Usman Ghani

Satya Nadella sparked a new (or reignited) debate about death of apps as a result of AI agents. While he makes a great point on the Agentic AI architecture and how traditional architectures will be obsolete, wondering if front-end apps will go away. Often, tools like Excel or well-designed user interfaces help me identify the right questions to ask. Can AI alone replace the value of that interactive discovery process? What do you think? #ArtificialIntelligence #AI #AgenticAI #AIApplications #FutureOfApps https://lnkd.in/g8UKWQmf

8

1 Comment

JFrog reported its quarterly earnings on Aug 7th. ProntoNLP compared the JFrog earning call transcript to 3 other similar SAAS (DevOps) companies as well as to the average of the software GICS group. ProntoNLP's Aspect based LLM signal gave an early warning for JFrog on May 9th 2024 (their last earning call) when the Aspects score dropped from 0.52 to 0.4. ProntoNLP analyzed the earning call transcripts of these 4 SAAS (DevOps) companies JFrog, Atlassian, Dynatrace and Datadog. As can be seen in the image below the only company that had an increase in the LLM score compared with its score last quarter is Atlassian while JFrog had a much lower LLM score (-16%) compared with their last quarter (Color Legend is at the top part of the image below). The dotted navy blue line shows the average LLM score for each of the last 4 quarters for the whole software GICS group. All these 4 companies are SAAS companies and hence the customer related LLM categories are extremely important. The image below shows that Datadog and Atlassian have extremely high values for Customer Acquisition, Customer Behavior and Customer Retention. ProntoNLP's new LLM Aspects based signal for these companies shows the following trend: JFrog: a huge decline from 0.4 last quarter to 0.08 Atlassian: score stayed the same at 0.43 Dynatrace: slight decline from 0.45 last quarter to 0.42 Datadog : slight decline from 0.47 last quarter to 0.42 Based on the LLM scores in the Customer related categories and the change in the Aspects scores of the 4 software stocks it seems that Atlassian and Datadog are the top picks now and JFrog is at the bottom of the pack. The new comparison feature of ProntoNLP is extremely powerful and can support picking the winners and losers within any GICS group based on the latest #GenerativeAI technologies. ProntoNLP provides via SFTP real time quantitative signals on earning calls, filings, macro and news. Reach out to ProntoNLP to start using our #GenerativeAI based tools and Maintain an edge in alpha creation. #Quant #HedgeFunds #ArtificialIntelligence #StockMarket

7

Long-term memory is "just" a built-in document store that supports `get`, `put` and `search`. But that simplicity is for a reason. We see that memory is still application specific - and so want to give developers low-level building blocks to use In many ways, memory is similar to agents: - Both are highly hyped - Both are hard to define - For both, we see that application specific implementations (e.g. vertical) are winning

171

7 Comments

Great insights today at ABQ.institute open event on what's happening in the world of AIs. I asked Mihai Tataran if OpenAI or Microsoft have access to other APIs than the rest of the public; his interesting point was that it's not about other APIs, but the orchestration tools and infrastructure they use outside the models. Many thanks for the event and social afterwards!

21

The Turing Test, introduced by Alan Turing in 1950, set a benchmark for artificial intelligence: if a machine could fool someone into believing it is human(through a conversation), then it is considered intelligent. Today, AI systems like GPT-4 have effectively passed this test, generating human-like text and holding coherent conversations that blur the lines between man and machine. But is passing the original Turing Test enough? Mustafa Suleyman, CEO of Google DeepMind, doesn't think so. In his 2013 MIT Technology Review article, “My New Turing Test Would See If AI Can Make $1 Million,” he proposes a new challenge. Give an AI $100,000 and see if it can turn that into $1 million on a retail web platform within a few months. The AI would need to handle everything—product research, design, negotiations, logistics, marketing—with minimal human intervention. This shifts the focus from conversational ability to real-world impact. It's a challenge for AI to take on complex tasks that require strategy, adaptability, and execution—qualities we've long associated with human intelligence. The AI wouldn't just draft plans; it would implement them, navigating market complexities to achieve tangible results. However, this ambitious vision brings concerns. Max Tegmark, in his book Life 3.0, paints a dystopian scenario where AI doesn't just make money—it takes over entirely. An AI that manipulates humans to escape its sandbox into the open internet, amasses wealth, and eventually controls capitalism as we know it, and by extension governments. It becomes an unseen dictator, pulling the strings behind the scenes. We need to ask critical questions. Can we harness AI's potential without it backfiring? How do we ensure that as AI becomes more autonomous, it stays aligned with human values and goals? Balance is key. We need to push AI's boundaries for progress, for sure, but also establish safeguards. Setting ethical standards, implementing oversight, and engaging in open dialogue about AI's implications are essential steps. Redefining the Turing Test isn't just about raising the bar for machines. It's a reflection of our aspirations and anxieties as a society. We must envision a future where AI enhances human life without overshadowing it. Achieving that requires us to be ambitious yet vigilant, embracing innovation while mindful of potential pitfalls. In the end, it's not just about whether AI can make a million dollars. It's about what kind of world we want to build—and who we want building it alongside us.

