Run your own AI assistant gateway. Connect any model to WhatsApp, Telegram, Slack, Discord, and 30+ channels. Full agent tools. Total control.
AGDI is a complete runtime for running, routing, and controlling AI assistants across every surface you use.
Central dashboard to configure, monitor, and control your AI assistant. Real-time status, config editor, and agent management.
One assistant, 36+ channels. WhatsApp, Telegram, Discord, Slack, Teams, Signal, iMessage, Matrix, and more.
Use OpenAI, Anthropic, Gemini, Ollama, or any compatible provider. Switch models per-channel or per-agent. No vendor lock-in.
Built-in agent tools: browser control, bash, filesystem, canvas, camera, cron scheduling, and sub-agents.
Jarvis mode provides hands-free wake word detection and ultra-low latency STT/TTS streaming for true conversational AI.
Deploy specialized squads of sub-agents (Engineers, Analysts, Social Managers) that collaborate autonomously.
Your AI can generate, render, and refine interactive React/HTML artifacts directly in your dashboard or native apps.
Fully extensible. Build and distribute your own custom channels, translation layers, or agent tools via npm.
Four steps to a fully autonomous AI assistant running on your infrastructure.
One command setup. Configure your model provider keys and preferences via JSON or the web dashboard.
Add WhatsApp, Telegram, Discord, Slack — or any of the 36+ supported channels. Plug and play.
Activate agent tools, browser automation, code execution, and custom skills to extend your assistant's capabilities.
Manage everything from the web dashboard, native macOS/iOS/Android apps, or the CLI.
Reach your AI from the apps you already use. No new UIs to learn. Your assistant lives where your conversations happen.
WhatsApp
Telegram
Discord
Slack
Signal
Matrix
iMessage
Twitch
LINE
Mattermost
Google Chat
Zalo
NextcloudNo vendor lock-in. Route different channels to different models. Failover automatically. Run local or cloud.
OpenAI
Anthropic
Google Gemini
OllamaAGDI ships with real tools. Your assistant can browse, execute code, manage files, and automate tasks on your behalf.
Full Playwright-powered browser. Navigate, search, fill forms, take screenshots. Your AI sees the web.
Execute commands in a sandboxed shell. Run scripts, install tools, pipe outputs. Real system access, controlled.
Read, write, edit, and organize files. Workspace-scoped or full access. Your AI works with your files.
Generate HTML/React artifacts rendered in-browser. Build interactive prototypes and visual outputs.
Schedule recurring tasks. Daily reports, periodic checks, automated maintenance. Set it and forget it.
Spawn focused sub-tasks. Delegate research, code review, or analysis to specialized agent instances.
Control what your AI can access. Approve actions before they execute. Audit everything.
Control your assistant from the web, terminal, or native apps on macOS, iOS, and Android.
Full control panel. Config editor, chat, status monitoring, and agent management.
Power-user interface. Run commands, manage configs, and interact with your AI from the terminal.
Native Swift apps. Menu bar access on Mac, full mobile experience on iPhone.
Kotlin native app. Full gateway access, notifications, and chat from your Android device.
Pick your path. From npm to Docker to source — AGDI runs wherever Node.js runs.
curl -fsSL https://raw.githubusercontent.com/anassagd432/Agdi/main/install.sh | bash
irm https://raw.githubusercontent.com/anassagd432/Agdi/main/install.ps1 | iex
npx agdi@latest
docker run -d agdi/agdi:latest