drift

we <“> robots

The moment a robot stops driving blind and starts sensing the world is the moment robotics gets real. We added a LiDAR to our Gazebo sim from a single prompt. URDF link, joint, ray sensor plugin, frames and /scan all lined up. The part that usually eats an hour, done in context!

Presenting : Project context - Drop a DRIFT . md or SKILLS . md in your repo and Drift uses it on every step. No more re-establishing context turn after turn. Here's quick video walkthrough:

At what point does a machine stop being a machine and become a robot? We went down this rabbit hole. The honest answer: nobody fully agrees. But almost every robot follows one loop. Sense → Think → Act. Hundreds of times a minute.

A small milestone at @GoDrift_ai this week - we received a new robot arm from @makermodsai and had it executing tasks within hours of opening the box. Two things made that possible: > Makermods built hardware that's genuinely easy to assemble. It's rare in this space and worth calling out > Drift is designed so that new hardware becomes useful immediately and not after weeks of integration We'll be running this arm on real-world tasks over the coming weeks and sharing what we learn, the good, the broken and the surprising. Stay tuned!

Opening a giant open-source robotics repo for the first time is genuinely intimidating. We pointed Drift at @huggingface's LeRobot and said: "Inspect the lerobot directory and give a summary." A few seconds later the entire architecture was mapped.

Empty ROS2 workspace. One prompt. A few minutes later: a differential drive robot with LiDAR, running in Gazebo. No packages, no URDF, no launch files to start. This is Drift!

Try it - link.godrift.ai/x

Drift v1.0.24 - for robotics engineers tired of re-explaining their project to agents every command. Shipped: 🧠 Project context - drop a DRIFT.md or SKILLS.md, it's used on every step 🤖 MuJoCo + Gazebo, end-to-end - URDF, MJCF, plugins, launch, debug, run, train ⚙️ Multi-step builds that hold context - robot -> world -> sim -> tune, one prompt

Last one month at San Francisco was one of those very fast and full of learning month ever in my life. And honestly, it was one of those months that sets the tone for everything after! And the most fun part was that here in San Francisco, nobody was asking whether robots or AI agents would work anymore. That question is already so behind. The only question I hear in SF is how fast can we convert our vision into reality and that's exactly the kind of place @godrift_ai needs to be 🚀

Try it: godrift.ai

Drift v1.0.24 is live! for robotics engineers tired of re-explaining their project to agents every command. Shipped: 🧠 Project context - drop a DRIFT.md or SKILLS.md, it's used on every step 🤖 MuJoCo + Gazebo, end-to-end - URDF, MJCF, plugins, launch, debug, run, train ⚙️ Multi-step builds that hold context - robot -> world -> sim -> tune, one prompt

@godrift_ai got featured by @techinasia ! We are grateful for the support :) Read full article here - techinasia.com/building-ai-co… Onwards 🚀

I took a Waymo my first week in SF and the most impressive part was that nothing happened! It was a no drama, boring smooth ride. And that's the achievement! Boring is the hardest thing to ship in robotics. Every edge case it handled without flinching was years of someone's work. Honestly, this is what the last mile actually looks like.

People ask why robotics. Why not just do software? Because the physical world has $50T of untouched surface area, and we've barely started simulation, is where we close the gap. That's where @godrift_ai lives!

Agreed, partially. We are seeing the new to robotics folks tackling the infra/devtool layer but the what they dont understand is RDLC workflows and context settngs is 100x harder than software. Robotics engineers are notoriously faithful to their stack and truly value the development side hustle. But if these new folks focus on deployment side optimization (use case refinement, sensor logging/monitoring) this is where value unlock for them is max. At @godrift_ai we are a team who has spent 10+ years building and solving for deep verticals such as locomotion, drone chaos testing, dexterity etc. We believe sim2real to be a chicken-egg problem. Hence building @godrift_ai to empower each open sim stack engineer to iterate development, pre-training and testing 100x faster, reliably.

@ctorobotics @cybernetic_lab True and we are trying to solve that @godrift_ai

We at @godrift_ai just hosted our first dev meetup in SF. What a night. “Physical AI night” SF - open mic for robotics and physical AI builders to demo and present. Great turnout, great energy, great connections made. We're just getting started. @AntlerIndia

Physical AI night in SF was a massive success. Packed room - engineers, researchers, founders, investors, students, creators all showing up for robotics and physical AI. Everyone walked out with people to build with. This is just the beginning. @AntlerIndia @NicolasSamDuval @pratiksharda2 @godrift_ai @Sanjil