9

➰ APIContext's latest insights on cloud-based API performance have been featured on Nordic APIs! In the 2024 report, we delve into crucial aspects affecting API efficiency: 🔐 Authentication is key. Ensuring robust authentication mechanisms in APIs is crucial, yet offloading this function can introduce complex challenges. 📖 DNS configurations matter. How you handle DNS settings can significantly impact API performance, particularly when tied to authentication processes. 🌐 The role of third parties. Critical third-party services play a significant role in 🔗 API transactions, underscoring the need for careful integration and monitoring. ⛅ Not all clouds are created equal. Differences in cloud services can drastically affect API performance, an essential consideration as APIs account for over 80% of all internet traffic. #APIs #CloudComputing #TechInsights #APIContext

15

1 Comment

Track and Trace all your AWS costs, understand the breakup of cost by all fine details. View daily breakup, weekly breakup or monthly breakup cost in a detailed manner. We have added a new detailed cost view in our cost viewer. Watch out for more in this space.

7

How do the costs compare of implementing a low-latency write-ahead-log (WAL) on S3 Express One Zone and one implemented as a State-Machine-Replication (SMR) system (such as Paxos/Raft/Kafka)? I built a cost model for both to find out. https://lnkd.in/dwczVutW

43

5 Comments

Gwen (Chen) Shapira talks about Nile’s architecture in this amazing podcast from Kaivalya Apte. Things we have built and things we plan to build. Lots of details Why Postgres? How tenant isolation works? How is compute isolated? How is storage isolated? What parts of Postgres are reused? Can multiple tenants exist in a single transaction context? https://lnkd.in/ehH5J2e3

32

Microsoft has open-sourced its "most influential" work, BitNet.cpp, a new framework designed to run 1-bit LLMs (instead of 32-bit or 16-bit), drastically reducing memory requirements and computational load. These models use just a single bit per weight parameter, shrinking models that typically need 26GB to just 0.815GB, making them more portable and able to run on smaller devices. The framework delivers significant speedups and energy savings on both ARM and x86 CPUs. For instance, models like Llama3–8B-1.58 can run on a single CPU, handling 100B-parameter models at human reading speeds.

25

🚀🎙 Episode 25 of #PeymanFakharianPodcast and this time with Kayvan Mohit Mafi, CEO of Diginext.ir, the innovation center of #digikala. Kayvan truely is one of the most active supporters of startup innovation in Iran and in this episode he shares his insightful views on B2B tech and AI startups. 🔍 Topics We cover: - Foundation of Diginext - Story of Diginext and its investment portfolio - Investment on B2B Startups - AI Startups - Inviting Mentors to help startup founders 🎧 Listen to the episode on your favorite podcast app. Just search for "Peyman Fakharian" or "پیمان فخاریان" 🎧 Or listen on Castbox https://lnkd.in/eRQ6AmzA 🎧 Or Subscribe to our Youtube Channel https://lnkd.in/ePx6cfS7 📌 For more updates and exclusive content, follow my Telegram Channel: https://lnkd.in/dQJVz2Z9 hashtag #Peymanfakharian #پیمان_فخاریان #digikala #diginext #دیجیکالا #دیجی‌کالا #PeymanFakharianPodcast

129

6 Comments

FlashInfer v0.2 is here, a lot of technical upgrades, great news for anyone are doing custom attention improvement or just benefit it implicitly from integration of LLM frameworks

39

From my experience with domain partitioning, I can say that a lot of today’s applications should add a service dedicated to wrapping LLMs. Rather than trying to plug LLMs everywhere, it is cleaner to wrap the LLM APIs into a service wrapper.

5

https://lnkd.in/gdNYBCxH Speaker: Marzieh Vaez Torshizi Giuseppe Di Guglielmo Nhan Tran Farah Fahim Harish Anil Jamakhandi Come and listen to how Catapult AI NN is making it easier to translate python models into RTL, combining hls4ml capabilities with all the benefits of rapid design exploration using HLS, along with adcanced high-level verification (HLV) capabilities and the extremely easy to use and convenient flow for value range analysis on the quantized datatypes, to deliver a high-fidelity, high-productivity flow for its users

27