AI use-case register

Record AI uses and AI agent authority: owners, permitted actions, tool permissions, approvals and activity evidence.

For
AI owners, governance reviewers and workflow leads.
You leave with
A repeatable use-case record and a portfolio review decision.

Choose your format

DOCX is editable in Word or a compatible editor. PDF is ready to print. No signup is required.

Inside this template

The full document follows these sections. The first prompts are shown below so you can check the fit before downloading.

Identify the use

Repeat this record for each use case; maintain a shared index by ID and owner.

Identity and purpose

Record ID, title, business objective, workflow, affected people, intended benefit and current lifecycle stage.

Accountable ownership

Name the business owner, decision authority, delivery owner and human reviewer. Record who can stop the use.

Data and dependencies

List data owners, sources, quality requirements, access restrictions, integrations and supplier or model dependencies.

Scope and boundaries

Record permitted users, uses and data, prohibited uses and the manual fallback. Link the supplier decision sheet.

Assess and approve

Use stages: proposal, assessment, pilot approval, evaluation, operation and review. Record changes and retirement.

Risk and oversight

Link privacy, security, legal and impact reviews as applicable. Record review standards, exceptions and escalation owners.

Decision and authority

Record the requested decision, authorised approver, date, rationale and conditions. Retain the previous decision.

Evidence and missing items

Link source records, supplier reviews, baseline measures, test results and open questions, with evidence owners and due dates.

Pilot boundaries

Specify test data, participants, duration, acceptance criteria, human review and stop conditions. Link the pilot scorecard.

Review the portfolio

Review each record together with the portfolio’s value, capacity and dependencies.

Benefits evidence

Compare observed quality, total work and review effort, costs and usefulness with the baseline. Distinguish observations from estimates.

Changes and incidents

Record supplier or model changes, data changes, incidents and lessons. Identify new approvals, controls or training needed.

Portfolio decision

Choose pilot, continue, change, scale or retire. Record rationale, approver, conditions and effects on other use cases.

Follow-up and retirement

Assign actions, owners and next review date. For retirement, plan access removal, data handling, supplier exit and a validated fallback.

Agent authority

Define the authority granted to this use. A technical specialist must validate that the configured controls enforce it.

Actions and authorising owner

Name who grants authority, the permitted and prohibited actions and their limits. In the fictional support example, reading approved guidance and drafting are allowed; sending replies and changing customer records are not.

Tool and data permissions

List tools, data, account or agent identity and read/write permissions. Name the technical owner and retain a configuration test. Do not assume a user’s full access belongs to the agent.

Approval events and evidence

Specify which action needs approval, by whom and on what evidence. Bind approval to the actual action and target; record agent, approver, time, outcome and exceptions with suitable access and retention.

Pause, revoke and reassess

Name who can pause actions and revoke delegated access. Check queued/in-flight work, retain logs and define recovery and reauthorisation. Reassess material changes before changed operation.

Need clearer footing for an AI decision?

Start with a focused conversation about a live AI use case, workflow bottleneck, training need, or governance gap.

Contact

Talk to Ada Studio

Have an AI adoption question? Send Adriana a short message.