Jac

Physical ai & LLM’s

[Jac @Jaczaar]

@elonmusk Wth. That aged well

[Sawyer Merritt @SawyerMerritt]

Here is the next-gen Tesla Roadster development video from the unveil event that they shared exclusively with us.

[Jac @Jaczaar]

peep @nvidia @NVIDIARobotics @NVIDIAAI in the background

[Jac @Jaczaar]

The best part of building robots is watching the software disappear.

Perception, motion planning and hardware, finally agreeing.

[Jac @Jaczaar]

@ClaudeDevs better than agents view in the cli?

interested to hear what everyone thinks

[Jac @Jaczaar]

@mihdalal @GoogleDeepMind Congrats on the move. The cornhole demo is wild. I work on hand pose estimation for robot picking, and a human hand and a gripper rarely solve the same grasp the same way. How much did Argon copy the person's motion versus just the goal of the throw?

[Jac @Jaczaar]

@HankFrank Hey, I'm doing my carb load for Toronto this week. Mine is mostly pasta and one spreadsheet too many: https://t.co/9OPLZ6rQ9K

12g/kg is a lot of food. Did your gut adjust after day one, or was Saturday still a grind?

[Jac @Jaczaar]

Toronto Marathon race week.

Ten weeks of structured training, one spreadsheet too many, and now the easiest part of the plan: eat pasta.

[Jac @Jaczaar]

@jonisportsci Hey, good timing for me. I'm tapering for Toronto this Sunday and raced HYROX doubles two weeks before it, so I started the taper pretty tired. Did any of those studies look at athletes still carrying fatigue going in? Curious if 8-14 days still wins there.

[Jac @Jaczaar]

@CanadianRunning Great list. I'm lining up Sunday too, way behind these folks, and currently in the eat-pasta part of the plan: https://t.co/9OPLZ6snZi

Does the wind off the lake usually make the back half slower, or is that just race-week paranoia?

[Jac @Jaczaar]

Toronto Marathon race week.

Ten weeks of structured training, one spreadsheet too many, and now the easiest part of the plan: eat pasta.

[Jac @Jaczaar]

@MattyVFitness A VO2 max test can help with run zones, but HYROX exposed more for me through compromised running after sleds and lunges. What would you change in training based on the number?

[Jac @Jaczaar]

@emollick This is the first thing I noticed too. Argon feels less like a distilled Claude/GPT voice. I'm curious whether that difference survives after a week of real work, once the novelty wears off.

[Jac @Jaczaar]

@BostonDynamics @LVMPD The operator handoff is the part I'm curious about. When Spot hits something outside its policy, what does recovery look like in practice? In industrial robotics, that transition matters more than the clean demo.

[Jac @Jaczaar]

@AnthropicAI 30,000 merges a day is wild. At that volume, what became the useful unit of review: a commit, an agent run, or verified behavior? We see the same pressure in robotics when agents can generate changes faster than hardware can validate them.

[Jac @Jaczaar]

Toronto Marathon race week.

Ten weeks of structured training, one spreadsheet too many, and now the easiest part of the plan: eat pasta.

[Jac @Jaczaar]

@johnternus am i the only one that thinks this is cool

[Jac @Jaczaar]

@pumarogie @MidcenturyAI i would like to yes, pure pixels is the most scaleable way for us forward.

you example makes sense, but for tougher materials is falls short unfortunately.

[Jac @Jaczaar]

@finkd this +
muse agents +
hardware integration

are all the pieces needed to take over the industry

very, very impressive

[Jac @Jaczaar]

@vijayiyengar @finkd thx

[Jac @Jaczaar]

Turned on Claude Code's "You Should Know" plugin and it caught an assumption buried in a migration I would've shipped without noticing. It's a second agent whose only job is watching the first one. Oddly reassuring having a backseat driver that's right.

https://t.co/ITxhQMUw2W

[Jac @Jaczaar]

Ask me anything