A few thoughts: - It feels like a microcosm of where we really are in agents right now - viral, chaotic, and culturally visible - rather than a mature product category. The project wasn't "perfect" engineering, but it rubbed the nervous system of a lot of people and made the idea of a personal agent feel inevitable. - Steinberger choosing the research frontier over bootstrapping a startup is an interesting signal. It's a tacit admission that these general-purpose assistants require deeper integration with model development and safety tooling than most indie projects can muster on their own. - The foundation move is nice in theory but also raises the usual question: who ends up shouldering liability when these agents get write access to real systems? There's an asymmetry between open source and accountability that the community hasn't really solved yet

This stuff is fun to see: https://x.com/steipete/status/2023440538901639287