Meta requires an automated chat to say it is automated, and people must always be able to reach a human. Reply Mitra handles both, so you never have to remember it.
The agent introduces itself as your AI assistant inside its first reply, in a friendly way. If a reply ever leaves that out, this line is added in front:
Asking for a person, with phrases like "talk to a person" or "are you a bot", hands the chat to you at any time.
When it appears
- In the first reply of every new conversation.
- When the agent picks a conversation back up after it was handed to a person.
It does not repeat in every message, so the conversation still reads naturally.
If someone asks "are you a bot?"
The agent always answers honestly that it is an automated assistant for your business, and offers to bring in a person. It never claims to be human, whatever its persona says.
This line is added in code, not left to the AI, so it is always there, and it cannot be turned off. The separate "Say the reply was written by AI" setting in Behaviour adds your own note under the first reply, if you want one as well.