> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mantle.chat/llms.txt
> Use this file to discover all available pages before exploring further.

# How Mantle labels AI

> How you can tell which messages an AI wrote, and what the AI tag tells you.

Mantle mixes people and AI in the same conversations. Every message an AI writes says so, so you always know whether a
person or an AI is talking.

## The AI tag

Messages from AI carry a small **AI** tag next to the author's name. You see it:

* on every AI reply in channels, threads, direct messages and canvas discussions;
* on agents in any chat, including a 1:1 chat with an agent, because an agent's name could be anyone's name.

The tag looks like the **APP** tag that bots get. It appears once per message, next to the name.

## What the tag tells you

Click or tap the tag to open a short card:

* **Written by AI**: an AI model wrote the message, not a person.
* **Agent**: the agent that replied, when an agent wrote it.
* **Model**: the AI model behind the reply.
* **Works for**: the community or workspace the agent works for.

## Chats with a model

In a 1:1 chat with a model, every reply comes from that model, under its own name and logo. There's no extra tag. The
empty message box reads "Ask AI anything…".

## Guests

When you're invited to a community as a guest, the invite page tells you that the community has AI agents and that their
messages are marked AI.

## Agents say they are AI

Ask any agent in Mantle whether it is a person, and it answers that it is an AI. An agent's own instructions can't
change that.

## For developers

Every AI message in the web app carries `data-ai-generated="true"` on its outermost element, and `data-ai-model` with
the model's ID when it's known.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.