AI vendor review worksheet

Review AI suppliers and AI agent permissions, total cost, activity evidence, revocation and exit before purchase.

For
Business owners, procurement, IT and governance teams.
You leave with
A documented review, unresolved questions and a clear purchase or pilot 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.

Define the proposed use

Start with the work and the people involved, before comparing product features.

Tool and proposed use

Name the supplier, product, version or service plan, intended workflow and expected benefit.

Owner and participants

Identify the business owner, users and the people reviewing security, privacy and procurement.

Scope and affected people

Describe where the output will be used, who may be affected and which uses are outside this review.

Decision and timing

Record the decision being sought, the target date and the person authorised to approve it.

Follow the data

Ask for written terms or evidence. Record the source and date of each answer.

Data entering the tool

List public, internal, confidential and personal data. Identify anything that must stay out.

Training and reuse

Can inputs or outputs be used for model training or service improvement? Record the default, settings and contractual terms.

Location and other providers

Where is data processed and stored? List relevant sub-processors and any questions for the privacy reviewer.

Retention and deletion

How long are prompts, files and outputs kept? Can the organisation delete or retrieve them, including when leaving the service?

Check control and review

Connect the supplier’s promises to the way your team will actually use the tool.

Access and security

Demonstrate allowed and denied tool/data permissions in the proposed configuration. Identify account ownership, authentication, the authorising owner and technical enforcement evidence.

Human review and traceability

Show a reviewer the sources and actual action/target before approval. Demonstrate denial without approval and exportable logs linking agent, approver, action, result and exceptions.

Changes and incidents

Demonstrate delegated-access revocation, queued/in-flight action handling and recovery. Agree change and incident notices, escalation contacts and who reauthorises operation.

Evidence still missing

List unanswered questions, the evidence needed, who will obtain it and when.

Decide and follow up

Approval should describe a specific use and its conditions, not every possible use of a product.

Pilot boundaries

Record permitted users, data, duration, success measures and conditions that would pause the pilot.

Decision and reasons

Choose: proceed, proceed with conditions, defer or decline. Record the rationale and any alternatives considered.

Actions and accountability

For each condition, record an owner, due date and evidence of completion. Add rows as needed.

Approval and review

Record approver, decision date and next review. Reassess material changes to purpose, data, tools, permissions, autonomy, model or supplier before changed operation.

Compare sourcing options

Assess business fit and the whole cost of the proposed use, with evidence for each option.

Business fit and alternatives

Compare buying, building, partnering and the manual process against the same business objective and acceptance criteria.

Total cost

Estimate setup, licences, integration, staff training, human review, support, maintenance and exit costs. Record assumptions and uncertainty.

Compatibility and support

Check data formats, integrations, capacity, support terms, escalation contacts and missing supplier evidence with your technical reviewers.

Exit arrangements

Test usable export of data and action logs, access and credential revocation, deletion evidence and manual fallback. Record migration effort, lock-in, notice periods, support and recovery arrangements.

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.