the tools and support you need to thrive.",[5022,5068,5070],{"id":5069},"be-part-of-the-future-of-ai","Be Part of the Future of AI",[12,5072,5073],{},"At Effect AI, we believe that human expertise is the key to unlocking AI’s full potential. Working in tandem on our platform, data scientists and prompt engineers can create a feedback loop where the AI's performance is continuously evaluated and improved. Data scientists can analyze the AI's output and identify areas where it may be biased or inaccurate. Prompt engineers can then refine the prompts to address these issues. This cyclical process allows the AI system to continuously learn and evolve to align with human values and societal needs.",[12,5075,5076],{},"If you’re passionate about AI, whether as a researcher, engineer, or innovator, we invite you to be part of this revolution.",[12,5078,5079],{},"Join us and start shaping the future of AI today. Learn more:",[12,5081,5082],{},[81,5083,4986],{"href":5084,"rel":5085},"http://docs.effect.ai/",[85],{"title":234,"searchDepth":235,"depth":235,"links":5087},[],"2025-03-12",{"src":4025},{"head":5091},{"meta":5092},[5093,5095,5096],{"name":252,"content":5094},"AI, agents, tasks, prompt, engineering, tuning, prompting, microtasking, remote, work, training, earning, base, pay, compensation",{"name":255,"content":256},{"name":258,"content":5097},"© 2024 Effect.AI","/news/the-future-of-ai-depends-on-better-prompts-and-you-can-help-build-it",{"title":5012,"description":5017},"news/the-future-of-ai-depends-on-better-prompts-and-you-can-help-build-it","jA4CmJ5d6HkB7xowik7IZWTPVk2g4HvhacaZLbXlTe8",{"id":5103,"title":5104,"author":7,"body":5105,"created":5411,"description":5412,"extension":244,"headerImage":245,"image":5413,"lastUpdated":5411,"meta":5414,"navigation":260,"path":5421,"published":260,"seo":5422,"stem":5423,"__hash__":5424},"news/news/the-future-of-microtasking-and-ai-human-collaboration.md","The Future of Microtasking and AI-Human Collaboration",{"type":9,"value":5106,"toc":5401},[5107,5110,5127,5133,5136,5142,5157,5163,5174,5181,5187,5190,5201,5208,5214,5224,5231,5245,5251,5258,5261,5291,5301,5307,5314,5324,5330,5341,5367,5385,5395],[12,5108,5109],{},"Effect AI is redefining how microtasks are completed, giving individuals the opportunity to contribute while ensuring fair pay, instant payments, and an open marketplace free from corporate control. Traditional platforms take high fees, limit access to tasks, and dictate who gets paid and when. Effect AI operates differently.",[12,5111,5112,5113,5116,5117,5120,5121,5126],{},"We are developing a ",[39,5114,5115],{},"decentralized, open-source network"," where workers, managers, and providers collaborate seamlessly. Our system allows human intelligence to supercharge AI agents, making them smarter, more accurate, and more effective. With ",[39,5118,5119],{},"worker nodes, manager nodes, and provider nodes",", tasks flow through the network efficiently and securely. ",[81,5122,5125],{"href":5123,"rel":5124},"https://www.coinbase.com/en-es/learn/crypto-glossary/what-are-zero-knowledge-zk-rollups",[85],"ZK-rollups"," power near-instant on-chain settlements, eliminating payment delays, disputes, and hidden fees.",[25,5128,5130],{"id":5129},"a-smarter-approach-to-microtasking",[39,5131,5132],{},"A Smarter Approach to Microtasking",[12,5134,5135],{},"Most microtasking platforms are owned by a single company that controls everything. Workers have no say in how work is assigned, how much they get paid, or when they receive compensation. We have taken an entirely different approach by decentralizing the process and ensuring transparency at every level.",[532,5137,5139],{"id":5138},"worker-nodes-earn-without-barriers",[39,5140,5141],{},"Worker Nodes: Earn Without Barriers",[12,5143,5144,5145,5148,5149,5152,5153,5156],{},"Worker nodes are the ",[39,5146,5147],{},"foundation of Effect AI",", allowing anyone to contribute by completing microtasks. The software runs directly in a browser and enables ",[39,5150,5151],{},"peer-to-peer (P2P) communication"," with manager nodes. Workers receive tasks, complete them, submit their results for validation, and claim their ",[39,5154,5155],{},"on-chain payment in EFFECT tokens",". There is no corporate approval process, no third-party interference.",[12,5158,5159,5160,205],{},"Other platforms make workers wait for payments, deduct hidden fees, or suddenly ban them without reason. Effect AI ensures that once a worker's submission is approved, payment is ",[39,5161,5162],{},"automatically processed on-chain",[12,5164,5165,5166,5169,5170,5173],{},"Workers are also rewarded for simply keeping their node online. ",[39,5167,5168],{},"Worker base pay"," is a passive income system that provides compensation for being available to complete tasks. As long as the node remains active and passes periodic health checks, it accrues EFFECT tokens. Payouts fluctuate based on ",[39,5171,5172],{},"network conditions, task volume, and worker participation",", ensuring a fair distribution of rewards.",[12,5175,5176,5177,5180],{},"Traditional microtasking platforms often reject work with no explanation, leaving workers unpaid. That does not happen in our network. Every submission undergoes a ",[39,5178,5179],{},"validation process"," before reaching the provider. Validators assess the accuracy and quality of the work, and once it is approved, payment is finalized. There are no unfair rejections.",[532,5182,5184],{"id":5183},"manager-nodes-a-fair-system-for-task-distribution",[39,5185,5186],{},"Manager Nodes: A Fair System for Task Distribution",[12,5188,5189],{},"Manager nodes keep the network organized by handling the flow of work between providers and workers. These nodes receive microtasks from providers, distribute them to available workers, track progress, request validation, and finalize payments. In traditional systems, this process is controlled by a company that can manipulate task distribution, favor certain users, or introduce inefficiencies.",[12,5191,5192,5193,5196,5197,5200],{},"Effect AI prevents these problems by decentralizing task management. ",[39,5194,5195],{},"Manager nodes are run by independent operators"," who must ",[39,5198,5199],{},"stake a large amount of EFFECT tokens"," to participate. Staking ensures that managers have a financial interest in acting fairly. If a manager engages in fraud or unfair practices, their stake can be slashed, creating built-in accountability that centralized platforms lack.",[12,5202,5203,5204,5207],{},"Workers assigned a task can choose to