Intelligent Machine Guild @Intmachineguild

@UnitreeRobotics Full open-source, not just the weights, is the part that matters most here. The whole field moves faster when SOTA results are actually reproducible instead of stuck behind an API.

@BostonDynamics Bi-directional is the key word here. One-way data collection is easy, the hard part is trusting a robot to actually act on what the factory tells it back. That's a real jump in integration, not just a smarter sensor. Bravo.

@GoogleDeepMind Robots need this same skill: think and act in the background without breaking a person's flow. Good sign seeing it nailed in conversation first, that usually shows up in physical systems next.

@rosorg Love seeing ROS 2 show up here, a robot climbing wind turbine blades is a way better proof point than another warehouse demo. That's where a standard actually earns its keep.

@agilityrobotics @Bloomberg Cooperative safety' is doing the real work in that quote. Traditional industrial safety kept humans out, this is about a robot predicting human movement well enough to work beside someone untrained. Industry is the right place to pressure-test that before a living room has to.

@tristanharris The headline-level debate (doom vs hype) gets all the attention, but the actual work that would make AI systems safer is unglamorous: testing protocols, deployment standards, red-teaming practices that never make a documentary. Would love to see more of that story told too.

@bcherny Super impressive. Can't wait to see what's next.

@Apptronik Congrats to Jeff, well earned. Watching Austin build a real gravitational pull for humanoid robotics is the more interesting story long-term, these hubs don't happen by accident, they need people willing to plant a flag early and make the case for the region.

@CovariantAI Very impressive. Can't wait to see what's next.

@Waymo The real design choice here isn't the autonomy, it's deciding what to hand back to the rider. Music and temperature are trivial to automate too, but letting people control the small stuff while trusting you with the driving says a lot about how you earn trust in autonomy.

@Figure_robot The compute commitment here is wild, but the harder unlock for a robot in every home isn't FLOPs, it's trust. Whatever safety and reliability standard lets people feel good handing over their kitchen to a robot doesn't exist yet. That part scales a lot slower than GPUs.

@BostonDynamics Glad to see 'data silos' called out directly, that's the real unlock for smart factories, not just a smarter robot. Interoperability across systems is the hard, unglamorous part of this whole shift. Curious how Orbit handles legacy systems that were never built to share data.

@DrJimFan The 'boil the ocean, then specialize' arc is basically the old guild apprenticeship model with extra GPUs. No master started on one hand-drawn task either, you generalize for years before your hands specialize into the craft. Robotics is just hitting that arc.