I just released an open-source, model-agnostic multi-agent deep research system!
It supports model plug-and-play via @LiteLLM , has built-in web search/extract with @tavilyai , summarization-based memory, and audit trails for every session.
Check it out: https://t.co/Cq7aEkVGDR
Just watched @OpenAI’s full fine-tuning workshop at @aiDotEngineer, super insightful breakdown of SFT, DPO, and RFT! https://t.co/ILK7RBu7nN
If you're building with AI, this is a must-watch.
“Being a multi-planet species greatly increases the probable lifespan of civilization, or consciousness, or intelligence (both biological and digital.”
@elonmusk’s @ycombinator AI Startup School talk with @garrytan just dropped!
https://t.co/y3y4SsvU7G
Just got back from YC’s AI Startup School in SF. It was a true eye-opener. So many mind-blowing moments and inspiring people. I’m sharing my AI notes from some of the talks I got to attend and transcribe. Shoutout to @superwhisper:
Sam Altman (@sama): https://t.co/r0JTN1XOSG
Andrew Ng (@AndrewYNg): https://t.co/jSNtEt8pqv
François Chollet (@fchollet): https://t.co/k95YuN94RS
Huge thanks to @ycombinator for this amazing event!
Wrapping up YC AI SUS with a fireside chat: @sriramk (Senior Policy Advisor for AI, The White House) joins @garrytan. Curious to learn about the policy side of AI!
Awesome follow-up: @_mohansolo co-founder & CEO of @windsurf_ai talking “Winning the AI Tooling Race.” As a dev tools nerd, I’m super hyped for this presentation!
@karpathy@garrytan Thanks, Andrej! Your talk was incredibly inspiring, I’m already excited to go home and start building with the insights I gained from your presentation
I didn’t have much experience with robotics, but Chelsea Finn’s talk was a real eye-opener.
Robots learning to fold laundry from demos, with a foundational model trained across homes and tasks, is mind-blowing.
From @AndrewYNg today at @ycombinator AI SUS:
“One of the most important skills in the future is the ability to tell a computer exactly what you want.”
A great insight from Andrew Ng today at his presentation:
“Engineering is getting so fast that product management and deciding what to build is becoming the bottleneck.”
He even mentioned that a member of his team proposed a ratio of 1 PM to 0.5 engineers! (Though he added he’s not yet sure if he agrees with it.)
Everyone is now a programmer. Natural language is the new interface.
To unlock this, we must re-architect software so LLMs can understand and act , with protocols, structured docs, and GUI feedback loops.
We're just getting started.
Just watched Andrej Karpathy’s(@karpathy) talk on "Software in the Era of AI" at @ycombinator AI Startup School and it’s packed with mind-expanding insights.
Here are my 5 biggest takeaways you missed🧵