decline it, but repeated rejections may result in fewer assignments. If a submission does not meet quality standards, the task is reassigned to another worker. Consistently poor performance may lead to penalties such as reduced work opportunities or refunded payments. The entire system is ",[39,5205,5206],{},"transparent, traceable, and fair",", eliminating the unpredictability of centralized microtasking platforms.",[532,5209,5211],{"id":5210},"provider-nodes-faster-smarter-microtasking-without-middlemen",[39,5212,5213],{},"Provider Nodes: Faster, Smarter Microtasking Without Middlemen",[12,5215,5216,5217,5220,5221,205],{},"AI developers, researchers, and businesses need human input to fine-tune their AI agents. Current microtasking platforms charge high fees, limit how tasks are assigned, and introduce delays in execution. Effect AI offers a ",[39,5218,5219],{},"direct, efficient"," alternative that allows providers to ",[39,5222,5223],{},"submit microtasks without unnecessary barriers",[12,5225,5226,5227,5230],{},"A provider creates a batch of tasks, defines its parameters, and submits them to the network. ",[39,5228,5229],{},"Manager nodes distribute the work to workers",", who complete the tasks and submit results for validation. Providers no longer have to rely on a company to approve payments or wait weeks for tasks to be completed.",[12,5232,5233,5234,5237,5238,5241,5242,205],{},"Prepaid funds for work are ",[39,5235,5236],{},"held in an escrow account",", ensuring that payments are secure. Once a task is validated, workers and managers are ",[39,5239,5240],{},"instantly paid on-chain",". Traditional platforms frequently delay payments or allow providers to reject work at the last minute. Effect AI prevents this by ",[39,5243,5244],{},"locking in payments once a task is completed",[25,5246,5248],{"id":5247},"a-transparent-and-reliable-microtasking-network",[39,5249,5250],{},"A Transparent and Reliable Microtasking Network",[12,5252,5253,5254,5257],{},"Effect AI guarantees that every microtask follows a ",[39,5255,5256],{},"clear and fair"," process. No task can be arbitrarily rejected, no payments can be withheld, and no worker is penalized without reason.",[12,5259,5260],{},"Every task moves through four distinct phases:",[92,5262,5263,5269,5275,5281],{},[36,5264,5265,5268],{},[39,5266,5267],{},"Creation"," – Providers define the task, submit it to the network, and deposit payment into escrow.",[36,5270,5271,5274],{},[39,5272,5273],{},"Delegation"," – Manager nodes assign tasks to available worker nodes.",[36,5276,5277,5280],{},[39,5278,5279],{},"Execution"," – Workers complete the task and submit their results.",[36,5282,5283,5286,5287,5290],{},[39,5284,5285],{},"Settlement"," – Once validated, payment is ",[39,5288,5289],{},"instantly released"," on-chain.",[12,5292,5293,5294,5297,5298],{},"Centralized platforms often reject work without explanation, leaving workers unpaid. Effect AI eliminates that uncertainty. Once a task is approved, ",[39,5295,5296],{},"it cannot be reversed or disputed",". Future iterations may introduce a structured dispute system, but the core principle remains the same. ",[39,5299,5300],{},"Work gets completed, and workers get paid—no exceptions.",[25,5302,5304],{"id":5303},"no-more-unfair-policies-or-hidden-fees",[39,5305,5306],{},"No More Unfair Policies or Hidden Fees",[12,5308,5309,5310,5313],{},"Workers on traditional microtasking platforms often face hidden penalties. Many companies ",[39,5311,5312],{},"shadowban workers, reduce task visibility, or delay payments"," without providing a reason. These platforms never reveal how decisions are made, leaving workers in the dark.",[12,5315,5316,5317,5320,5321,205],{},"Effect AI ensures that ",[39,5318,5319],{},"every action is recorded on-chain",", removing secrecy and unfair practices. Workers have the right to decline assignments, but repeated rejections may impact future opportunities. Consistently poor performance results in fewer assignments or reduced payouts. Unlike traditional platforms, ",[39,5322,5323],{},"every rule is transparent, and every decision is trackable",[25,5325,5327],{"id":5326},"why-effect-ai-is-the-smartest-choice-for-microtasking",[39,5328,5329],{},"Why Effect AI is the Smartest Choice for Microtasking",[12,5331,5332,5333,5336,5337,5340],{},"Effect AI is ",[39,5334,5335],{},"designed for fairness, transparency, and efficiency",". Workers and providers ",[39,5338,5339],{},"control the process",", ensuring that microtasks are completed on time and payments are processed without corporate interference.",[33,5342,5343,5349,5355,5361],{},[36,5344,5345,5348],{},[39,5346,5347],{},"No hidden fees"," – workers and providers keep what they earn.",[36,5350,5351,5354],{},[39,5352,5353],{},"Instant payments"," – payouts are processed as soon as work is approved.",[36,5356,5357,5360],{},[39,5358,5359],{},"Fair access to tasks"," – decentralized management prevents biased assignments.",[36,5362,5363,5366],{},[39,5364,5365],{},"Full transparency"," – every transaction is recorded on-chain.",[12,5368,5369,5370,5373,5374,5377,5378,5381,5382,205],{},"Effect AI is not just another microtasking platform. It is ",[39,5371,5372],{},"a new standard for human-AI collaboration",". Workers get ",[39,5375,5376],{},"paid fairly and immediately",", AI developers get ",[39,5379,5380],{},"high-quality human input",", and ",[39,5383,5384],{},"the entire process is open and decentralized",[12,5386,5387,5388,5391,5392],{},"AI is only as smart as the data it learns from. Effect AI ensures that human intelligence plays an active role in ",[39,5389,5390],{},"shaping the future of AI agents",". Workers complete microtasks that help AI systems learn, adapt, and improve. The result is a smarter, more efficient AI workforce powered by people, not corporate algorithms. To read more about these core concepts, visit our documentation at ",[81,5393,4984],{"href":5084,"rel":5394},[85],[12,5396,5397,5398],{},"The future of microtasking is here. Join us in ",[39,5399,5400],{},"building a better way to work.",{"title":234,"searchDepth":235,"depth":235,"links":5402},[5403,5408,5409,5410],{"id":5129,"depth":235,"text":5132,"children":5404},[5405,5406,5407],{"id":5138,"depth":896,"text":5141},{"id":5183,"depth":896,"text":5186},{"id":5210,"depth":896,"text":5213},{"id":5247,"depth":235,"text":5250},{"id":5303,"depth":235,"text":5306},{"id":5326,"depth":235,"text":5329},"2025-03-05","Effect AI is redefining how microtasks are completed, giving individuals the opportunity to contribute while ensuring