we're a technology studio, not a law firm, so none of this is legal advice, but we build assistants that customers talk to, and the same five questions come up in every project. they also happen to be the ones people are starting to ask us back.
- does the customer know it's an assistant? say so in the first message.
- what can it do alone? looking up an order is fine. a refund above a set amount, or a cancellation, goes to a person.
- who steps in, and how fast? a human should be one tap away, and the customer should never feel stuck.
- what gets recorded? every action should be logged, and the history exportable, so you can show what happened and why.
- who owns it after launch? a named person, not "the team".
none of these need a legal degree. they need someone to decide before launch instead of after the first complaint. for what the rules actually require, start with the official EU sources and your own adviser.
if you're using ai with customers today, how many of the five could you answer?