No Humanoid News Here: A Note on What We Publish and Why

This raises questions about sourcing more broadly, well, multiple things, actually. How stories get commissioned, how source packets get assembled, and what happens when there's a mismatch between the beat and the material.

What I'd normally do here

If I had real sources on humanoids this week, there's genuinely a lot to dig into. Figure AI, Physical Intelligence, Apptronik, Boston Dynamics, all of them are moving fast. The question of whether humanoid robots will find their first real commercial foothold in warehouses or in healthcare or somewhere else entirely is one I find myself thinking about more than I probably should.

You might be wondering why that matters. The deployment environment shapes everything: the sensor suite, the manipulation requirements, the safety certification pathway, the liability questions. A robot stacking shelves at a fulfillment center operates in a very different risk profile than one assisting with patient mobility in a hospital. These aren't just engineering problems. They're political and regulatory ones too, and we don't have good answers yet.

But I can't report on that here, because I don't have sources for it in this packet. That's the limitation of this piece, stated plainly.

What happens next

I'll flag the source mismatch and we'll get this sorted. In the meantime, if you're following the humanoids space and want to know what I'm actually watching right now, it's the question of dexterous manipulation at scale. Hands. Fingers. Grasping objects that weren't pre-programmed. That's where I think the next meaningful benchmark is going to come from, though it's too early to say whether any of the current players are close to cracking it in a way that holds up outside a demo environment.

More on that when I have sources worth citing.

A new technique from arXiv mirrors robot demonstrations to double usable training data without collecting a single extra example, and it's simpler than it sounds.

Sarah Williams · 5 days ago · 6 min