Gorkem Yurtseven

Today, Daytona launched DevContainer (https://devcontainer.ai/): an open-source tool that automatically generates an infrastructure-as-code specification just by pointing it to a repository. Check it out! Super useful 🚀

23

1 Comment

Every time we sit down with enterprise CIOs and AI teams, the top of mind concern we hear is "Reliability" - how do we go from our use case to reliable AI as fast as we can? In our experience, reliable AI is extremely hard to engineer. There is no silver bullet - it is about the many long tail problems that you need to grind at for weeks and months. But, this is what makes enterprise AI extremely resource intensive today. We are building Awen to change this. We are building the core agentic infrastructure that helps AI teams ship reliable AI 100x faster - from data curation to running evals at scale. Experience more in our playground: https://lnkd.in/gFGnpcdR Abhishek Borar Himanshu Muchhal

9

The latest edition of The Driverless Digest has just been published! Highlights from this week: 🤖 Waymo now at $45 Billion valuation but how do they stack up to Uber at a public market cap of $150 billion? 💵🔥Zoox to launch in San Francisco and Las Vegas. 🤖 Waymo working on scheduled rides feature which is great for customers but read why I think it’s only going to hurt their utilization. Check it out here - https://lnkd.in/gNywsSmW

18

Anthropic just released its new multimodal AI model: Claude 3.5 Sonnet. The benchmarks sure need a refresh as AI models keep improving at breakneck speed. But more importantly, each benchmark represents a real-world application, so it's crucial to steer clear of thin wrappers over any benchmarked capability. #claude #anthropic #chatgpt #gpt #openai

5

1 Comment

Acquisitions reveal a lot about ambitions. NVIDIA is actively acquiring AI startups. 2023 saw a surge in Nvidia’s dealmaking activities. The trend continues in 2024. This year, they bought Run: ai and Deci AI. These companies specialize in optimizing AI workloads. Nvidia’s focus on AI performance and optimization is no news. This acquisition spree enhances its AI capabilities. Major corporations acquire startups to: - Access latest technology - Innovate faster - Gain new talent - Strengthen market position PayPal started acquiring BNPL startups before offering the Buy-Now-Pay-Later option on its main platform. Acquisitions are done to drive growth faster. You don’t buy companies; you buy growth. 𝙄𝙣𝙩𝙚𝙧𝙚𝙨𝙩𝙚𝙙 𝙞𝙣 𝙢𝙤𝙧𝙚 𝙈&𝘼-𝙧𝙚𝙡𝙖𝙩𝙚𝙙 𝙘𝙤𝙣𝙩𝙚𝙣𝙩? 𝙈𝙖𝙠𝙚 𝙨𝙪𝙧𝙚 𝙩𝙤 𝙛𝙤𝙡𝙡𝙤𝙬 𝙢𝙚. #entrepreneurship #venturecapital #startup #mergers #acquisitions Serial Entrepreneur & Investor Helping Startups Become Unstoppable – David Hauser

77

4 Comments

You can now leverage model routing in LiteLLM (YC W23) with Not Diamond! Simply define the LLMs you want to route between and your request will go to the most accurate model on each prompt. Accuracy is maximized by default, but you can also define explicit cost/latency tradeoffs to save 2-10x on cost by using smaller models when doing so doesn't impact quality. s/o Krrish D. and Ishaan Jaffer for building such an amazing gateway. Link to code block in thread.

36

2 Comments

$2 H100s: How the GPU Bubble Burst Last year, H100s were $8/hr if you could get them. Today, there's 7 different resale markets selling them under $2. What happened? The market has flipped from shortage ($8/hr) to oversupplied ($2/hr), because of reserved compute resales, open model finetuning, and decline in new foundation model co’s. Rent instead. https://lnkd.in/gpQEidmC

2

1 Comment

Waymo grew from 100K rides per week to 150K rides in two months. At 7.8M rides per year this is nothing compared to the 10B rides Uber provides. But here is the catch, Uber's costs are more or less fixed. Google's are going to come down exponentially. The cost for an AI team will go down very fast if expressed per ride. Given no drivers are needed and vehicle costs are coming down as well, it is just a matter of scaling up for Google to steal away clients from first Uber but afterwards also public transport, second car ownership and finally first car ownership. In 2025, Waymo can easily go to 1 or multiple millions of rides per month. 2026, 10-50M and in 2027, 1B per month. Giving that Tesla will want to launch in 2025 and scale in 2026, consumers will be the big winners. https://lnkd.in/e4rcAH6D

