Moltbot: Your Personal AI Assistant Across Every Messaging Channel

What if you could have a single AI assistant that lives on your own devices and answers you on every messaging platform you already use — WhatsApp, Telegram, Slack, Discord, Signal, iMessage, Google Chat, Microsoft Teams, and more? Moltbot is exactly that: a local-first, personal AI assistant with a unified gateway architecture that connects to 14+ messaging channels, supports voice interaction, includes browser automation, and runs entirely under your control. No cloud dependency, no vendor lock-in — just a fast, always-on assistant that feels native to every platform it touches.

The Problem

AI assistants today force you into their ecosystem. ChatGPT lives in a browser tab. Claude lives in another. Siri lives on Apple devices. Each has its own interface, its own context, and its own limitations. None of them meet you where you already are — in your WhatsApp group chats, your Slack workspace, your Discord server, or your Signal conversations. If you want AI help, you have to context-switch to a separate application, copy-paste your question, and then copy-paste the answer back.

The product is the assistant, not the gateway. The gateway is just the control plane that makes the assistant feel omnipresent.

Beyond the fragmentation problem, cloud-hosted AI assistants raise privacy concerns. Your conversations, your data, and your queries all live on someone else’s servers. For personal use — managing your life, your work, your private conversations — this is an uncomfortable tradeoff.

The Solution

Moltbot is a self-hosted personal AI assistant built on a WebSocket-based gateway architecture. You install it on your own machine (macOS, Linux, or Windows via WSL2), pair it with your messaging accounts, and it handles the rest. The gateway acts as a control plane that routes messages between your channels and the AI agent, maintaining context across conversations.

The assistant supports Anthropic Claude (recommended: Claude Pro/Max with Opus 4.5) and OpenAI models, with automatic failover between providers. It is designed for a single user — this is your personal assistant, not a chatbot for your customers.

Key Features

14+ Messaging Channels

Moltbot connects to an impressive array of platforms, each with purpose-built integrations:

• WhatsApp (via Baileys) — Full DM and group support
• Telegram (via grammY) — Bot API integration
• Slack (via Bolt) — Workspace integration with thread support
• Discord (via discord.js) — Server and DM support
• Google Chat (via Chat API) — Workspace integration
• Signal (via signal-cli) — Privacy-focused messaging
• iMessage (via imsg) — Native Apple ecosystem
• Microsoft Teams — Enterprise communication
• Matrix, BlueBubbles, Zalo, WebChat — Extension channels

Voice and Speech

Voice Wake provides always-on speech detection on macOS, iOS, and Android. Talk Mode enables continuous conversation with ElevenLabs text-to-speech. The assistant can listen and respond naturally, making it feel like a real conversation partner.

Live Canvas

An agent-driven visual workspace called A2UI (Agent-to-UI) lets the assistant push interactive content to a canvas that you can see and interact with. This goes beyond text responses — the assistant can render visual information in real-time.

Browser Automation

Moltbot includes a dedicated Chrome/Chromium instance with CDP (Chrome DevTools Protocol) control. The assistant can browse the web, take snapshots, fill forms, and upload files on your behalf.

Security by Design

Since Moltbot connects to real messaging surfaces, inbound DMs are treated as untrusted input by default. The pairing system requires unknown senders to provide a code before the bot processes their messages. Public inbound DMs require explicit opt-in. The moltbot doctor command surfaces risky configurations.

Automation and Skills

• Cron jobs and wakeups — Schedule recurring tasks
• Webhooks — Trigger actions from external services
• Gmail Pub/Sub — React to incoming emails
• Skills platform — Bundled, managed, and workspace-level skills with install gating
• Camera and screen — Snap photos, record screen via iOS/Android/macOS nodes

Technology Stack

Architecture

Moltbot’s architecture centers on a single Gateway WebSocket control plane running on ws://127.0.0.1:18789. This gateway manages all sessions, channel connections, tool integrations, and event routing. Connected to the gateway are:

• The Pi agent runtime — Executes AI interactions in RPC mode with tool streaming and block streaming
• Channel connectors — Each messaging platform has its own connector that translates between platform-specific protocols and the gateway’s unified message format
• Companion apps — macOS menu bar app, iOS node, Android node
• WebChat UI — Browser-based chat interface served directly from the gateway
• CLI surface — moltbot gateway, moltbot agent, moltbot send, and utility commands

WhatsApp / Telegram / Slack / Discord / Signal / iMessage / Teams / ...
               |
               v
     +-------------------------+
     |        Gateway          |
     |    (control plane)      |
     |  ws://127.0.0.1:18789   |
     +------------+------------+
                  |
    +---+---------+--------+------+
    |   |         |        |      |
   Pi  CLI    WebChat   macOS   iOS/Android
  agent              app     nodes

The session model supports main sessions for direct chats and isolated group sessions with configurable activation modes. Multi-agent routing allows different channels or accounts to be routed to isolated agents with their own workspaces and sessions. For remote access, Moltbot auto-configures Tailscale Serve (tailnet-only) or Funnel (public) while keeping the gateway bound to loopback.

Getting Started

Installation is straightforward with Node.js 22+:

npm install -g moltbot@latest
moltbot onboard --install-daemon

The onboarding wizard walks through gateway setup, workspace configuration, channel pairing, and skill installation. The daemon runs as a launchd (macOS) or systemd (Linux) user service, keeping the gateway always running. Development follows stable/beta/dev channels with moltbot update --channel for channel switching.

Looking Ahead

Moltbot represents a different philosophy for AI assistants: local-first, channel-agnostic, and user-controlled. Instead of building another chat interface, it meets you where you already communicate. The growing ecosystem of channels, tools, and companion apps makes it a genuine operating system for personal AI — one that respects your privacy while delivering the convenience of having an intelligent assistant available everywhere you go.