The AI That Lost a Customer: Fixing Over‑Questioning in Chatbots

The customer wanted a booking link. The AI wanted their industry, team size, budget, and annual revenue. It answered every question and still lost the lead. In this 24-minute workshop, follow a fictional customer chat where an AI agent turns a simple booking request into an endless questionnaire. Then watch how to repair the conversation. Test your judgment with 14 practical challenges covering: • Recognizing what the customer actually wants. • Asking useful questions without creating unnecessary barriers. • Making optional qualification genuinely optional. • Providing the booking route without promising an unconfirmed appointment. • Handling failed forms, human-help requests, and customer refusals. • Measuring progress beyond completed questionnaires. TIMESTAMPS 00:00 The customer walks away 03:00 Diagnose the friction 06:00 Repair the booking flow 09:00 Qualify without interrogating 12:00 Handle tool failures and handoffs 15:00 Stress-test the conversation 18:00 Measure outcomes honestly 21:00 Write the repair brief Pause at each challenge, choose your response, and compare it with the worked explanation. Which question would you remove first and what should the agent say instead? Share your improved reply in the comments. Subscribe to The AI Scope Lab for practical lessons on AI agents, evaluation, data annotation, and quality assurance. This visual workshop uses on-screen text and instrumental music, with no spoken narration. All chats, service rules, links, and results are fictional teaching examples. Illustrations are AI-generated. #AIAgents #AIEvaluation #ConversationalAI