A team can pass internal evals and still build the wrong agent. Customer interviews catch a different class of risk: unclear jobs, hidden approval needs, missing trust signals, and workflows that users would rather keep manual.
Start with the current workflow
Ask the customer to describe the last time the task happened. Do not begin with the proposed agent. The useful detail is in the messy current process: who starts it, what data they need, where they hesitate, and what makes the work painful.
Find the cost of a wrong action
A launch team needs to know what happens if the agent is confidently wrong. Ask what a bad answer, wrong status update, mistaken refund, or unsafe recommendation would cost in time, money, trust, or compliance overhead.
Test the trust boundary
Ask which actions the customer would allow automatically, which actions need review, and which actions should never be delegated. This is often more useful than asking whether they “””like AI.”””
Look for procurement language
If the customer mentions audit logs, permissions, data retention, vendor review, security review, or legal approval, capture those words. They reveal what the buying committee will care about later.
End with a concrete next step
Ask whether the customer would try a small version of the workflow, review a sample report, or share a redacted failure case. Real next steps are better evidence than polite compliments.
Quick checklist
- Ask about the last real workflow, not a hypothetical future workflow.
- Record exact phrases customers use for risk, trust, and approval.
- Separate user pain from buyer requirements.
- Ask what would make them stop using the agent.
- Close with a small concrete commitment.
How to use this
Use this guide as a working review aid, not as a compliance certificate. A team preparing an AI agent for production should pair it with traces, test cases, policy decisions, and a named owner for each unresolved risk.
Related IBBS resources
- AI Agent Readiness FAQ for Founders
- AI Agent Readiness Self-Assessment
- Sample AI Agent Readiness Audit Report
AI Agent Customer Interview Questions Before Launch: A practical interview guide for learning whether an AI agent solves a real customer problem before the team expands scope. Read it at https://ibbs.ai/2026/07/07/ai-agent-customer-interview-questions/
If your team is preparing an AI agent for production and wants an outside review, start with the AI Agent Readiness Self-Assessment or review the sample audit report.