New: Cookbooks and AI ExplanationsStep-by-Step recipes to solve problems connected to Roadmaps and Cheat Sheets. Need more details? Use AI buttons for structured and simple explanations with concrete examples throughout the whole platform.Take a look
Your support agent keeps a confused customer talking for ten more turns instead of just handing them to a person.
What you'll have at the end
A rule set that hands a conversation to a person on low confidence, missing data, or a task it isn't allowed to do
You need
An agent already answering real user conversations, with a way to see each reply and whatever record or tool call produced it.
Not covered
Escalating because of sentiment, VIP status, or a customer's own request for a person, which are real triggers handled by different rules than the three built here.
Should your agent ask before it deletes, sends, or pays?
Covers the authority question for any agent action in general; this recipe narrows it to the moment a support conversation has to hand off.
Your agent never checks its own answer before handing it back
A different fix for the same doubt this recipe's confidence check catches: checking the answer itself instead of comparing two independent tries.
Should you build your own agent, or buy one?
The governance work either path still owes you covers the same approval-gate and audit-logging pattern this recipe builds for one narrower case.
Checked 15 Aug 2026
Part of the AI Engineering cookbook