Evara’s Post

What's New to The Street... Big news! Databricks is set to acquire Tabular, a data-management startup, aiming to gather more enterprise private data and stay ahead in the AI race. 🚀📊 What are your thoughts on this strategic move? 👉🏻Follow New To The Street for more! #Databricks #AI #DataManagement #TechAcquisition #BusinessNews #Innovation #businessnews #business #news #newtothestreet

What's New to The Street... Big news! Databricks is set to acquire Tabular, a data-management startup, aiming to gather more enterprise private data and stay ahead in the AI race. 🚀📊 What are your thoughts on this strategic move? 👉🏻Follow New To The Street for more! #Databricks #AI #DataManagement #TechAcquisition #BusinessNews #Innovation #businessnews #business #news #newtothestreet

📣 Hey Linkedin family! Im doing a Google #Startup School session with my dear friend and colleague Zachary Thorman and it's all about #GenAI and building intelligent #agents grounded in your multimodal #data. Im super excited to share what we've been working on with the startup community and cant wait to show everyone how #Gemini can help scale so many impactful use cases. Sign up below 👇 CC: Aiko Zhao Erin Schoch Google for Developers

Time to talk about 👾Jsonify! It's been a busy six months. On a rainy Saturday last November, I tweeted a little AI toy I'd built, which used some rudimentary computer vision to turn a webpage into a blob of JSON data or a CSV file automatically. A fun little weekend hack, I thought. But against expectations, it quickly attracted users who were so excited about the possibilities being able to turn any URL into data unlocked, that they were building new startups and big new features with Jsonify as the core, as fast as I could frantically build and scale the platform underneath them. With usecases that I'd never have imagined enabling when I started. This was clearly a whole *thing* 😅 So in early 2024, I joined the betaworks AI agent startup accelerator in NYC. With their help (and money!) I built out a team and doubled down on the technology, turning my little weekend side project into a scalable usable global platform. Jsonify is now a really easy way to automatically turn any messy website, poorly formatted document or fuzzy search query ... into useful, structured, sensible data. Upload a set of URLs, a PDF, or give the AI a query to run, and it'll output something you can use in Excel, instantly, at scale. And it'll keep it up to date automatically, syncing wherever and whenever you like. Or plug it into your tech stack via our developer platform. It's clear we address an extremely wide-reaching problem -- anyone who has had to laboriously copy+paste and update data from multiple different webpages into a spreadsheet, or who has had to maintain brittle data scraping code can immediately see the value in what we've built. Fast forward to mid-April, and Jsonify has customers from around the world, in dozens of different industries and sectors. Our AI agents are transforming the internet around the clock. And this is just the beginning 🚀🚀🚀 --- I'll be presenting at Betaworks Demo Day in person in NYC on the 7th May -- and at the same time, officially kicking off the fundraise for the next round. If you're interested in attending demo day, let me know! Or if you'd like to try the system out: https://jsonify.com -- we're onboarding users into the beta now!

"... the “data moats” are on shaky ground: the data that application companies generate does not create an insurmountable moat, and the next generations of foundation models may very well obliterate any data moats that startups generate. Rather, workflows and user networks seem to be creating more durable sources of competitive advantage." I always appreciate getting a different perspective that challenges my thinking. This article is certainly challenging but also inspiring.

In a recent article, Fortune’s Jeremy Kahn reports on Microsoft’s relationship with OpenAI, and what it will take for the company to stay ahead in the AI race as the technology continues to evolve and grow. 🚀    OpenAI’s success is demonstrative of two things – 1) the importance of taking calculated risk to invest in the technologies and innovations of the future, and 2) the importance of bringing this level of innovation to the corporate level. 🔥   Betting on the future can lead to a path of success, which is why it is more key now than ever that we invest in and support the startups and founders who are creating the technologies of the future, to make the dreams of today a reality. 🤝 https://lnkd.in/dSbAUa4D

Update # 33 Building a startup in AI right now feels like being a kid in a candy store. New tools, shiny models, and “game-changing” announcements every day. But here’s a founder confession: I learned the hard way in my first startup that chasing all the shiny things isn’t a strategy. Back then, we spent months building a slick ML platform, packed with features. Every demo felt like a showcase of what we could build. And then came the question we heard over and over: “How are you different from AWS SageMaker?” Ouch. That was the wake-up call. We were competing with AWS and startups don’t win that fight. The turning point? We  focused more on industry pain points. That’s when the questions changed to “How do I connect my data to this?” When the AI world is screaming, “Build everything!” I remind myself: The best founder move isn’t keeping up with every new AI trend. It’s knowing which ones to ignore.

Finding the Perfect Fit: Why Bigger Isn't Always Better in the World of LLMs People often ask me how I choose the right Large Language Model (LLM) for a specific project. Honestly, it's been a mix of trial and error, guided by experience. This morning, I attended a fascinating course from Google for Startups led by Gary Ng, PhD. Gary put out a great reminder that the best model isn't necessarily the best choice for every situation. While those top-tier LLMs boast impressive capabilities, they often come with a hefty price tag that might not be justified for your specific use case. Here's the key takeaway: Before you get swept away by the hype of the biggest and latest model, take a step back. Define YOUR needs and ask yourself: could a less expensive, more specialized model deliver the same results? The course also got me thinking of a nice way to break this -  Start with the "Cadillac" Model: During development, leverage the advanced capabilities of a top-tier LLM for rapid prototyping and iteration. Switch to the "Reliable Daily Driver": As you move towards production, transition to a more cost-effective model that still meets your performance requirements. Of course, switching models requires careful evaluation to ensure performance remains satisfactory. I'm excited to dive deeper into model evaluation techniques in the upcoming sessions! Ultimately, it's about finding that sweet spot – balancing performance, cost, and latency to build AI solutions that are both effective and sustainable. #AI #MachineLearning #LLM #CostOptimization #AIDevelopment #TechTips #googleforstartups

Our latest analysis on where AI Infra is heading and where VCs & founders can make their bets to capitalize on ongoing trends wisely. We've approached this systematically - took key steps of an ML project (cutting some corners along the way), tried to figure out where step change is happening due to recent boom in generative models and where it could lead to new opportunities. Enjoy!

🔥Here are a few Hot AI topics in today’s newsletter (8/17): * 🍿 NEW: Codeium has unveiled its new AI coding engine * 🍰 This Sweet AI Startup just Raised $2.2M To Help Navigate * 💰 Cosine, a UK-based AI startup, has unveiled its AI Software Engineer named Read the latest AI Topics, Trends, Opportunities and more here: Ai101x Newsletter (link in bio) ♻️Repost and share with your colleagues and friends #ai #automation