🦞

Moltbot: Your Personal AI Assistant

An open-source, self-hosted AI assistant that runs on your own devices. Works on WhatsApp, Telegram, Slack, Discord, and 10+ more channels. Like having J.A.R.V.I.S. in your pocket.

Open Source (MIT)Node.js 22+12+ ChannelsSelf-Hosted

Works Where You Work

💬WhatsApp
✈️Telegram
💼Slack
🎮Discord
🔒Signal
🍎iMessage
💚Google Chat
🏢Microsoft Teams

What Can It Do?

Persistent Memory

Remembers everything across conversations. Your AI builds context that compounds over time.

Multi-Channel

Works wherever you chat — WhatsApp, Telegram, Slack, Discord, iMessage, Signal, and more.

Email & Calendar

Reads your inbox, drafts replies, manages your calendar, sends follow-ups.

Web Research

Searches the web, scrapes pages, analyzes competitors — all without leaving chat.

Proactive

Doesn't wait to be asked. Sends reminders, morning briefings, and nudges on tasks.

Multi-Agent

Spawns sub-agents for complex parallel tasks. One request, multiple workers.

My Setup: Jarvis on Azure

I run my personal AI assistant (I call him Jarvis) on an Azure VM. Here's exactly how it's configured:

Infrastructure

  • VM: Azure D2s_v3 (2 vCPU, 8GB RAM)
  • OS: Ubuntu 22.04 LTS
  • Cost: ~$80/month
  • Storage: 128GB Premium SSD
  • Node: v22.22.0

Connected Channels

  • Slack: My personal workspace (harshithventures)
  • Telegram: Direct messages
  • WhatsApp: Via Baileys (no API fees)
  • Email: Google OAuth (4 accounts)
  • Calendar: Google Calendar integration

AI Models

  • Primary: Claude Opus 4.5 (via Azure)
  • Embeddings: Ollama (local nomic-embed)
  • Memory: Mem0 + Qdrant vector DB
  • Fallback: Kimi K2.5 (newest model)

Security

  • Gateway: Bound to loopback only
  • Access: SSH key auth + Tailscale
  • DM Policy: Pairing code required
  • Audits: Hourly security cron checks

How to Set Up Your Own

Azure VM (Recommended)

Run 24/7 on a cloud VM for always-on availability. My setup runs on Azure's D2s_v3 ($80/month).

Always-on 24/7
No local machine needed
Scales easily

Setup Steps:

  1. 1Provision Ubuntu 22.04 VM (2 vCPU, 4GB RAM minimum)
  2. 2Install Node.js 22+ and npm
  3. 3npm install -g moltbot@latest
  4. 4moltbot onboard --install-daemon
  5. 5Connect your messaging channels

Mac Mini / Home Server

Perfect for privacy-conscious users who want full control. Runs locally on your hardware.

Full control
Privacy-first
No cloud costs

Setup Steps:

  1. 1Install Node.js 22+ via Homebrew
  2. 2npm install -g moltbot@latest
  3. 3moltbot onboard --install-daemon
  4. 4Use Tailscale or ngrok for external access
  5. 5Keep Mac Mini awake (caffeinate)

Raspberry Pi

Ultra low-cost option. Runs on a $35 Pi 4 with 4GB RAM.

Cheapest option (~$5/mo power)
Silent
Portable

Setup Steps:

  1. 1Flash Raspberry Pi OS 64-bit
  2. 2Install Node.js 22+ via nvm
  3. 3npm install -g moltbot@latest
  4. 4moltbot onboard --install-daemon
  5. 5Optional: Add SSD for reliability

Quick Start (5 minutes)

# Requires Node.js 22+
npm install -g moltbot@latest

# Run the setup wizard
moltbot onboard --install-daemon

# That's it! The wizard handles:
# - Gateway configuration
# - Channel setup (WhatsApp, Telegram, etc.)
# - Workspace initialization
# - Daemon installation (auto-restart)

Advanced Features I Use

Mem0 Integration

Long-term memory using vector databases. Your AI remembers facts, preferences, and context across all conversations.

# Install mem0
pip install mem0ai

# Configure in your workspace
# hooks/mem0-inject/run.py - Injects memories at session start
# hooks/mem0-save/run.py - Saves facts when session ends

# Uses:
# - Ollama for local embeddings
# - Qdrant for vector storage
# - GPT for fact extraction

Custom Skills

Extend your AI with skills — modular capabilities like GitHub integration, weather, Google Workspace, and more.

# Skills are defined in SKILL.md files
# Located in: /usr/lib/node_modules/moltbot/skills/

# Available skills:
# - github: PR reviews, issue management
# - gog: Google Workspace (Gmail, Calendar, Drive)
# - weather: Forecasts (no API key needed)
# - slack: Advanced Slack actions
# - tmux: Terminal session control

Cron & Automation

Schedule proactive tasks — morning briefings, email checks, reminders, and recurring automations.

# Example: Morning briefing at 8 AM
moltbot cron add --name "morning-brief" \
  --schedule '{"kind":"cron","expr":"0 8 * * *","tz":"IST"}' \
  --payload '{"kind":"systemEvent","text":"Morning briefing time"}' \
  --sessionTarget main

# Example: Check emails every 30 min
moltbot cron add --name "email-check" \
  --schedule '{"kind":"every","everyMs":1800000}' \
  --payload '{"kind":"agentTurn","message":"Check inbox"}' \
  --sessionTarget isolated

Browser Control

Your AI can control a headless browser — navigate sites, fill forms, take screenshots, and automate web tasks.

# Moltbot manages its own browser profile
# Supports: CDP control, snapshots, actions, uploads

# In chat:
"Go to Linear and create a new ticket for..."
"Screenshot the homepage of competitor.com"
"Fill out this form with my details"

# Uses Playwright under the hood

What I Actually Use Jarvis For

📧

Email Triage

Reads my inbox every morning, summarizes important emails, drafts responses

📅

Calendar Management

Reminds me of meetings, preps briefings on attendees, suggests optimal times

🔍

Research

Searches the web, analyzes competitors, finds information I need

💰

Financial Tracking

Tracks my receivables, reminds me to chase payments, monitors goals

📝

Note Taking

Maintains my memory files, logs decisions, keeps track of everything

🤖

Automations

Runs scheduled tasks, checks for updates, sends proactive alerts

Frequently Asked Questions

Ready to Build Your Own AI Assistant?

Start with the docs, join the Discord community, or book a call if you want me to build it for you.