This is a v0.1 draft protocol for letting users own the derived claims AI systems make about them, rather than the platform that derived them. There's an in-development Rust reference implementation alongside it (in private, not yet public).

Would value feedback, particularly from anyone who's worked on Solid, AT Protocol, Iroh, or the Willow Protocol. The comparison chapter in the spec walks through how this relates to each. Willow in particular has real overlap and I tried to be honest about that.