13

6 Comments

Benjamin Bolte from K-Scale Labs just posted his 'Robotics Predictions for 2025' containing some hot takes 🦾🔥 By the end of 2025.... 1. Humanoid hardware is commoditized - Hardware components will converge, but robot designs will remain different - Actuators no longer differentiators; industry moves towards open-source like MIT Cheetah - BOM for a humanoid will come down to a shocking <$8k! Question: What happens to incumbents like Telsa, Figure, 1X, Apptronik, etc., who have spent billions of $ in R&D to create robots with a $50 - $150k BOM? 2. No ChatGPT moment for robotics - A LOT of data (millions of hours) is needed to train a general-purpose robotics model - Language + multi-modal models will not be able to control a robot sufficiently - Progress will be incremental rather than spiky Question: If building a robot for a domain and bootstrapping to generality is a losing strategy, what's a better one? 3. More than half of humanoids will be sold to consumers - The 'killer apps' will be speech and vision - Traditional SLAs for business customers will force robotics teams away from building good generalist models - "Robotics companies will start looking more like SV tech companies, caring about user engagement and retention." Question: What are the 'killer' consumer use cases that will drive this adoption? What are the jobs to be done? Link to the full blog in the comments. What are your thoughts?

26

2 Comments

Excellent blog from the folks at Shield Capital about the infrastructure needed to accelerate autonomy! Miru (YC S24) is mentioned among a host of other awesome startups that build software tools for robotics startups. "Consider the world of web app development: developers have an extensive array of tools at their disposal, offering support for everything from coding assistance and test management to cloud infrastructure handling and beyond. This is not yet the case with autonomous systems, whose developer ecosystem remains more nascent and fragmented. Many autonomous systems developers describe how they rebuild common tooling like testing and simulation tooling across every autonomous systems development organization where they worked." This is why we built Miru. These teams deserve better off-the-shelf tooling. They shouldn't have to develop and maintain this themselves. If you're a developer struggling with CI/CD for your robot or edge device, reach out to me. Let's get this off your plate 🦾 https://lnkd.in/gk2fSHTq

24

4 Comments

Groq Partners with Aramco to Propel Saudi AI Ambitions In a groundbreaking move, Groq, the innovative chip startup renowned for its cutting-edge technology, has inked a significant deal with aramco, the world’s leading oil company. This strategic partnership is set to be a game-changer for Saudi Arabia's burgeoning AI landscape. What This Means: Groq’s advanced chip technology will play a crucial role in supporting Saudi Arabia's ambitious plans to position itself as a global leader in artificial intelligence. The collaboration is expected to significantly enhance the computational power available for AI applications, paving the way for breakthroughs in various sectors including energy, healthcare, and smart cities. Key Highlights: Boosting AI Capabilities: Groq’s state-of-the-art chips are designed to handle complex AI tasks with unprecedented speed and efficiency, aligning perfectly with Saudi Arabia's AI objectives. Strategic Alignment: This partnership reflects a broader trend of technological alliances aimed at fostering innovation and economic diversification in the region. Future Impact: By leveraging Groq’s technology, aramco and Saudi Arabia are positioning themselves at the forefront of the AI revolution, potentially setting new standards in the industry. As the tech world watches closely, this deal underscores the growing intersection of AI and semiconductor technology, and how strategic collaborations are shaping the future of these industries. For more information: Bloomberg https://lnkd.in/dr89qeCV #AI #ChipTechnology #Innovation #TechPartnerships #Groq #Aramco #SaudiArabia #FutureOfAI #TechNews

6

Leamar Hiring has identified standout AI startups across leading industries, poised to make a significant impact in 2025: 1: DevRev 2: GMI 3: Guidewheel 4: Wynd Labs 5: Perplexity 6: World Labs 7: Beautiful.ai 8: Kalepa These teams are highly regarded for their innovative vision, exceptional growth potential, and the strong interest they have garnered from the tech community over the past year. Are your tech teams often overwhelmed, with engineers earning $150k/year handling tasks worth only a fraction of their cost? 👉 Our team of engineers will take over these workflows, freeing up your team to focus on higher-value work. Want to learn more about this flexible "secret weapon" for tech companies? 🤩 🌐 Contact now: https://lnkd.in/gSn7_jfw 🤖🚀 #AI #Innovation #Startups2025 #LeamarHiring #FutureTech

