Hunchful the axes what is this the protocol ↗
what is this

Your AI's memory of you shouldn't belong to your AI

A portable, revisable model of how you like to collaborate — readable by any assistant, owned by you.

Every AI is building its own private memory of you: ChatGPT's memory, Claude's preferences, Gemini's personalization. Useful — but siloed, per-vendor, and opaque, right as most people use several assistants. Your "how I like to work" ends up scattered and locked in.

A different shape: one model, any AI

A Hunchful model is a small, public, revisable snapshot of your collaboration style — not your identity. It lives at one URL. Any assistant can read it and adapt: paste a link, or connect over MCP. It's a Cognitive Model Protocol document — an open protocol, not a product lock-in.

Built to be honest

Every claim is a hunch, never a label — each carries a confidence, an uncertainty, and a falsifier (what would change its mind). You can inspect, correct, export, or delete it any time. It maps onto a small shared vocabulary of collaboration axes, which is what makes models comparable without profiling anyone.

Why it's not a dossier

Hunchful only ever describes collaboration — how you like to work with an AI. Never demographics, beliefs, health, or other sensitive attributes; those are screened out by design. Evidence stays private to you.

Make your model — about two minutes →