When to use this log
- A chatbot answers with the wrong price, availability, policy, refund, service detail, or escalation path.
- A customer asks for a human but the conversation remains stuck in automation.
- An AI draft looks plausible but lacks the source note, CRM context, or owner approval needed before replying.
STOP AUTOMATION: do not let the bot send follow-up promises about refunds, legal/policy decisions, health/safety issues, payments, warranties, access, scheduling, or service availability until a human owner reviews the source record.
Copy/paste chatbot handoff failure card
AI chatbot handoff failure log
Date/time and timezone:
Customer or ticket ID:
Channel: website chat / SMS / inbox / social DM / support portal
Conversation link or transcript:
Customer request in one sentence:
What the chatbot/AI draft said or failed to do:
Risk label: low / medium / high / legal-policy-finance-safety-sensitive
Correct source record or policy URL:
CRM/order/job/account record checked:
Human owner:
Customer impact if not fixed:
Immediate customer-safe update needed? yes/no
Root cause: missing source / stale source / weak prompt / no escalation rule / integration gap / unclear owner
Fix owner:
Fix due date:
Re-test prompt or scenario:
Next review date:
Customer-safe handoff note
Hi [Name] — thanks for flagging this. I am moving this to a person so we can check the right source before replying with a firm answer.
What we are checking: [policy/order/job/source]
Human owner: [name/team]
Expected update window: [safe window]
We do not want the automated reply to overpromise. We will confirm the accurate answer after review.
Weekly fix review
- Group failures by root cause: stale source, missing escalation rule, unclear customer-status field, or risky prompt wording.
- Update the public/internal source page before changing the prompt; otherwise the chatbot may keep guessing.
- Add a handoff rule for sensitive topics: billing, refunds, legal/policy, medical/safety, warranty, identity, tenant/access, pricing, and schedule guarantees.
- Re-test the exact customer scenario and keep screenshots or transcripts as proof.
AI review prompt
You are reviewing an AI chatbot handoff failure for a small business. Compare the transcript, source record, proposed customer update, and escalation rule. Flag unsupported promises, missing source proof, sensitive topics needing human review, and unclear owner/due-date fields. Return: (1) safe response draft, (2) hold-for-owner reasons, (3) source/prompt/integration fix, and (4) re-test scenario.
Product bridge
This worksheet is a practical sample of the AI workflow owner, human-review, and customer-safe prompt patterns inside The Small Business AI Profit Kit. Use it when one chatbot failure reveals a wider need for AI workflow guardrails and source-proof checklists.