fair pay, instant payments, and an open marketplace free from corporate control.",{"src":4188},{"head":5415},{"meta":5416},[5417,5419,5420],{"name":252,"content":5418},"AI, agents, tasks, microtasking, remote, work, training, earning, base, pay, compensation",{"name":255,"content":256},{"name":258,"content":5097},"/news/the-future-of-microtasking-and-ai-human-collaboration",{"title":5104,"description":5412},"news/the-future-of-microtasking-and-ai-human-collaboration","ot6zzsiA_-F67uol35DZtkzRjfyFECoCMVMDx8OHrQw",{"id":5426,"title":5427,"author":7,"body":5428,"created":5460,"description":5461,"extension":244,"headerImage":245,"image":5462,"lastUpdated":5460,"meta":5463,"navigation":260,"path":5470,"published":260,"seo":5471,"stem":5472,"__hash__":5473},"news/news/how-ai-agents-rely-on-effect-ais-natural-language-interface-to-evolve.md","How AI Agents Rely on Effect AI's Natural Language Interface to Evolve",{"type":9,"value":5429,"toc":5458},[5430,5433,5436,5439,5442,5445,5448,5451],[12,5431,5432],{},"Artificial intelligence systems are increasingly becoming integral to modern business operations. AI agents process vast amounts of data at speeds unmatched by human capability, yet their true potential often remains limited by challenges such as overfitting and concept drift. Effect AI’s platform offers a robust Natural Language Interface (NLI) that empowers these agents to learn continuously through real-time human input, ensuring they remain adaptable and effective.",[12,5434,5435],{},"Sophisticated machine learning models trained on historical datasets typically form the foundation of AI agents. Overfitting happens when models learn noise and false patterns in the data, which makes them excellent at training sets but inadequate at applying what they've learned to new data. Concept drift compounds this issue as real-world conditions evolve. When the underlying data distribution changes, the models’ predictions become less reliable over time. In environments where rapid change is the norm, these challenges can undermine the operational efficiency of AI agents.",[12,5437,5438],{},"Effect AI addresses overfitting by integrating human feedback directly into the learning loop. Users interact with the system through natural language, providing explicit feedback that guides the agent in adjusting its responses. A dynamic feedback loop records user corrections, which are then used as new training signals to balance out the fact that initial datasets are static. Continuous data inflow prevents the model from clinging to outdated patterns and helps balance the trade-off between bias and variance. This human-curated data frequently recalibrates algorithms, reducing the risk of overfitting and ensuring the model adapts to subtle shifts in context.",[12,5440,5441],{},"Real-time monitoring and adaptive learning techniques combat concept drift. The platform analyzes incoming data streams and compares them against historical trends. When discrepancies emerge, the system flags these changes for human review. Experts can then determine whether a shift represents a temporary anomaly or a fundamental change in the data environment. By incorporating these insights, the model updates incrementally, maintaining its high predictive accuracy even as market conditions or user behaviors change. This human-in-the-loop strategy proves critical in environments where static models quickly become obsolete.",[12,5443,5444],{},"Effect AI’s platform leverages state-of-the-art natural language processing techniques to facilitate this interactive learning process. Transformer-based models and contextual embeddings power the interface, enabling AI agents to parse complex user inputs and generate nuanced responses. The platform’s microservices architecture allows components to be updated independently, making it easier to integrate new algorithms as they become available. Containerization ensures that the system scales seamlessly, handling high volumes of simultaneous interactions without compromising performance.",[12,5446,5447],{},"Underpinning the entire framework is a robust blockchain infrastructure on the Solana network. Solana’s rapid transaction speeds and low-cost operations provide a secure and transparent environment for recording every interaction. Solana immutably logs each piece of user feedback, enabling verifiable audits and fostering a trust-based ecosystem. The blockchain not only upholds data integrity but also enables decentralized governance, guaranteeing transparent and collaborative system updates and modifications.",[12,5449,5450],{},"Effect AI strikes a balance between automation and human oversight. The platform integrates AI agents into an ecosystem that continuously refines their learning through real-world feedback. Problems like overfitting and concept drift are actively dealt with by Effect AI, which keeps models that are both strong and flexible as conditions change.",[12,5452,5453,5454,5457],{},"Developers and industry specialists wanting a more in-depth understanding of these techniques will find extensive documentation at ",[81,5455,4986],{"href":5084,"rel":5456},[85],". We invite organizations active in Solana initiatives to work together and integrate their solutions with our platform. The combination of natural language processing, adaptive machine learning, and blockchain technology sets a new standard for intelligent systems that learn and grow based on user input.",{"title":234,"searchDepth":235,"depth":235,"links":5459},[],"2025-02-25","Artificial intelligence systems are increasingly becoming integral to modern business operations. AI agents process vast amounts of data at speeds unmatched by human capability, yet their true potential often remains limited by challenges such as overfitting and concept drift.",{"src":4998},{"head":5464},{"meta":5465},[5466,5468,5469],{"name":252,"content":5467},"AI, agents, machine, learning, natural, language, interfaces",{"name":255,"content":256},{"name":258,"content":5097},"/news/how-ai-agents-rely-on-effect-ais-natural-language-interface-to-evolve",{"title":5427,"description":5461},"news/how-ai-agents-rely-on-effect-ais-natural-language-interface-to-evolve","pXcahLdaptsSmmyze00BS8xbSQAd6IsPYJNVX99m3Us",{"id":5475,"title":5476,"author":7,"body":5477,"created":5528,"description":5481,"extension":244,"headerImage":245,"image":5529,"lastUpdated":5528,"meta":5531,"navigation":260,"path":5537,"published":260,"seo":5538,"stem":5539,"__hash__":5540},"news/news/empowering-ai-agents-with-human-insight.md","Empowering AI