Mesachat Documentation
Mesachat connects AI agents to the places people already talk — Telegram and email — and gives you an admin console to configure them.
What is Mesachat?
- AI conversations — context-aware replies drawn from the conversation's recent history
- Two channels — Telegram and email. Others are extensible but not built.
- Model choice — OpenAI or Anthropic, per bot
- Bot builder — configure and connect bots without writing code
- Multi-tenant — isolated workspaces for teams and organizations
- API access — a REST API for programmatic control
Conversations happen in your messaging app. There is no web chat window; the web app is a companion surface, and several of its screens are not built yet.
Quick Navigation
| Section | Description |
|---|---|
| Quick Start | Get up and running in under 5 minutes |
| Key Concepts | Understand tenants, bots, agents, and channels |
| Using Mesa | Talking to bots, channels, apps and tools |
| The web app | What each screen does today |
| Admin console | Creating bots, integrations and keys |
| Enterprise | Multi-tenant, SSO, and admin features |
| API Reference | REST API documentation |
How It Works
graph LR
A[User Message] --> B[Channel]
B --> C[Bot]
C --> D[AI Agent Router]
D --> E[Specialized Agent]
E --> F[Response]
F --> B
- A user sends a message through a channel (Telegram or email)
- Mesa checks the chat is authorized and this bot is assigned to it — if not, nothing happens
- The message is routed to the assigned bot
- The bot's router selects an agent, by rules or by asking a model
- The agent processes the request using its tools and the conversation's recent history
- The response is sent back through the same channel