THE END-TO-END PLATFORM FOR AI COMPLIANCE

The system of record for AI compliance.

Register every AI system and agent once. Documents, questionnaire answers and your public trust page all derive from that record — with a citation behind every claim.

Request a demo Explore the platform

One operating record

systems, agents, owners, approvals and controls held in one live registry

Framework-ready

global regulations, standards and buyer controls mapped from the same source

Evidence-linked

documents, answers and trust disclosures stay tied to underlying records

OPERATING RECORD

LIVE

AUR-114Autonomous underwriting agentHigh risk

OWN-008Owners and approvals assigned4 teams

EVD-031Controls and evidence linked14 records

Coverage

EU AI ActMapped

ISO/IEC 42001Mapped

Buyer security reviewReady

Trust state

One source powers the full stack, clearly.

Inventory, obligations, documents, diligence answers and trust disclosures remain in sync because they all resolve to the same reviewed record.

THE FULL STACK, AT A GLANCE

Classify

Risk tier, obligations and owners

Document

Technical files, policies and packs

Answer

Buyer, security and procurement reviews

Trust

Public disclosures and gated evidence

The standard for AI governance

Registry, classification, documentation, buyer diligence and public trust — unified on one evidence base.

Systems and agents share the same obligations, so Hael governs both from one record. The work that sits today across legal, security, engineering and compliance runs on a single source of evidence — built to the standard auditors, buyers and regulators already expect.

THE PLATFORM

One registry. Systems and agents. Everything derives from it.

Systems or agents, the obligations are the same — and Hael governs both from one registry. Register an AI system or an autonomous agent once: classification, documents, questionnaire answers and your public trust page all draw from the same record, and stay in sync because they share a source.

AI SYSTEMS + AGENTS · ONE REGISTRY

UND-014 · Underwriting Copilot (agent) · High

FRD-021 · Fraud Scoring Model · Limited

CHT-031 · Customer Support Chat (LLM) · Minimal

Audit-ready files

The Annex IV, ISO 42001 or model-card pack — written, sourced, ready to send.

Buyer questionnaires

Security and procurement answers, drawn from your own evidence.

Public trust page

A live page customers and buyers can read before they ask.

Regulator filings

What the EU AI database expects — prepared in the same place.

Five modules. One source of truth.

Who we are

The category standard for AI compliance.

Hael is the platform organisations use to prove every AI system and every AI agent they run is lawful, tested and accountable — designed so the people who read the file can trust it, and the people who write it can defend it.

One registry for systems and agents

Models, LLM features and autonomous agents all sit on the same record, under the same obligations — not a separate module bolted on later.

Regulatory counsel in the loop

Shaped by practising counsel — not consultants writing about regulation from the outside.

Mapped across global AI frameworks

Outputs tie back to the specific articles, controls and functions your team is working against — not a hand-wavy framework badge.

Security-grade buyer trust

Trust-center sharing, review trails and evidence-linked answers give buyers the feel of a mature assurance programme, not ad hoc AI paperwork.

Purpose-built for AI compliance

Built from day one to produce the documentation regulators, auditors and procurement teams actually read — not a GRC tool with an AI module bolted on.

Built to earn trust quickly

The homepage, trust center and buyer outputs are designed to signal control, seriousness and completeness the moment a prospect or auditor lands on them.

Continuous, not point-in-time

When a system or agent changes, dependent documents are flagged stale and regenerated with a visible diff. Your file is current the day the regulator asks for it.

WHY HAEL

Built for the documentation, not the checklist.

Evidence first

Every answer, document and disclosure resolves to a record, control, approval or source in your operating registry. Buyers and auditors can follow the trail without asking your team to reconstruct it by hand.

Reviewable by design

Hael is built so legal, compliance, security and auditors can inspect what changed, who approved it and what evidence supports it before anything is shared.

Always current

When a system or agent changes, dependent documents, questionnaire answers and trust-center disclosures are flagged stale and regenerated from the updated source record.

Product · Registry

Every AI system, on the record.

One record per system: what it does, what data it touches, who owns it, what risk it carries. Seven questions to register; five minutes; no statute reading. The registry is the single source every document, answer and page derives from — change a system, and everything built on it knows.

