Question 1

What does Meridian Attest do?

Accepted Answer

It sits between an AI agent and the systems it acts on. Every action the agent proposes is classified against closed enumerations, risk is derived by a fixed rule, the action is routed to an accountable human role, declared policy is enforced, and a signed, hash-chained record is sealed in an append-only ledger.

Question 2

Does Meridian Attest certify legal compliance?

Accepted Answer

No. Meridian Attest attests what happened and whether declared policy was followed. It never judges whether your organization complies with the law.

Question 3

Does a language model make the decisions?

Accepted Answer

No. Decisions come from tables: a versioned doctrine with closed enumerations, flags, a derivation rule and policy rules that only harden the default verdict. A model may act as the demo agent that proposes actions; it never decides.

Question 4

How can I verify the evidence?

Accepted Answer

Every event is canonicalized, hash-chained to its predecessor and HMAC-signed, and cites the doctrine sha it ran under. Verify recomputes every hash and signature of a run and reports the first sequence number where the chain breaks, if any.