AI products change.
Customer assurance does not.
HumanWox links every AI system to an accountable owner and a connected record of its risks, controls, evidence, decisions and reviews. When a customer asks how the system is run, the answer is already there.
Every release changes what you would have to show.
Models get swapped. Features ship. Owners change. Reviews happen, or they slip. Each of those changes leaves the last assurance answer a little less true.
The difficulty is not answering the first questionnaire. It is answering a later one about a system that has changed since anyone last looked at it.
Everything starts with the AI system.
Each AI system carries its own record: who owns it, where it sits in its lifecycle, what it is built on, who provides oversight. The system is the object. Everything else connects to it.
Every decision stays connected to what informed it.
Risks lead to controls. Controls are evidenced. Evidence supports decisions. Decisions are reviewed.
Ask why a system was approved and the record answers with the risk it addressed, the control applied, the evidence behind it and the person who signed it off.
Every answer stays connected to the decisions that created it.
- Risk
- Control
- Evidence
- Decision
- Review
The report your customer asked for.
Generate an assurance report for any AI system: its register entry, the chain behind it, its current status and its review history.
Your customer receives a document carrying a verification code. They scan it and confirm the report has not been altered since it was issued.
The report is not a certification and it is not an audit opinion. It is your record, with proof that it has not been changed.
Tamper-evident verification runs at verify.humanwox.com. Anyone holding a report can check it without an account.
Built in alignment with BS 10008.
A person approves every record.
Every record carries its origin: declared by a person, detected by the platform, or suggested by the platform and approved by a person.
Nothing reaches the accountability record without someone approving it. The record you hand a customer is a record a named person stood behind.
Point-in-time record.
See how the record stood on any date.
An incident, a customer review, an audit. Each asks the same question: what did you know then, and what had you done about it?
Choose a date and the record returns its state as at that date, with the origin of each entry.
AI Discovery.
You cannot account for a system nobody registered.
Detection reconciles what your register says against what the organisation is actually using. Attestation campaigns ask owners directly and record what they answer.
Anything that turns up is staged for approval. Nothing is written into the record automatically.
Obligations.
Track requirements, actions and changes across connected records.
Controls, reviews and framework requirements carry due dates and owners. What has slipped surfaces here, before a questionnaire finds it.
Frameworks are optional. Your systems come first.
The accountability record stands on its own. Activate a framework when a customer or an auditor asks for the mapping.
Supported: ISO/IEC 42001, the EU AI Act, NIST AI RMF and DSIT AIME. Nothing in the product depends on activating one.
Access, API and audit trail.
Role-based access scoped to your organisation. An append-only activity log covering every change. An inbound event API with keys, so governance signals from your own systems reach the record. Partner and client workspaces where you need them.
What shapes the price.
Pricing reflects:
- How many AI systems sit in your register
- How many people need access
- Which framework overlays you need, if any
- How many assurance reports you issue, and who receives them
- Whether you need partner or client workspaces
A walkthrough sets it. Twenty minutes, no deck.
Questions,
answered.
Trust should keep up with the product.
HumanWox maintains the accountability record behind every AI system, so customer assurance, audit and regulatory review do not start from scratch each time something changes.