← Back to resources

Briano buyer toolkit

A practical AI evaluation guide

Use the same questions to assess the experience, the controls and the work required to run it.

Planning worksheet. Any notes typed here stay in this page and are not sent to Briano or saved after you close it. Use Print / save as PDF to keep a copy.

1. Can it finish the intended job?

Choose a concrete request and define a useful outcome. Ask for a walkthrough from the first input to the final record in your system.

• Which parts are demonstrated, configured or still proposed?
• What source supports the answer or action?
• What happens when information is missing, conflicting or outdated?

2. Does it know when to involve a person?

Test an ambiguous request and an action that requires approval. Inspect the handoff, not only the automated response.

• Can the next person see the request, evidence and work already completed?
• Who is responsible for decisions and exceptions?
• How are customers told what will happen next?

3. What data and permissions does it need?

Map sources, processing and destinations. Ask for evidence specific to the configuration you are considering.

• Which systems and fields are needed? Can access be limited?
• Where is data processed, by which providers and for how long?
• What are the model-training, deletion and end-of-contract terms?

4. How will you know it works?

Agree the test set, baseline, sample size, quality review and success criteria in advance.

• Are resolution and automation metrics clearly defined?
• Do reported savings include review, rework and operating costs?
• Can you inspect failures and test again after a change?

5. What will your team need to operate it?

Put the implementation and ongoing responsibilities alongside the product demonstration.

• Who owns integrations, knowledge, rules, monitoring and support?
• What is included in the commercial proposal?
• What is the process to pause, roll back or expand the workflow?

Capture the decision