Platform
An AI register that stays true as things change
Aicura's AI register is one up-to-date list of every AI system and agent your organisation uses, with who owns it, where it came from, the models it runs on and its current risk rating. Each change that matters saves a new version, so you can see any system exactly as it stood on any date.
Included in every plan, starting with the free Register plan. AI use requests are in Pro. Compare plans
Know what AI you use, every time someone asks
The board asks what AI the organisation uses. So does the insurer's questionnaire, the customer's due diligence form and the auditor scoping next year's work. A spreadsheet answers once and is out of date by the next meeting. The register keeps answering, with an owner and a history behind every line.
AI systems and agents in one list
Each line on the register, called an entry, is an AI system or an agent. It shows the business unit, where the system came from (its origin), the risk rating from its latest completed assessment, its lifecycle stage and its status.
- Origin is built in-house, bought from a vendor, or a mix of both.
- The lifecycle is where the system is in its life, from Proposed through In Development, Testing and Live to Superseded or Retired.
- An entry with no completed assessment reads Not Assessed. It is never assumed to be low risk.
Roles can be limited to business units, so each team sees its own systems and everything that belongs to them.
Start from the library, not a blank form
Adding an entry opens a product search first. Pick a known AI product or agent from Aicura's library and the form arrives prefilled, so you check and correct rather than type from nothing.
For a system the library does not know, download the helper document, have the technical team fill it in offline and upload it. Or start blank.
Everything about a system on one page
An entry's page holds the system's whole profile and the work done against it. Tabs cover its details and owners, the data and tools it uses, the agent identities it runs under, provider warnings about its models, the documents written for the people it affects, evidence, guardrails, sign-offs, testing, supporting files and versions.
- Owners over time. Designated owner, executive accountability and technical contact, each with a dated history. Pick an "as at" date to see who owned it then.
- Profile. What data it uses, what it decides and who that affects, how people oversee it, and what you tell people about it.
- Supporting files. Specifications, vendor confirmations and other documents attached to the entry.
See any system exactly as it stood on a past date
Saving a change that matters keeps the old version and saves a new one, then re-runs every scan and assessment that depends on it. Open any version to see the entry exactly as it stood.
A save never moves the lifecycle stage. Changing stage is its own step, recorded with the version, who moved it and a reason. Retired, Declined and Withdrawn entries become Inactive and stay on the register.
Every change to a record is written to the audit log on every plan, with what changed, who changed it and when.
Know which models you rely on, and what providers say about them
The Models tab lists the models your AI systems and agents run on. A model is a known provider model, your own fine-tune of one, or a model you built yourself.
- Your own status for the model (Approved, Restricted, Deprecated, Retired) sits beside the provider's stated status, and the two can disagree.
- Used by shows every system and agent that runs on the model.
- When a warning, called an advisory, is published against a model, each entry records its response as Acknowledged, Replacing the model, Accepting it or Not applicable.
Pro
Let staff ask before they start using new AI
The Requests tab is the front door for new AI and for changes to what is already registered. A request saves a proposed version of the entry, and the reviewer's decision moves it forward or stops it.
- Aicura flags points from the answers that need a closer look, such as training on your own data or named data sets holding personal information.
- Reviewers can ask for clarification, discuss, attach files, or pass it to a committee whose decision is recorded back on the request.
- Approval routes name the approvers for each business unit. Set requests to Optional, or Required so every new system arrives through one.
Questions
What is AI register software?
AI register software keeps an up-to-date list of every AI system and agent an organisation uses. Each one has a named owner, a note of whether it was built in-house or bought, the models it runs on, where it is between proposed and retired, and its current risk rating. Good register software saves a new version as each system changes, so you can say what it looked like on any past date, and it connects each system to the assessments, rules and monitoring that depend on it.
Is an AI register the same as an AI inventory?
They describe the same list. Inventory is the usual word for finding out what AI you run. Register is the usual word for the version you keep up to date, with an accountable owner, a history and a stage from proposed to retired for each system. Aicura does both jobs in one place. Discovery finds what you run and the AI register keeps it current.
Is the AI register free?
Yes. Aicura's AI register for systems and agents, discovery, prefill from the product library, the audit log and AI system and agent reports are all in the free Register plan. Paid plans add assessments, policies, enforcement and monitoring on top of the same register, so nothing is rebuilt when you move up.
Does anything join the AI register without a person deciding?
No. Discovery suggests and a person decides. A system joins Aicura's AI register when someone adds it, accepts something discovery found or approves an AI use request. If a connected source created the entry, that source is named on it.
Can staff request new AI through the register?
Yes, from the Pro plan. AI use requests in Aicura let anyone propose a new system, a new agent or a change to one already on the register. The request goes to a reviewer, who can ask for clarification or pass it to a committee, and it ends approved, declined or withdrawn. Requests can be optional, or required so that every new system arrives through one.
Start your AI register today
Create a free account, add your first system or agent from the library, and connect a discovery source to find the ones nobody mentioned.