All use cases
AI Credit Decisioning

An AI credit model declines a mortgage. The borrower appeals. One signed decision record, full provenance, defensible.

The situation

An AI credit model declines a mortgage application. The borrower appeals and the lender must defend the decision, the inputs that drove it, the model version, and proof the record reflects what actually happened, not a reconstruction. Aegis Trace produces one signed decision record with full provenance, so the appeal is met with evidence rather than narrative.

The regulations

  • FCA CONC
  • Consumer Duty
  • GDPR Article 22
  • Equality Act
  • ECOA (US)

The evidence

The credit decision is recorded with its full provenance.

POST /v1/decisions
{
  "decision": "decline_mortgage",
  "regime": "FCA_CONC",
  "inputs": { "affordability_ratio": 0.61, "ltv": 0.92 },
  "model": { "name": "credit-engine", "version": "7.4.0" }
}

A defensible, signed decision record.

{
  "certificate_id": "crt_cr_9e33",
  "status": "sealed",
  "decision": "decline_mortgage",
  "regime": "FCA_CONC",
  "issued_at": "2026-03-14T09:21:07Z",
  "inputs_hash": "sha256:3f8a…c1d2",
  "model": { "name": "decision-engine", "version": "4.2.1" },
  "signature": {
    "alg": "ed25519",
    "value": "MEUCIQD…Qz9w",
    "anchored": "rfc3161"
  }
}

What the regulator receives

A single signed record showing the affordability inputs, the model version, and tamper-evident proof of provenance, a decline that holds up on appeal.

Evidence the moment the decision is made.