Built for the EU AI Act · 2024/1689

Compliance you can prove, not just claim

Run an audit against the laws that apply to you, then hand regulators a signed binder backed by tamper-evident evidence.

Explore the platform
  1. Step 1Register your AI

    List every model, vendor and system you run.

  2. Step 2Map it to the law

    We match each system to the controls that apply.

  3. Step 3Collect the evidence

    Every result is chained so it cannot be altered.

  4. Step 4Sign the binder

    Hand regulators a signed, timestamped binder.

Backed by an append-only SHA-256 hash chain, Merkle proofs and write-once storage

WHAT AXIS DOES

Six things, in plain language

Audit against every law that applies

Pick your regulations and run one audit. We scan your code and AI systems against the EU AI Act, SOC 2, NIST AI RMF and OWASP LLM, and map each finding to a control.

Evidence nobody can quietly change

Every result is written to an append-only record, chained with SHA-256 hashes and timestamped. If a record changes, the chain breaks.

A signed binder for your regulator

Get a regulator-ready binder, digitally signed and archived as PDF/A, that an auditor can check without trusting us.

Your AI inventory, always current

Register every model, vendor and asset once. We map each one to its obligations and export an AI bill of materials on request.

Compliance posture at a glance

See your coverage per framework, your open findings and a plan of actions and milestones, so you know what to fix next.

Move legacy systems safely

Migrating old COBOL estates? We mask personal data first, build a twin for review, and keep a rollback path for every step.

THE DIFFERENCE

The annual scramble, or an audit you can repeat

Today

  • Screenshots and spreadsheets gathered by hand
  • Evidence that can be edited after the fact
  • No clear view of which laws apply
  • Findings discovered by the regulator first

With Axis

  • One audit across every law you choose
  • Append-only, hash-chained evidence
  • A posture score and a plan of actions
  • A signed binder you can hand over
01 — POSTURE

Know exactly where you stand.

Covered, partial and gap controls for each framework, updated as evidence arrives — and flagged when it drifts.

EU AI Act · controls (example)

  • Risk management system✓ Covered
  • Data governance◐ Partial
  • Human oversight✕ Gap
  • Record-keeping✓ Covered

Regulatory audit · your repository (example)

  • src/model/train.py:42✕ EU AI Act Art. 10
  • api/predict.py:118✓ passes
  • config/logging.yaml:7✕ Art. 12 record-keeping
  • docs/model-card.md✓ passes
02 — AUDIT

Audit your code against the law.

Point Axis at a repository and get a per-law report — complies or doesn't, down to the folder and line.

03 — PROVE

Hand auditors proof, not promises.

Signed PDF/A-3 binders and evidence packs your auditors can verify independently.

Audit binder (example)

  • BinderEU AI Act — evidence pack
  • SignatureXAdES-LTA ✓
  • FormatPDF/A-3
  • TimestampRFC 3161 ✓
5compliance frameworks mapped side by side
SHA-256hash-chained, append-only evidence
RFC 3161trusted timestamps on every record
XAdES-LTAlong-term signatures on audit binders
WHY NOW

The rules we check you against

Real penalties, real dates. The 2026 Digital Omnibus moved several deadlines. It did not remove the requirements.

€35M or 7%Maximum fine for prohibited AI practices, whichever is higher.
€15M or 3%Maximum fine for most other obligations, including those for high-risk systems.
2 Dec 2027New date for most high-risk (Annex III) obligations, moved from 2 Aug 2026.
2 Aug 2028New date for high-risk AI built into regulated products (Annex I).

Sources: Gibson Dunn · AI Act Explorer · Orrick

Ready when your auditor is.

Sign in to see your posture, evidence and audits.