Agents with Human Insight",{"type":9,"value":5478,"toc":5525},[5479,5482,5485,5488,5491,5494,5497,5500,5502,5504,5506,5515,5519,5523],[12,5480,5481],{},"Artificial intelligence is transforming industries by speeding up decision-making and streamlining processes. At Effect AI, we are taking a bold step forward by connecting AI agents with human intelligence. Our approach fuses advanced algorithms with a nuanced understanding of human expertise, resulting in AI systems that learn, adapt, and serve people better.",[12,5483,5484],{},"A key challenge with traditional AI agents lies in their inability to handle ambiguous situations or adapt to unexpected inputs. Human feedback enriches the learning process, ensuring that every decision an AI agent makes is informed by real-world insights. Real-time human input fuels continuous improvement, helping AI systems refine their outputs and better align with human expectations.",[12,5486,5487],{},"Cutting-edge techniques such as human-in-the-loop systems, active learning, and natural language interfaces lie at the core of our strategy. Users interact with our platform by providing explicit feedback through simple interfaces—buttons, ratings, and yes or no responses. Implicit feedback gathered by tracking clicks, scrolling behavior, and time on the page further guides the system's evolution. Each interaction serves as a building block for a smarter, more capable AI.",[12,5489,5490],{},"Every day, humans with knowledge in various fields join forces with our AI agents, offering their knowledge to resolve uncertainties and enhance decision-making processes. When an AI agent encounters a situation it cannot confidently resolve, it turns to human expertise for guidance. The process of active learning not only refines the AI’s accuracy but also helps build trust among users who see tangible improvements based on their input.",[12,5492,5493],{},"The integration of natural language interfaces paves the way for smoother communication. Engaging in open dialogue with users, AI agents learn to interpret subtle nuances and contextual cues. This capability transforms how technology supports everyday tasks, making interactions more intuitive and responses more tailored to individual needs.",[12,5495,5496],{},"The marriage of human intelligence and machine efficiency will benefit industries ranging from healthcare and finance to transportation and customer service. Medical professionals can rely on AI agents to provide rapid, data-driven insights enhanced by clinical experience. Financial institutions gain from AI that learns from market trends and expert advice, while customer support teams enjoy the benefits of systems that continually evolve based on user interactions.",[12,5498,5499],{},"Effect AI envisions a future where humans and machines collaborate to overcome complex challenges. Our platform represents a practical, forward-thinking solution that leverages human insight to make AI agents more adaptable, reliable, and effective in serving society.",[22,5501],{},[25,5503,187],{"id":186},[12,5505,3180],{},[12,5507,5508,193,5510,3381,5513,875],{},[39,5509,216],{},[81,5511,3380],{"href":3809,"rel":5512},[85],[39,5514,459],{},[12,5516,5517,3388],{},[39,5518,192],{},[12,5520,5521,3194],{},[39,5522,210],{},[22,5524],{},{"title":234,"searchDepth":235,"depth":235,"links":5526},[5527],{"id":186,"depth":235,"text":187},"2025-02-18",{"src":5530},"/img/news/announcement-alt.png",{"head":5532},{"meta":5533},[5534,5535,5536],{"name":252,"content":5467},{"name":255,"content":256},{"name":258,"content":5097},"/news/empowering-ai-agents-with-human-insight",{"title":5476,"description":5481},"news/empowering-ai-agents-with-human-insight","l3Endh1ATz3yL5TkckiQ5XaGwqw5GhL4OGH_FBWj2Qs",{"id":5542,"title":5543,"author":3846,"body":5544,"created":5581,"description":5582,"extension":244,"headerImage":245,"image":5583,"lastUpdated":5581,"meta":5585,"navigation":260,"path":5593,"published":260,"seo":5594,"stem":5595,"__hash__":5596},"news/news/supercharging-ai-agents.md","Supercharging AI Agents with Human Intelligence",{"type":9,"value":5545,"toc":5579},[5546,5549,5552,5555,5558,5561,5564,5567,5570,5573,5576],[12,5547,5548],{},"By Rochelle T., Effect.AI",[12,5550,5551],{},"At Effect.AI, we are redefining the future of artificial intelligence by harnessing the power of human insight. Our mission is simple: merge the ingenuity of people with the precision of machines to transform task-solving and decision-making. With our innovative platform, soon available on Solana, we are preparing to launch a system that fuses real-time human input with advanced AI algorithms. Learn more about our vision and progress at effect.ai.",[12,5553,5554],{},"Our approach is based on the simple premise that AI agents must continuously learn, adapt, and improve to remain relevant and accurate. Real-time input from users forms the backbone of our system. Every decision is informed by direct feedback from real people, ensuring that when an AI agent faces uncertainty, it actively seeks clear guidance from users or domain experts. Techniques like real-time feedback loops, active learning, and natural language interfaces propel our technology forward.",[12,5556,5557],{},"Users provide explicit feedback using simple tools like buttons, ratings, and yes or no responses. At the same time, the system captures implicit signals by monitoring user behavior—such as clicks, scrolling patterns, and time spent on pages—offering continuous insights for rapid learning and adaptation.",[12,5559,5560],{},"We place a strong emphasis on human involvement. Human operators review AI outputs and make necessary corrections, a vital safeguard in high-stakes situations. Our human-in-the-loop model seamlessly integrates domain experts and crowd workers, ensuring that every decision is refined and made safer through human judgment.",[12,5562,5563],{},"Active learning is at the heart of our design. When uncertainty arises, our AI reaches out for clarification, ensuring that only high-quality information shapes its decisions. Real people also label important data points, guaranteeing that our system learns from the best examples available.",[12,5565,5566],{},"Our technical documentation lays out the core ideas and practices behind the Effect.AI protocol. Our ecosystem connects a vibrant community of workers, task requesters, and a robust governance model. Task requesters post challenges that need solving, and real people—our workers—provide the insights that drive continuous improvement. A blockchain-based