Embodied AI

A useful framework for what "autonomy" actually means when people throw the word around. Saves you from nodding along when someone conflates teleoperation with full autonomy.

A solid sector overview if you want to understand the competitive landscape and where the major players sit relative to each other.

The best single overview of where the field actually stands. Worth bookmarking and revisiting — it ages well because it deals in evidence rather than projections.

Captures the distance between what physical AI companies promise in demos and what they deliver in production. If you've ever watched a robotics demo reel and thought "that can't be the whole story" — this is the rest of the story.

Shows what a real world model company looks like when there's actual technology behind the claims. A useful benchmark for evaluating anyone else who says they're building world models.

Physical Intelligence is one of the best-funded teams in this space. This piece shows how their general-purpose models are being deployed by real partners to solve real problems — a good reference for what "world models meeting enterprise" should actually look like.

A thoughtful, wide-angle essay on why robotics is where many of the most interesting threads in AI converge. Good weekend reading — the kind that reframes how you think about the whole field.

A thorough walk through the history, theory, and potential of world models — from what they are to why they matter for physical AI. If you only read one thing to understand why everyone keeps saying "world models," make it this.

Packy does what Packy does best — takes a sprawling, technical landscape and makes it genuinely fun to read. If the other pieces here are the syllabus, this is the one you actually enjoy doing the reading for.