Small Business AI Pros buyer guide
Guide 5: Interview an AI consultant or implementer
Use consistent questions to compare how candidates discover, build, test, communicate, and hand off work.
Begin with discovery
Ask how the candidate learns a workflow, finds the system of record, identifies stakeholders, and distinguishes a process problem from a software problem. Ask what would make them advise against an AI approach.
Listen for questions about the people doing the work, the information involved, the review burden, and the final owner. A candidate who starts with a product demonstration may still be capable, but the business needs evidence that the operating problem comes first.
Probe the delivery boundary
Ask how access will be granted, how test data will be chosen, what environments will be used, how changes are reviewed, and how a failed test is contained. Ask who owns documentation, training, deployment, monitoring, and support.
Request examples of assumptions and risks in a normal proposal. The goal is not to force a perfect answer during an interview. It is to see whether the professional exposes uncertainty and gives the business useful decisions.
Record evidence consistently
Use the same core questions and a simple evidence sheet for each candidate. Separate the answer, supporting evidence, required follow-up, and your own unresolved concern. Do not score charisma as technical evidence.
A final decision may also require references, contract review, security review, financial checks, or a paid discovery stage. Keep those gates visible instead of treating the interview as complete verification.