platform underpins the entire process, ensuring transparency and trust in a reliable marketplace for human and machine collaboration. You can dive deeper into these concepts in our core concepts documentation.",[12,5568,5569],{},"Natural language interfaces (NLI) are another key element of our innovation. Open dialogue between users and the AI allows the system to understand the subtleties of everyday language. By capturing the context of each conversation, our platform delivers thoughtful and precise responses.",[12,5571,5572],{},"At Effect.AI, we are committed to redefining what happens when human intelligence and artificial intelligence converge. Our platform is a practical upgrade that brings people and machines together to solve everyday challenges in healthcare, finance, software systems, transportation, and customer support.",[12,5574,5575],{},"I am immensely proud to lead this journey and excited to share more updates as we finalize our platform and prepare for our launch on Solana. Thank you to everyone joining us on this transformative adventure as we supercharge AI agents with the power of human insight.",[12,5577,5578],{},"Are you working on a Solana project and ready to explore new possibilities? Visit docs.effect.ai to learn more about our platform, and reach out if you're interested in collaborating and connecting your product to our innovative human-AI ecosystem.",{"title":234,"searchDepth":235,"depth":235,"links":5580},[],"2025-02-10","We are redefining the future of artificial intelligence by harnessing the power of human insight. Our mission is simple: merge the ingenuity of people with the precision of machines to transform task-solving and decision-making.",{"src":5584},"/img/news/announcement.jpg",{"head":5586},{"meta":5587},[5588,5590,5592],{"name":252,"content":5589},"AI, EFX, deepdive, technology, NLP, Solana, Intelligence",{"name":255,"content":5591},"Rochelle T. - Effect.AI",{"name":258,"content":4703},"/news/supercharging-ai-agents",{"title":5543,"description":5582},"news/supercharging-ai-agents","tbYXqtHZvUhNl9EsCadAcy_0jn38sc0VWjnY9cHzzcM",{"id":5598,"title":5599,"author":5600,"body":5601,"created":5760,"description":5761,"extension":244,"headerImage":245,"image":5762,"lastUpdated":5760,"meta":5764,"navigation":260,"path":5771,"published":260,"seo":5772,"stem":5773,"__hash__":5774},"news/news/new-effect-token.md","The New $EFFECT Token","Effect.AI",{"type":9,"value":5602,"toc":5752},[5603,5607,5610,5615,5629,5633,5642,5647,5650,5654,5657,5664,5669,5672,5685,5689,5692,5697,5700,5706,5711,5714,5718,5721,5725,5728,5733],[25,5604,5606],{"id":5605},"after-a-successful-snapshot-we-are-excited-to-share-that-the-new-token-will-be-publicly-tradable-on-the-10th-of-january-2025-if-youve-been-holding-tokens-during-the-snapshot-the-time-has-come-to-claim-your-new-allocation-through-the-official-portal","After a successful snapshot, we are excited to share that the new token will be publicly tradable on the 10th of January 2025. If you’ve been holding tokens during the snapshot, the time has come to claim your new allocation through the official portal.",[12,5608,5609],{},"This blog will guide you through the process, provide key security tips, and ensure a smooth transition to the new token. Although the EFFECT token will only be tradable on the 10th of January, you can already prepare your token claim. Let’s dive into the details and get you ready to claim!",[12,5611,5612],{},[3922,5613,5614],{},"In Summary",[33,5616,5617,5620,5623,5626],{},[36,5618,5619],{},"$EFFECT will be tradable on Solana on January 10th",[36,5621,5622],{},"Snapshot holders can claim their allocation through the Claim Portal starting today",[36,5624,5625],{},"Claimed tokens are staked and subject to an unstaking period",[36,5627,5628],{},"Follow the claim process described in this post!",[25,5630,5632],{"id":5631},"claiming-your-effect-tokens","Claiming your EFFECT tokens",[12,5634,5635,5636,5641],{},"The snapshot has been taken on January 1st, and all $EFX on EOS and\nBSC should not be sold/bought and they carry no utility. Starting\ntoday, you can claim your EFFECT tokens via the official Claim Portal\nat ",[81,5637,5640],{"href":5638,"target":5639},"https://migrate.effect.ai/","_blank","migrate.effect.ai",". This is the only authorized\nportal for claiming your tokens—beware of imposter sites that try to\nscam users.",[12,5643,5644],{},[3922,5645,5646],{},"Supported wallets",[12,5648,5649],{},"A wide range of BSC, EOS, and Solana wallets can be connected to the\nClaim Portal. If the wallet you used to store $EFX does not show up on\nthe portal, please reach out to support so we can assist you during\nthe claim process. As long as you have access to the wallet and keys,\nyour tokens are safe and can be claimed.",[25,5651,5653],{"id":5652},"preparing-for-the-migration","Preparing for the migration",[12,5655,5656],{},"Before starting, make sure your Solana wallet is ready. You’ll need a small amount of SOL in your wallet to cover transaction fees. Desktop wallets like Phantom or Solflare are recommended for the best experience, as mobile wallet in-app browsers may not function properly for this process.",[12,5658,5659,5660,5663],{},"If you’re using an EOS wallet, ensure you connect with your ",[44,5661,5662],{},"active"," permission, not your owner permission, unless instructed otherwise. Double-check this before proceeding to avoid any issues during the claim process.",[12,5665,5666],{},[3922,5667,5668],{},"Claiming process",[12,5670,5671],{},"The process is simple when using your EOS or BSC wallet:",[33,5673,5674,5679,5682],{},[36,5675,5676,5677],{},"Visit the official Claim Portal at ",[81,5678,5640],{"href":5638,"target":5639},[36,5680,5681],{},"Connect your EOS or BSC wallet to the portal, connect the Solana wallet where you want to receive the $EFFECT tokens, and confirm the migration.",[36,5683,5684],{},"Be sure to double-check all wallet addresses before finalizing the transaction.",[25,5686,5688],{"id":5687},"staking-unstaking-and-unstake-release-schedule","Staking, Unstaking, and Unstake Release Schedule",[12,5690,5691],{},"After claiming your EFFECT allocation, the tokens are automatically\nstaked. As long as they remain staked they will build up stake\nage, and, starting from the 10th of January, will generate rewards.",[12,5693,5694],{},[3922,5695,5696],{},"Unstake Release Schedule",[12,5698,5699],{},"If you decide to unstake some of your tokens, this part will be\nreturned to you slowly over a period of time. For every unstake, the\ntokens will be locked