Robot-use agents can now control real robots. But they are slow.
We made the GPT-6 Astra + Codex agent run much closer to real time:
- cup task: 11.1 min → 3.4 min
- towel folding: ~18 min → ~6 min
- robot moving 84% of the time, up from 24%
Blog: https://t.co/BIaiZ2P5ch
Code: https://t.co/MvWxzfMXiK
Introducing Frostbite: a new class of CPU built around interaction nets.
Given only interaction net theory, our agents co-designed the ISA and chip architecture from scratch, running large swarms to design, verify, and optimize the CPU down to the final GDS.
On our first set of target workloads, Frostbite is 4.6x more performant per watt than modern traditional CPUs at similar area.
Congrats to everyone at @AnthropicAI on going public today!
The @pollenrobotics team and I have been working on a little robot inspired by Claude.
Figured today was a good day to introduce him 🦀
Say Hi to Claudius :)
Agents are very good at learning from examples, so we added more of them to our website
Closed loop linkages, solver coupling, hydroelastic contacts, mimic joints, and so on!
#Felis Preview (3/3)
Part1: 🐱Felis is dribbling a ball⚽️ between pallets.
Part2: Its waist is a universal joint, which lets it turn around in a tight spot and walk its way out.
Policy trained via sim2real.
#Robot#Cat#Simulation#Robotics
For the first time in human history, we may have to consider whether slowing technological progress could cost more lives than it saves.
I’m all for pushing forward. Let the age of abundance begin.
Nick Bostrom says delaying superintelligence carries risks of its own. Every year of delay could postpone medical breakthroughs that extend human life and reduce suffering.
Furthermore, AI's upside could extend far beyond longer lives and higher living standards, opening up new possibilities for human intelligence, emotional depth, creativity, and connection.
Nick joins a16z's Eddy Lazzarin and Elena Burger to discuss what people misunderstand about his views, what “deep utopia” means, and what human purpose might look like in a world with superintelligence.
00:00 Intro
00:56 What we've learned about AI
02:51 Misconceptions about Bostrom’s views
04:12 Why he wrote Deep Utopia
07:29 How humanity should navigate superintelligence
11:00 Aligning superintelligence
15:51 The cost of delaying superintelligence
23:09 Building superintelligence safely
27:10 AI research isn’t the Manhattan Project
30:27 Why our vision of the future matters now
32:49 Does life have a meaning, and can AI answer it?
YouTube: https://t.co/k5tmfaj1LF
@eddylazzarin@VirtualElena
Steel 0.2 is out:
• 81 simulation environment examples, interactive docs, and migration instructions for your coding agent.
• Magnetic contacts that latch, hold, and release.
• Closed-loop joints for grippers and parallel robots.
Build the intelligence around the job, not the job around the intelligence.
RoboForce is going model plus robot (RF-0 plus TITAN), aimed at the hardest industrial work, and solar is first.
RoboForce chose the hardest work as the starting point for robot intelligence. Here’s why.
Building intelligence for real work starts with understanding the work itself—the conditions, the challenges, and what customers need to get the job done.
That understanding shapes what we choose and what we build: RF-0, our robot foundation model, and TITAN, our general-purpose AI robot for the hardest industrial tasks.
See our commercial progress in solar and hear directly from the customer in the field.
The hardest work is where we begin. Solar is just the start.
Read the full story: https://t.co/ByjiubWc4V