# I’m building a travel agent that takes money. What do I lock down before it talks to real customers? — retained record

> Retained reddit record from reddit:AI_Agents.

Canonical: https://highsignal.app/data/records/735d5d7c710b1e9b

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:AI_Agents)  published 2026-11-11 10:49 ingested 2026-11-12 02:37

[original source ↗](https://www.reddit.com/r/AI_Agents/comments/1xteqmy/im_building_a_travel_agent_that_takes_money_what/)

## retained text



```
I’m building a small travel advisor. An agent plans the trip, charges a planning fee and sends people to book through partner links. No human in the chat.

The planning part is fine. I am more concerned around money.

It can promise a price that’s gone an hour later. It can tell someone a booking is refundable when it isn’t. An airline already lost a case because its chatbot made up a refund rule. And travel customers paste in all kinds of stuff like emails and screenshots. Any of that could carry instructions meant for my agent.

My plan so far is that the agent can’t state refund terms on its own and can’t change a paid plan without me. That doesn’t feel like enough.

If you run an agent that talks to paying customers, what did you lock down first? And what bit you that you didn’t see coming?
```



document fetched 2026-11-12 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1xteqmy/im_building_a_travel_agent_that_takes_money_what