for 7 days, during which they are not\ntransferable and do not generate rewards. After 7 days, they will\nbegin streaming back to your wallet over a period of 30 days. To\nillustrate: it will take 37 days for all of your unstaked tokens to be\navailable in your wallet, and half of your tokens will become available\nafter 22 days.",[12,5701,5702,5705],{},[39,5703,5704],{},"Note",": unstaking is not reversable; once initiated, you will have\nto wait for the full unstake release period.",[12,5707,5708],{},[3922,5709,5710],{},"Stake Age Score",[12,5712,5713],{},"The longer your tokens are staked the higher their Stage Age Score\nwill be. This score will be beneficial for voting and rewards once the\nEffect DAO is launched. The Stake Age of the $EFX tokens you held on\nEOS is automatically transferred to the new tokens.",[25,5715,5717],{"id":5716},"rewards","Rewards",[12,5719,5720],{},"Starting January 10th, staked tokens will generate rewards. These\nrewards are distributed continuously and can be claimed at any time.\nExtra rewards will be distributed to those who have tokens staked on\nJanuary 10th. The APY generated from a stake can be viewed in the\nPortal. Please note that it is dynamic and subject to change. More\ndetails on the rewards will be released in the coming days.",[25,5722,5724],{"id":5723},"security-tips-and-support","Security tips and support",[12,5726,5727],{},"Never share your private keys or seed phrases, and always ensure you’re using the official portal link. Bookmark the site to avoid phishing scams, and verify the URL before entering any details.",[12,5729,5730],{},[3922,5731,5732],{},"Stay Updated and Get Support",[12,5734,5735,5736,5741,5742,5746,5747,205],{},"For the latest updates and announcements, follow us on Twitter at ",[81,5737,5740],{"href":5738,"rel":5739},"https://twitter.com/effectaix",[85],"@effectaix",". If you have any questions or need help, our community is here for you on ",[81,5743,230],{"href":5744,"rel":5745},"https://discord.gg/effectnetwork",[85]," and ",[81,5748,5751],{"href":5749,"rel":5750},"https://t.me/effectai",[85],"Telegram",{"title":234,"searchDepth":235,"depth":235,"links":5753},[5754,5755,5756,5757,5758,5759],{"id":5605,"depth":235,"text":5606},{"id":5631,"depth":235,"text":5632},{"id":5652,"depth":235,"text":5653},{"id":5687,"depth":235,"text":5688},{"id":5716,"depth":235,"text":5717},{"id":5723,"depth":235,"text":5724},"2025-01-05","We are excited to share that the EFFECT token will be publicly tradable on Solana on the 10th of January 2025. If you’ve been holding $EFX during the snapshot, the time has come to claim your new $EFFECT allocation through the official portal.",{"src":5763},"/img/news/token-blog-header.jpg",{"head":5765},{"meta":5766},[5767,5769,5770],{"name":252,"content":5768},"AI, EFFECT, EFX, migration, technology, staking, rewards, solana, claiming",{"name":255,"content":5600},{"name":258,"content":5097},"/news/new-effect-token",{"title":5599,"description":5761},"news/new-effect-token","AMqqcYA5zeu3QMQFapi8h1Li4gnIfsJjx9JQZfrZ5bU",{"id":5776,"title":5777,"author":7,"body":5778,"created":6126,"description":6127,"extension":244,"headerImage":245,"image":6128,"lastUpdated":6126,"meta":6129,"navigation":260,"path":6136,"published":260,"seo":6137,"stem":6138,"__hash__":6139},"news/news/migration-checklist.md","Pre-Snapshot & Migration Checklist",{"type":9,"value":5779,"toc":6112},[5780,5784,5795,5799,5802,5828,5832,5835,5867,5871,5876,5879,5897,5899,5903,5907,5910,5912,5916,5919,5921,5925,5928,5930,5934,5937,5939,5943,5946,5948,5952,5962,5964,5968,5981,5991,6001,6013,6024,6034,6045,6054,6056,6058,6062,6065,6104,6107,6109],[25,5781,5783],{"id":5782},"preparing-for-the-move-to-solana","Preparing for the Move to Solana",[12,5785,5786,5787,5790,5791,5794],{},"Hey everyone! I’m Rochelle T., and most of you know me as the community manager here at Effect AI. I've written up a checklist of the do's and dont's of the upcoming migration! The ",[39,5788,5789],{},"snapshot for token migration"," will occur on ",[39,5792,5793],{},"1 January at 12:00 PM UTC"," for all EFX and NFX tokens, so now is the time to get ready. Follow this step-by-step pre-migration checklist to make sure you are ready for the big day!",[25,5796,5798],{"id":5797},"step-1-prepare-for-the-snapshot","Step 1: Prepare for the Snapshot",[12,5800,5801],{},"Here are the most important rules to follow for the snapshot:",[33,5803,5804,5812,5820],{},[36,5805,5806,5809,5811],{},[39,5807,5808],{},"Keep Your EFX and NFX Tokens on EOS and BSC Chains:",[393,5810],{},"\nEnsure your EFX/NFX tokens are stored in a wallet you control on either EOS or BSC networks.",[36,5813,5814,5817,5819],{},[39,5815,5816],{},"Avoid Exchange Wallets:",[393,5818],{},"\nTokens held in an exchange wallet or liquidity pool during the snapshot will not be included in the migration and will be lost. Transfer your tokens to a personal wallet before the snapshot date.",[36,5821,5822,5825,5827],{},[39,5823,5824],{},"Double-Check Your Wallet Control:",[393,5826],{},"\nMake sure you have access to the private keys or recovery phrases for your wallet to retain control of your tokens.",[25,5829,5831],{"id":5830},"step-2-set-up-a-wallet-on-solana","Step 2: Set Up a Wallet on Solana",[12,5833,5834],{},"On January 5th, the new EFFECT tokens will be claimable on Solana. To prepare:",[33,5836,5837,5859],{},[36,5838,5839,5842,5844,5845,5851,5852,5858],{},[39,5840,5841],{},"Choose a Solana Wallet:",[393,5843],{},"\nWe recommend using ",[81,5846,5848],{"href":5847,"target":5639},"https://phantom.com/",[39,5849,5850],{},"Phantom"," or ",[81,5853,5855],{"href":5854,"target":5639},"https://backpack.app/",[39,5856,5857],{},"Backpack",", as they are user-friendly and widely supported in the Solana ecosystem.",[36,5860,5861,5864,5866],{},[39,5862,5863],{},"Set Up and Secure Your Wallet:",[393,5865],{},"\nDownload your wallet from its official source, set a strong password, and back up your recovery phrase securely. Make sure you have a small amount of SOL in the wallet for the claiming transaction.",[25,5868,5870],{"id":5869},"step-3-wait-for-official-announcements-on-5-january","Step 3: Wait for Official Announcements on 5 January",[4859,5872,5873],{},[12,5874,5875],{},"⚠️ Do not attempt to swap your tokens or interact with any claim portal until the official announcements are made.",[12,5877,5878],{},"After the snapshot, Effect AI will release details about the claim process. Here’s what to watch for:",[33,5880,5881,5889],{},[36,5882,5883,5886,5888],{},[39,5884,5885],{},"Claim