Chris Rilling

robot that sorts the M8-1.25 bolt drawer at Lowe’s

@___Dario_____ what top 3 episodes that be in listening history?

I’m convinced I could run a a top decile VC fund by only using podcast listening history as the criteria for investment.

mounted & moving ✅

@QianWangX2robot epic! how do we get this in SF?

Book it like food delivery. The robot shows up on time and gets household chores done. In Shenzhen, you can now book our robot cleaning service for about $20.

@GB_Drone @jimbelosic yep, when starting from scratch just do one thing really really well first

@chrisrilling @jimbelosic THIS. Literally one of my favorite restaurants only makes made-from-scratch dumplings. They ask you if you want beef or potato dumplings. That's it. They are very good, fast, and cheap. If I wanted a burger I would go to a burger place.

“Be an In-N-Out, not a Cheesecake Factory” - @jimbelosic

@jimbelosic the glare you’ll get if you take VC funding too early ^

@chrisrilling Omg that look on my face

Excited for this team

Anyone in SF got a spare parallel gripper I can buy?

signs of life

prediction: robotics will go the same way there’s more value to be captured selling, maintaining & servicing robots than manufacturing them

@parisbayarea can confirm

handling risk, chaos and uncertainty is specific knowledge - they don't teach you that at Stanford (or maybe they do I haven't been)

Claude 🤝 finding a metal shop in SF open on the weekend (shout out to Bay Metals) Documenting the process of spinning up a basic VLA pick & place robot (FR5)

@n0homie congrats - you guys are doing some awesome work!

successful opening night

We bring the ML expertise, GPUs, and infra. We’ll help do the work. We’ll give you up to $5,000 in GPU credits. You keep all the code and models. Robotics teams: if you’re trying to get models out of the lab and onto real robots, we want to work with you. We help teams doing hard deployment work: compression, fine-tuning, distillation/quantization, and evals on real edge silicon. We’ve helped teams move to cheaper Jetsons and cut latency in half. All we ask for is blunt feedback. Send a DM, email hello@, or book via hapticlabs.ai/deploy if interested!

@robotforamerica @MachinaLabs_ @EdwardMehr @standardbots @evanbeard @GrayMatterRobot @newindustrials can you open up your DMs?

@chrisrilling @MachinaLabs_ @EdwardMehr @standardbots @evanbeard @GrayMatterRobot @newindustrials It won't - but we'll definitely do a written recap!

This Friday: AI+ Expo in D.C. Join us from 10:30-11:45 AM and hear from @MachinaLabs_ CEO @EdwardMehr @standardbots CEO @evanbeard, & @GrayMatterRobot Director of Strategy Nick Ayala. + @newindustrials moderating See you soon!

what do you think is holding people back? My guess id downsides like losing a big customer if a pilot fails, wanting a strong public impression (for later funding rounds, talent acq), shooting for an over ambitious first use case, etc. How do we normalize / de-risk shipping imperfect robots & kickstarting the flywheel?

btw I got flamed on here for saying at the start of humanoid-mania (esp by hardware people) that robotics are still quite far from widespread adoption (despite hardware progress) since (despite fancy demos) we are in a very nascent state of software, and we don't have straightforward answers to outstanding challenges (e.g. model architecture, training data constraints) > it's a couple of years later, and this is still the case (imo we're not that much further along) > imo the constant factor is that in the US we seem to have an aversion to shipping any robots (which is backwards imo since shipping crappy AI systems, collecting real-world data, and retraining with said data is the best way to make a system robust in the real-world) > until we actually decide to ship things, we can pontificate about solutions and plow capital into walled-garden lab-based solutions that look cool on X, but we won't actually produce any meaningful economic impact (other than trading secondaries of inflated vaporware stock) If you're a founder building a robot, just fucking ship it

@YorkYang5050 on making deployment N+1 faster than deployment N, do you see it as a coordination / logistics problem (on-site data collection, tooling, customer ops)? A research problem of how best to incorporate deployment data? Or something else entirely

Toward general dexterity We are training robot policies that learn a rich and coherent physical representation of the world by conditioning on multimodal observations Here’s a small glimpse of what we’ve been building: Task: Pick the ramen cup and place it in the box Given the current world state, the model generates a multimodal trajectory; here we show the decoded video and the corresponding actions executed on the humanoid