Thanks to visit codestin.com
Credit goes to CertifiedData.io

CertifiedData.io
EU AI Act-aligned evidence layer · Decision records

AI Decision Ledger for audit-ready AI systems.

Log, sign, verify, and export AI decisions as tamper-evident audit records for audit-readiness, governance, and enterprise trust.

Built for teams that need to prove what happened, which system acted, what context was used, and whether the record was altered later.

RFC 8785 canonical JSONSHA-256 record hashesEd25519 signaturesIndependent verification

Trust jump

Don't trust the dashboard. Verify the record yourself.

The Decision Ledger demo generates a real signed record, then reveals a terminal command you can copy and run independently. If the signature or hash fails, verification fails.

$ curl -sX POST https://certifieddata.io/api/demo/decision/verify \
    -H 'content-type: application/json' \
    -d @signed-record.json | jq

{
  "signature_verified": true,
  "hash_verified": true,
  "key_id": "cd_dl_root_v1",
  "algorithm": "Ed25519"
}

No account. No API key. The demo signs a real record, then hands you the full payload + signature so you can run the same check anywhere.

The audit problem

Logs are not enough when the question becomes evidence.

Most AI systems already produce logs. The problem is that internal logs are usually mutable, application-bound, and not independently verifiable by an external reviewer.

1Show every decision your AI system made about this subject.
2Show which model or agent produced each decision.
3Show what context, policy, or artifact the decision depended on.
4Prove the record has not been modified since it was written.

This is not observability. This is evidence.

From internal logs to independently verifiable records.

Traditional logging

  • • Mutable records that can be edited or deleted
  • • Internal-only evidence that requires platform trust
  • • No deterministic payload hash
  • • No public signature verification

Decision Ledger

  • • RFC 8785 canonical payloads
  • • SHA-256 hashes linked into an append-only chain
  • • Ed25519 signatures for record authorship
  • • Independent verification without trusting the dashboard

How it works

AI decision → signed record → verified evidence.

01

Step 01

Log the decision

Send the decision event, actor, entity, outcome, rationale, and model context to Decision Ledger.

02

Step 02

Sign the record

The payload is canonicalized (RFC 8785), SHA-256 hashed, Ed25519 signed, and linked into an append-only chain.

03

Step 03

Verify yourself

Use the public verifier or the copyable terminal command to confirm hash and signature validity outside the dashboard.

04

Step 04

Export evidence

Share JSON bundles, public proof URLs, and audit-ready records with compliance, procurement, or regulators.

Live public evidence

Decision records are visible, signed, and re-verifiable.

Public logs help external reviewers inspect proof records without requiring access to the private system that produced them.

Decisions

Decision log

Signed AI decision records and public verification metadata.

Recent activity

dec_900a8033…·dataset_certified·2h ago
dec_818a5371…·dataset_generated·2h ago
View decision log →

Certificates

Certificate log

Artifact and dataset certificates remain available as supporting infrastructure.

Recent activity

cert_24e8c586…·issued·6h ago
cert_d40272ab…·issued·6h ago
View certificate log →

Datasets

Dataset registry

Certified datasets remain part of the proof layer, but not the homepage wedge.

Recent activity

cert_24e8c586…·sha256:sha256:a…·6h ago
cert_d40272ab…·sha256:sha256:a…·6h ago
View dataset registry →

Start with the proof moment

Generate a signed AI decision record and verify it yourself.

Run the live Decision Ledger demo, inspect the signed payload, copy the terminal verifier, and see the evidence layer work before you integrate anything.

Codestin Search App