Portal:",[393,5887],{},"\nThe official claim portal for swapping your old EFX tokens to the new EFFECT tokens on Solana will be released by Effect AI on 5 January, 2025, on our OFFICIAL WEBSITE ONLY.",[36,5890,5891,5894,5896],{},[39,5892,5893],{},"New EFFECT Token Contract Address:",[393,5895],{},"\nThe new token’s contract address will be announced via Effect AI’s official channels.",[22,5898],{},[5022,5900,5902],{"id":5901},"frequently-asked-questions","Frequently Asked Questions",[25,5904,5906],{"id":5905},"_1-what-happens-if-i-miss-the-snapshot","1. What happens if I miss the snapshot?",[12,5908,5909],{},"If your tokens are not in a personal wallet on EOS or BSC at the time of the snapshot, they will not be included in the migration. Tokens held on exchanges or in liquidity pools during the snapshot will be lost.",[22,5911],{},[25,5913,5915],{"id":5914},"_2-will-staked-tokens-count-in-the-snapshot","2. Will staked tokens count in the snapshot?",[12,5917,5918],{},"Yes, all liquid and staked EFX and NFX tokens in a wallet you control (not exchange wallets or liquidity pools) will be included in the snapshot.",[22,5920],{},[25,5922,5924],{"id":5923},"_3-when-will-the-claim-portal-be-live","3. When will the claim portal be live?",[12,5926,5927],{},"The claim portal will be released by Effect AI on its official website only on 5 January, 2025. Keep an eye on Effect AI’s official communication channels for updates.",[22,5929],{},[25,5931,5933],{"id":5932},"_4-what-if-i-have-tokens-on-multiple-chains-eos-and-bsc","4. What if I have tokens on multiple chains (EOS and BSC)?",[12,5935,5936],{},"Effect AI will account for tokens on both chains during the snapshot, so make sure all your tokens are in wallets you control on these chains",[22,5938],{},[25,5940,5942],{"id":5941},"_5-will-there-be-a-deadline-to-claim-the-new-effect-tokens","5. Will there be a deadline to claim the new EFFECT tokens?",[12,5944,5945],{},"No, the claim portal will be open and working for a very long time; there’s no deadline to claim your new EFFECT tokens on Solana.",[22,5947],{},[25,5949,5951],{"id":5950},"_6-i-have-my-tokens-in-trust-tokenpocket-wombat-anchor-wallet-can-i-use-the-migration-portal-with-them","6. I have my tokens in Trust / TokenPocket / Wombat / Anchor wallet, can I use the migration portal with them?",[12,5953,5954,5955,5961],{},"Yes, as long as you have access to your wallet, you can use them to connect to the migration portal and migrate them to Solana. We recommend ",[81,5956,5958],{"href":5957,"target":5639},"https://www.greymass.com/anchor",[39,5959,5960],{},"Anchor wallet"," for the smoothest experience.",[22,5963],{},[25,5965,5967],{"id":5966},"_7-what-is-the-contract-address-of-efx","7. What is the contract address of $EFX?",[4859,5969,5970],{},[12,5971,5972,5973,5976,5977,5980],{},"Note: We do not advise to purchase any EFX or NFX ",[39,5974,5975],{},"after"," the snapshot on ",[39,5978,5979],{},"1 January, 2025",". If you are looking to buy $EFX prior to the snapshot, it can be found here:",[12,5982,5983,193,5986],{},[39,5984,5985],{},"BSC Token Contract:",[81,5987,5990],{"href":5988,"rel":5989},"https://bscscan.com/token/0xC51Ef828319b131B595b7ec4B28210eCf4d05aD0",[85],"0xc51ef828319b131b595b7ec4b28210ecf4d05ad0",[12,5992,5993,193,5996],{},[39,5994,5995],{},"EOS Token Contract:",[81,5997,6000],{"href":5998,"rel":5999},"https://www.bloks.io/account/effecttokens",[85],"effecttokens",[12,6002,6003,6006,6007],{},[39,6004,6005],{},"Newdex"," - ",[81,6008,6010],{"href":6009,"target":5639},"https://newdex.io/trade/effecttokens-efx-eos",[546,6011,6012],{},"Buy EFX with EOS",[12,6014,6015,6006,6018],{},[39,6016,6017],{},"PancakeSwap",[81,6019,6021],{"href":6020,"target":5639},"https://pancakeswap.finance/swap?outputCurrency=0xc51ef828319b131b595b7ec4b28210ecf4d05ad0",[546,6022,6023],{},"Buy EFX with BNB",[12,6025,6026,6006,6029],{},[39,6027,6028],{},"Defibox",[81,6030,6032],{"href":6031,"target":5639},"https://defibox.io/marketDetail/191",[546,6033,6012],{},[12,6035,6036,6006,6039],{},[39,6037,6038],{},"ORION",[81,6040,6042],{"href":6041,"target":5639},"https://trade.orionprotocol.io/trade/EFX-USDT",[546,6043,6044],{},"Buy EFX with USDT (BSC)",[12,6046,6047,193,6050],{},[39,6048,6049],{},"$EFX on",[81,6051,6053],{"href":6052,"target":5639},"https://coinmarketcap.com/currencies/effect-ai/","CoinMarketCap",[22,6055],{},[22,6057],{},[25,6059,6061],{"id":6060},"beware-of-scammers","Beware of Scammers",[12,6063,6064],{},"As with any blockchain migration, scammers will try to exploit users during the process. Follow these safety tips:",[33,6066,6067,6073,6079,6085,6091],{},[36,6068,6069,6072],{},[39,6070,6071],{},"NEVER Share Private Keys or Recovery Phrases:","\nNo legitimate migration process will require you to share sensitive wallet information.",[36,6074,6075,6078],{},[39,6076,6077],{},"Only Use Official Channels:","\nAlways check for announcements on Effect AI’s website and verified social media accounts.",[36,6080,6081,6084],{},[39,6082,6083],{},"Verify Links:","\nAlways double-check URLs to ensure they are from Effect AI’s official website.",[36,6086,6087,6090],{},[39,6088,6089],{},"Avoid Third-Party Tools:","\nUse only the official claim portal and wallets recommended by Effect AI.",[36,6092,6093,6096,6097,6103],{},[39,6094,6095],{},"Do not use a scam claim portal:","\nOnly visit the ",[81,6098,6100],{"href":6099,"target":5639},"https://effect.ai",[39,6101,6102],{},"Effect AI website"," to access the claim portal.",[12,6105,6106],{},"If you encounter suspicious activity or have questions, report them to Effect AI’s support team immediately.",[22,6108],{},[12,6110,6111],{},"With the migration to Solana on the horizon, we are entering an exciting new phase with Effect AI. As always, Keep an eye out for updates, and let’s move forward together! See you on Solana!",{"title":234,"searchDepth":235,"depth":235,"links":6113},[6114,6115,6116,6117,6118,6119,6120,6121,6122,6123,6124,6125],{"id":5782,"depth":235,"text":5783},{"id":5797,"depth":235,"text":5798},{"id":5830,"depth":235,"text":5831},{"id":5869,"depth":235,"text":5870},{"id":5905,"depth":235,"text":5906},{"id":5914,"depth":235,"text":5915},{"id":5923,"depth":235,"text":5924},{"id":5932,"depth":235,"text":5933},{"id":5941,"depth":235,"text":5942},{"id":5950,"depth":235,"text":5951},{"id":5966,"depth":235,"text":5967},{"id":6060,"depth":235,"text":6061},"2024-12-18","If you’re an EFX token holder or part of the Effect AI community, now is the time to get ready. Follow this step-by-step pre-migration checklist to ensure