What information should a developer collect from the business before designing an AI agent?

An AI agent should not start with the question “Which AI technology should we use?”—it should start with the business problem and expected outcome.
What information should developers collect about users, processes, data sources, business rules, permissions, integrations, risks, success metrics, and human approvals?
Understanding these requirements early can prevent expensive redesigns when the agent moves from prototype to production.
Explore the key questions developers and solution architects should ask the business before designing an enterprise AI agent.