1/ Introducing Isaac 0.1 — our first perceptive-language model. 2B params, open weights. Matches or beats models significantly larger on core perception. We are pushing the efficient frontier for physical AI. https://t.co/dJ1Wjh2ARK

Introducing Seda: the AI agent for data and analytics. Work with data in natural language questions in plain English—Seda can write SQL, runs it on your warehouse, and gives you instant, customizable charts. Try Seda free today at https://t.co/KbvGCfSRMm and see what happens

To kick off the new year, I invited @jrdntgn to Infinite ML. He's the cofounder and CEO of MotherDuck. They've raised $100M in funding from investors like @a16z, @felicis, @MadronaVentures, and Altimeter. We talk about how they build and ship product at MotherDuck. And how they

Two novel storage solutions added to the @vercel marketplace: ▪ @niledatabase: multi-tenant serverless postgres with unlimited free databases. Think each customer / workspace getting its own sharded database. ▪ @motherduck: serverless duckdb, an incredibly efficient way of

💥 MotherDuck's native integration is now LIVE on @vercel Marketplace 💥 Developers can now deploy MotherDuck for their Vercel projects when building embedded analytics or data applications and components for the web. Learn More: https://t.co/Y7f0Kt7Fn8

I'm super excited about this launch... it solves one of the scalability gaps we had in MotherDuck, which was "what happens when I put up a dashboard that gets hammered?"

Who says the only people who can build foundation models are megacorps?

To our community: marimo wouldn't be where it is today without your help. Your code contributions and feedback shape marimo for the better; the awesome notebooks you make, and the blue-sky ideas you share, inspire us to keep building. Thank you!

Here is a link to the benchmark with all of the hosted platforms included:

I don't like to make too many comparisons but we're faster than things that cost a couple of orders of magnitude more. (at this one benchmark, your mileage may vary)

A few months ago, I wrote that the important thing about a database is how fast it is improving, not how fast it is today. Since then, MotherDuck has gotten 7x faster, according to ClickBench. We still have a lot of work to do, but not bad so far.

TLDR

Is Big Data dead? @jrdntgn and the @motherduck and @duckdb teams are leading the “small data” movement for faster and simpler data analytics *YouTube: https://t.co/PotdFVAGRz *Spotify: https://t.co/BYxUuiR2WS * Apple Podcasts: https://t.co/BawC8XGMoc

Shout out to @ywelsch quietly slipped this into today's release.

Most MotherDuck queries got almost 2x faster today. One of our engineers figured out how to do dual execution in one round trip instead of 2, so I'm seeing sub 50ms queries measured from the client in NYC. (even a 🦆 can't exceed the speed of light)

This week I attended @smalldatasf . For folks who know me, you know that I’m not a fan of most conferences. I find that they’re often the major league version of “this could just have been an email”. But I decided to still go to check it out for two reasons: 1️⃣ It was a first

Just got back from @smalldatasf. It’s fascinating how we're seeing this shift from "big" to "small" — not in terms of scale but in terms of focus and efficiency. Here are a few takeaways: - Less is More (And Always Has Been): There’s a lesson we keep relearning: simplicity

Develop locally, ship joyfully. This is part of the @smalldatasf manifesto and something we 💯 fully embrace at Rill with our BI-as-code approach. Eric Okuma, Product Manager, is attending & can't wait to connect with developers to learn new ways to build meaningful analytics.

Episode #8: @jrdntgn on The Analytical Language of John Wilkins by Jorge Luis Borges https://t.co/Oqe8MBlMwG

🎉 We’re officially sold out for Small Data SF next week, though added a few more tickets in case some stragglers want to join as we celebrate and collaborate on the small data + ai movement. https://t.co/GWcTsyx7ZB