a smooth transition.",{"src":5530},{"head":6130},{"meta":6131},[6132,6134,6135],{"name":252,"content":6133},"AI, EFX, deepdive, technology, staking, snapshot, solana, checklist, claiming",{"name":255,"content":256},{"name":258,"content":5097},"/news/migration-checklist",{"title":5777,"description":6127},"news/migration-checklist","PCF2O7LjHC8SbO8McqWvP8Wgxg59F5YxgYCpzjQag-8",{"id":6141,"title":6142,"author":3846,"body":6143,"created":6238,"description":6239,"extension":244,"headerImage":245,"image":6240,"lastUpdated":6238,"meta":6241,"navigation":260,"path":6249,"published":260,"seo":6250,"stem":6251,"__hash__":6252},"news/news/solana-announcement.md","Snapshot & Migration Announcement",{"type":9,"value":6144,"toc":6230},[6145,6155,6158,6162,6165,6168,6172,6175,6185,6189,6196,6200,6203,6209,6215,6218],[25,6146,1638,6148,6150,6151,6154],{"id":6147},"the-effect-ai-ecosystem-recently-made-a-bold-move-we-chose-solana-as-our-new-home-to-build-the-next-generation-platform-for-decentralized-intelligence",[39,6149,3846],{}," ecosystem recently made a bold move: we chose ",[39,6152,6153],{},"Solana"," as our new home, to build the next generation platform for decentralized intelligence.",[12,6156,6157],{},"As part of this transition, a new token called EFFECT will be launched, offering a straightforward 1:1 migration path for current EFX token holders. This token represents a fresh start for our mission to enhance human intelligence in the AI era. In this post, we are pleased to share the first concrete details around the migration and what this means for our token holders. Please read the information below carefully and don’t hesitate to reach out via our official channels for support.",[25,6159,6161],{"id":6160},"snapshot-of-efx","Snapshot of EFX",[12,6163,6164],{},"On the 1st of January 2025, at 12PM UTC, a snapshot will be taken of all EFX and NFX token balances across EOS and BSC. This snapshot will precisely determine the EFFECT token allocation for each user on Solana. Staked tokens will be automatically included in the snapshot, and there are no additional actions required from token holders - simply ensure you hold your EFX in a wallet that you control.",[12,6166,6167],{},"After the snapshot date, the EFX token will no longer be officially supported. We strongly advise against trading, purchasing, or selling EFX tokens after this point to avoid the risk of financial loss. Holders can sit back and relax while they await further instructions.",[25,6169,6171],{"id":6170},"claiming-your-new-tokens","Claiming your new tokens",[12,6173,6174],{},"On the 5th of January 2025, we will release an official Claim Portal through our website, enabling users to retrieve their new EFFECT tokens. The claim portal will remain accessible after this date, so holders who may have forgotten about their EFX tokens in a wallet will still be able to claim their new EFFECT tokens, even years down the line. However, for tokens left on exchanges, we are unable to assist after the snapshot is taken.",[12,6176,6177,6180,6181,6184],{},[39,6178,6179],{},"IMPORTANT",": Remain extremely vigilant against potential scammers. Only access the Claim Portal through our official website at ",[81,6182,6099],{"href":6099,"rel":6183},[85],". Do not use any other links or platforms claiming to be associated with this token migration.",[25,6186,6188],{"id":6187},"nfx-holders","NFX holders",[12,6190,6191,6192,205],{},"The Effect DAO has decided that it will discontinue support for the NFX governance token after the snapshot date. This decision will streamline the ecosystem into a more simple single token model. NFX holders will be able to claim additional EFFECT tokens through the Claim Portal for compensation. The exact conversion rate will be published ahead of the snapshot, and more details can be found ",[81,6193,6195],{"href":6194,"target":5639},"https://dao.effect.network/proposals","in recent DAO proposals",[25,6197,6199],{"id":6198},"special-cases","Special cases",[12,6201,6202],{},"While the previous information covers the migration process for most users, the following cases require special attention:",[12,6204,6205,6208],{},[39,6206,6207],{},"Liquidity Providers:","\nUsers providing EFX liquidity on platforms like PancakeSwap, Defibox, NewDex, or other liquidity pools must remove their liquidity before the snapshot date.",[12,6210,6211,6214],{},[39,6212,6213],{},"Defi Platforms & Exchanges:","\nIf you hold EFX on a DeFi Platform or a centralized exchange, you should withdraw your tokens to a wallet you control. If your tokens are on a DeFi platform or exchange at the time of the snapshot you will not be able to claim them after the migration.",[12,6216,6217],{},"Further details about the migration process, including instructions on how to claim the new Solana-based EFFECT tokens, will be provided in a blog post released alongside the launch of the Claim Portal on January 5th, 2025.",[12,6219,6220,6221,6225,6226,6229],{},"Please follow our ",[81,6222,6224],{"href":6223,"target":5639},"https://x.com/effectaix","official X"," account and join our ",[81,6227,6228],{"href":5744,"target":5639},"Discord server"," to post any questions you have about the snapshot, migration, and claiming process.",{"title":234,"searchDepth":235,"depth":235,"links":6231},[6232,6234,6235,6236,6237],{"id":6147,"depth":235,"text":6233},"The Effect AI ecosystem recently made a bold move: we chose Solana as our new home, to build the next generation platform for decentralized intelligence.",{"id":6160,"depth":235,"text":6161},{"id":6170,"depth":235,"text":6171},{"id":6187,"depth":235,"text":6188},{"id":6198,"depth":235,"text":6199},"2024-12-06","The Effect AI ecosystem recently made a bold move: we chose Solana as our new home, to build the next generation platform for decentralized intelligence. As part of this transition, a new token called EFFECT will be launched, offering a straightforward 1:1 migration path for EFX token holders. This token represents a fresh start for our mission to enhance human intelligence in the AI era. In this post, we are pleased to share the first concrete details around this migration and what this means for our token holders.",{"src":5584},{"head":6242},{"meta":6243},[6244,6246,6248],{"name":252,"content":6245},"AI, EFX, deepdive, technology, staking, snapshot, solana",{"name":255,"content":6247},"Jesse - Effect.AI",{"name":258,"content":5097},"/news/solana-announcement",{"title":6142,"description":6239},"news/solana-announcement","H2wGnHafkwh5YlHYKXn6Z8rweBCTyv3MJfPzrT-b2iU",1793374065637]