A stated confidence on every answer. A person below the line.

The model is the maker and a named person is the checker. With elan1 Decide and elan1 Govern, AI triages AML alerts, KYC reviews and disputes with a calibrated confidence; answers below your threshold go to a reviewer, and any resulting write is approved and recorded before it reaches a core system. The checker is never the requester.

RBI's FREE-AI framework, published in August 2025, expects board oversight, human accountability and auditable records for AI in financial services. elan1 helps you produce evidence for those expectations: a named reviewer for unsure decisions, a Decision Ledger entry for every answer and a proof record for every write. It does not make an institution compliant, and we do not say RBI has reviewed elan1.

Every answer is kept in the Decision Ledger with its type, calibrated confidence, route and reviewer. When a decision leads to a write, the ledger entry's digest is bound into that write's proof record, so the decision and the action can be read together. From L2 Attested, the records are signed and checkable offline.

It goes to a person, with the reason it was routed. Routes are auto, review, a slower second pass, or hold. No answer changes a record on its own authority.

No. A model proposes the answer and states how sure it is. Whether any record changes is decided by elan1 Govern, with versioned rules and no model in the allow path. Above the autonomy ceiling, a named person decides.

Yes, at L2 Attested: a signed bundle can be checked offline with the free verifier by your auditor, a lender's diligence team or an insurer. We do not say any regulator has approved elan1.

Yes. collect1 is a configuration pack for RBI-related collection rules, priced as a regulatory pack. Packs configure Govern; they are never separate products.

In one Indian residency domain, “in”: AWS ap-south-1, Mumbai. The proof record holds a hash of the payload, not the payload.