Explore Registry →

Classification · UND-014

Underwriting copilot agent

High risk · Annex III(5)

Plain-English verdict

This is high-risk because the agent materially influences a consumer's access to credit.

Maintain risk management and post-market monitoring

Evidence human oversight before deployment approval

Document training data controls and testing results

Obligation packs

Art. 9Open · 3 controls outstanding

Risk management

Art. 10Ready · evidence mapped

Data governance

Art. 14Review · owner assigned

Human oversight

Product · Classify

Turn AI inventory into obligations, instantly.

Hael classifies every system and agent, explains the decision in plain language, and maps the exact obligations, owners and next actions across the frameworks that apply.

Explore Classify →

Product · Documents

Produce documentation that stands up to scrutiny.

Technical files, transparency notices, readiness packs, model cards and buyer evidence sets — generated from the live record, reviewed before release and kept aligned as the system changes.

Explore Documents →

Generated set

Annex IV file

Ready for review

Article 50 notice

Approved

ISO 42001 pack

Evidence linked

Buyer packet

Pending sign-off

Annex IV technical documentation

AUR-001 · draft v7

second-eye review passed

§3.2 Risk management measures3 sources attached

Residual model risk is controlled through challenger-model review, weekly population sampling and quarterly governance sign-off against the approved risk plan.

Source

Registry control AUR-001 / Art. 9(2)(d)

Reviewer

Independent QA counsel check completed

Status

Ready for export and buyer review

Buyer diligence · AI governance and security review

18 / 24 drafted

How do you govern AI systems and agents in production?

Each system and agent runs from a named operating record with owners, approval history, mapped controls and linked evidence.

Registry-backed4 linked sources

How do you evidence review and oversight before release?

Hael shows approvers, review dates, control status and the underlying records used to clear the release.

Review trail3 approvals

Answer controls

Approvals

Legal, security, compliance

Sources

Registry, controls, decision logs

Share state

Ready after final review

Ready to send

The same record powers the answer, the file and the trust center.

Product · Questionnaires

Answer diligence at the speed of your evidence.

Upload the questionnaire. Hael drafts each response from approved records, shows exactly what supports it and keeps your answers aligned with the documents and trust center the buyer will inspect next.

Explore Questionnaires →

Product · Trust Center

Publish an AI assurance layer buyers recognise immediately.

Give customers a serious trust center for AI: disclosed systems, framework coverage, review status, named owners and controlled evidence access — all generated from the same live registry your documents and diligence answers use.

Explore Trust Center →

Trust Center · live assurance portal

Show the AI programme buyers expect to see.

Live registry-backedNDA-gated evidence

Autonomous underwriting agent

High risk

Status

Reviewed

Coverage

EU + ISO

Evidence

9 linked controls

Claims triage copilot

Limited risk

Status

Reviewed

Coverage

EU + ISO

Evidence

6 linked controls

Customer support assistant

Minimal risk

Status

Reviewed

Coverage

EU + ISO

Evidence

4 linked controls

Assurance summary

Latest review state

Reviewed by legal, security and compliance.

Buyer-facing disclosures, control mappings and requested evidence are generated from the same approved record.

Framework coverage

EU AI Act, ISO/IEC 42001, NIST AI RMF, GDPR and buyer security controls mapped from one source.

Available packages

Buyer diligence pack, control mapping export, technical file excerpts and review trail — all access controlled.

Trust signals

Named owners, latest approvals, control status and evidence counts visible before a prospect asks for them.

Annex IV technical file

Underwriting copilot · v2.4

100% grounded

The system is classified as high-risk under Annex III(5) of the EU AI Act. Residual model risk is controlled through challenger-model review, weekly population sampling and quarterly governance sign-off against the approved risk plan. Human oversight is documented under control AUR-001.

Source · AUR-001 Risk managementReviewed · counsel

Weekly override sampling cadence — recorded in the system registry, mapped to Article 9(2)(d) of the EU AI Act.

Every sentence linked to a source record14 / 14 cited

Product · Provenance

100% ground truth. Every line, sourced.

Every sentence in a Hael document is tied to a record in your registry — the system it describes, the obligation it satisfies, the person who approved it. Your reviewer, your auditor and your buyer can follow that trail end-to-end.