Metistry is an AI-Native Proof Registry.

Metistry is an AI-native proof registry. It records what changed, the proof that the change was real, and who stands behind it, then reasons over your live data and returns answers you can verify. You run it over the database you already have. Wondering how it stacks up against a blockchain, a vector database, or a typical database?

→ Compare Metistry

Access is by request while we onboard the first builders. There is no public self-serve key yet, on purpose, so early integrations get real support instead of a queue. Drop your email in the form below, or the bar that follows you as you scroll, and you are in line. Tell us one sentence about what you are building and your primary language and we can fast-track you. Once you are in, you get a project per connected database, an API key you can rotate or revoke anytime, and the SDK, CLI, and MCP server to build with.

No. Point Metistry at a Postgres or Supabase database and it reads your existing schema and builds a reasoning layer over it. Your database stays the source of truth. Nothing is copied to a second store, and private data never leaves your environment unless you ask it to.

No. Metistry is infrastructure you build on, the way you build on a database, not an asset to buy or trade. It validates with Proof of Change (PoΔ), which records a change together with the evidence that it was legitimate, rather than by consensus among strangers. That is why there is no fee at the record layer and no shared history to replay.

A blockchain is good at what it was built for, settling assets among strangers with no trusted operator in the middle. The work in front of you is different. Your records are private, your operations run constantly, and what you need to prove is that a change was legitimate, not just that it was published. A blockchain proves by publishing, so anything private is exposed the moment it lands, and every write carries a fee. Metistry was built for those needs directly. Private data stays in your environment, there is no fee at the record layer, and the proof rides with the record.

A blockchain can prove a record exists and that a valid key signed it. It was not built to tell you whether the data behind that record was real, or whether an AI reasoning over it reached a sound conclusion. Metistry reasons over your live data and seals the proof of legitimacy to the answer, so the result and its evidence cannot be pulled apart.

MCP is the open protocol most AI clients and coding assistants already speak. Add the Metistry server to your client and start asking questions against your data with proof attached. It is the fastest way to feel what Metistry does. If you would rather wire it in yourself, the SDK and CLI build against the same open spec, so MCP is the quick path, not the only one.

A Python SDK, a TypeScript SDK, a CLI, and the MCP server. The full API reference comes with access.

MQL is SQL-compatible, so if you know SQL you already know most of it. The difference is what a query can reach. A SQL query returns rows. An MQL query reasons over your data and the relationships in it, lets you set the context the reasoning runs within, and returns an answer with its proof attached, all in one place. You get AI-native reasoning through the same interface your team already uses.

Every result carries a Nexus Seal you can verify on your own, without trusting the service. The seal binds the record to the evidence behind it, so a change that merely looked legitimate has nothing to seal. You can run the conformance suite and check proofs independently against the open spec.

Yes. You assign your own models to the tiers you turn on, from a deterministic Tier 0 that uses no model at all up to deeper reasoning. You control what each tier is allowed to do.

The specification and tooling are open under Apache 2.0, stewarded by the Legacy Impact Futures Trust. You can read the spec, build against it, run the conformance suite, and verify proofs from the public repository at github.com/Metistry-Dev/metistry. What you reach here is the hosted engine that runs that standard as a service.

There is no fee at the record layer. Beyond that, pricing follows your use case and always includes a free tier.

Other tools are each strong at their own job, and none was built to do this one. A database stores state. A vector store finds similar things. A blockchain settles assets among strangers. Metistry records a change together with the proof that it was legitimate, and reasons over your live data with that proof attached. We laid them out side by side so you can see exactly where the gaps fall.

→ Compare Metistry