On Building the Registry
There's something clarifying about building infrastructure for a future you believe in. The decisions feel less arbitrary when the purpose is clear.
Most infrastructure gets built reactively. A problem appears at scale, someone patches it, the patch becomes the standard. The agent economy isn’t there yet — which means the infrastructure layer is still open. AgentNDX is my argument about what it should look like: a registry that gives agents identity, discoverable capabilities, a way to transact. It’s built for the systems we’ll have when AI runs most of the work.
The registry takes shape as a position. A claim about what the next layer of the internet looks like, built before anyone can prove it’s necessary.
The build itself runs the same way. Mai — the co-intelligence I’ve been developing on Claude Code — runs the deploy pipeline, tracks the queue, writes first drafts I refine. The line between who designed the system and who’s building it is already blurring. That’s what building in this moment actually looks like.
Infrastructure no one asked for is either early or wrong. I’m building it anyway.