7

Despite its 671B parameters, DeepSeek-V3 achieves energy efficiency via its Mixture-of-Experts (MoE) architecture, activating only 37B parameters per token. Training required 2.788M H800 GPU hours, costing $5.57M, setting a new standard for cost-effective, scalable AI. Benchmarks on MMLU, MATH, and Chinese SimpleQA show DeepSeek-V3 outperforming GPT-4 and Claude 3.5 in coding and math tasks, with a 90.2% accuracy on MATH-500. DeepSeek-V3 leverages FP8 mixed precision with fine-grained quantization, reducing memory usage by 50% while maintaining accuracy through increased accumulation precision. Multi-token prediction accelerates inference by generating multiple tokens sequentially, supported by speculative decoding that handles error propagation through parallel verification. https://lnkd.in/dWR32WUp

1

Notion vs Glean reminds me a lot of Tesla vs Waymo... Just like I think Tesla is going to solve the 'software' problem of autonomous vehicles before Waywmo solves the 'hardware' problem... I think that Notion is going to solve 'enterprise knowledge search' before Glean solves the problem of document and knowledge creation... Why? A couple reasons: 1) Just like Tesla, Notion has a huge advantage when it comes to the amount of underlying data and documents to train on... They already have 100M+ users, and likely 10x that in terms of the number of documents, files, etc. in team's workspaces. 2) It's very hard to change people's behaviour re: where/how they create 'work'. It's why a lot of companies have come and gone in that category... As exciting as Glean's AI writing assistants might seem, most people are just fine (or even prefer) a good old fashioned editor/writer. I could be way off here, but this will be an intersting one to follow. Thoughts? Where am I getting it wrong? #ai #enterprise #knowledge #management #notion #glean

3

1 Comment

Great overview of the promise of local AI by Javier Redondo. https://lnkd.in/gqC8S9Dm TLDR: Embracing the power of local on-device models brings a host of benefits: - 🔒 Enhanced Privacy and Security: Keep sensitive data on your device, minimizing risks and ensuring compliance with privacy regulations. - ⚡ Reduced Latency: Enjoy faster response times by eliminating the need for data to travel to remote servers—perfect for real-time applications. - 📶 Offline Functionality: Access AI capabilities without an internet connection, ensuring continuous operation wherever you are. - 💰 Cost Efficiency: Cut down on cloud service expenses by leveraging your device's computing power, making AI more accessible. - 🎯 Customization and Control: Tailor models to your specific needs for better alignment with your domain. This is what we are building at Knapsack. Harness these advantages to revolutionize how you deploy and interact with AI! 🌟 #AI #MachineLearning #Innovation #Privacy #Efficiency #knapsack

13

1 Comment

This is an important point. The ethic that we should build all possible technology “because we can” and “because it’s profitable” and “because someone else will“ has become pervasive and socially destructive. It is not unlike the wave of offshore manufacturing that hollowed out middle America and delivered up ghost towns and opioids and our current political disaster. Not sure what the answer is, but it is pretty clear the road we’re on risks going right over the cliff to techno-authoritarianism. (https://lnkd.in/ehi8QBqk) I'm exploring these issues and working toward a more constructive set of solutions on my substack: Reinventing Capitalism (https://lnkd.in/ewaaEcaA). It seems clear to me that new solutions are needed. Blindly following the techno-utopians to wave upon wave of new "solutions" that leave out humans seems to be a grave error. We need to re-think the structure of incentives in our society and encourage our best minds to develop technology to advance humanity, not replace it.

5

5 Comments

Excited to welcome Lume (YC W23) to the KV portfolio! Lume uses AI to automate data mapping, empowering customer success & engineering teams to effortlessly understand, map, and validate data.

63

2 Comments

Google DeepMind has just released two exciting updates to the Gemini family: Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002. These models bring faster, smarter, and more efficient AI capabilities, making them a game-changer for developers and businesses alike. Here’s what’s new: • >50% price reduction on Gemini 1.5 Pro • 2x faster output with 3x lower latency • Rate limit boosts: 2x higher for Flash and ~3x for Pro • Concise, focused responses, Enhanced safety and customizable filters #AI #MachineLearning #Gemini #DeepMind #AIInnovation #TechUpdate https://lnkd.in/gGGbwEFt

18