Hael
Sign inRequest a demo
PRODUCT · CLASSIFY

Every system, classified against every law that applies to it.

Classify determines the risk tier and the exact obligations for each AI system across every framework in scope — generated from the registry record, in plain language, with the citation behind every decision.

hael.app / classify / SYS-0148
Classification — Underwriting decision model
Generated from the registry record. Each decision is cited.
EU AI Act
High-risk
Annex III §5(a) — credit scoring
ISO/IEC 42001
In scope
Clauses 6.1.2, 8.3 — AI risk & impact
NIST AI RMF
Govern + Manage
Functions GV-1, MG-2
GDPR Art. 22
Applies
Solely automated decision with legal effect
Colorado AI Act
Consequential
§6-1-1701(3)(b)(III) — financial services
NYC LL144
Out of scope
No employment decision
THE PROBLEM

Classification is where compliance teams stall.

An AI system can be high-risk under the EU AI Act, in scope of ISO/IEC 42001, subject to GDPR Article 22, and a consequential decision under the Colorado AI Act — all at once. Working that out by hand, per system, per framework, is where governance programmes either slow to a halt or quietly produce answers no one can defend.

A classification you cannot trace back to a clause and a fact in the registry is not a classification — it is an assertion. Boards and regulators ask for the reasoning, not the answer.

HOW IT WORKS

One engine. Every framework. Every system.

Classify runs the same decision logic across every framework you are in scope of, against the same registry record — so the answers are consistent, traceable, and current.

01
Read the registry record
Purpose, data, third parties, jurisdictional scope and deployment context come straight from the system's record — no parallel questionnaire to fill in.
02
Run every framework in scope
EU AI Act, ISO/IEC 42001, NIST AI RMF, GDPR Article 22, Colorado AI Act, NYC LL144, Texas TRAIGA, California ADMT and the rest run together against the same inputs.
03
Return the decision in plain language
Each result names the tier or status, the obligations triggered, and the clause or article that produced it — readable by an executive, defensible to an auditor.
04
Re-run on change
When the registry changes — new data, new use, new jurisdiction — affected classifications re-run and surface the delta for owner review.
THE SYSTEM-OF-RECORD SPINE

Classification is a view of the registry, not a separate database.

The registry is the source. Classify reads from it; documents, questionnaires, the trust page and monitoring all read from the same record — so the tier shown to the auditor matches the tier shown to the buyer matches the tier in your Annex IV file.

Registry
Each system's record supplies the facts that drive every classification — owner, data, third parties, lifecycle.
Documentation
Annex IV technical files, FRIAs and model cards are produced against the classification — so the artefact matches the decision.
Questionnaire answers
Buyer questionnaires answer 'what is its risk tier' from the same classification, with the same citation.
Trust page
The public trust page surfaces the classifications buyers ask about, drawn from the same source.
Monitoring
Material changes to a system re-trigger classification and flag dependent artefacts as stale until re-reviewed.
Agent governance
Agents are classified alongside systems — purpose, tools and scope feed the same engine.
ONE ENGINE, THIRTEEN FRAMEWORKS

Classify once. Every framework in scope is decided against the same record.

The same registry record is classified against thirteen frameworks at once. Adding a new regime — a state law, a sector code — is a configuration change to the engine, not a rebuild of how your AI systems are described.

One classification engine
EU AI Act
GDPR
ISO/IEC 42001
NIST AI RMF
Colorado AI Act
Texas RAIGA
NYC LL 144
California ADMT
Illinois HB 3773
Utah AI Policy Act
Korea AI Basic Act
China Gen-AI / Algo
UK AI principles
Thirteen frameworks · one decision per system · adding a regime is configuration
CLASSIFY

Bring a system. See the decisions, with the citations.

Pick one of your live AI systems. We'll register it on the call and show every framework decision it triggers — the tier, the obligations, and the clause behind each.