How to Set Up OpenClaw: The Open-Source AI Personal Assistant

If you've been looking for a self-hosted AI assistant that actually fits into your existing workflow — your terminal, your messaging apps, your infrastructure — this OpenClaw setup guide is exactly what you need. By the end of this post, you'll have OpenClaw running on your machine, connected to Telegram, and ready to take commands.

What is OpenClaw?

OpenClaw is an open-source AI personal assistant CLI that runs on your own hardware and connects to the AI models and messaging platforms you already use. Unlike cloud-only assistants, OpenClaw puts you in control — your data stays on your infrastructure, your keys are your own, and you can extend it with custom skills and automations. Think of it as a programmable, self-hosted Jarvis for developers and operators.

Prerequisites

• Node.js 18 or higher — Check with node -v. Use nvm to upgrade if needed.
• npm — Comes with Node.js. Verify with npm -v.
• A messaging account — Telegram is the easiest to start with.
• An AI API key — OpenClaw supports OpenAI, Anthropic, GitHub Copilot models, and others.
• Linux or macOS — Windows users can use WSL2.

Installation

Install via npm

npm install -g openclaw

Verify the install:

openclaw --version

If you get command not found, add npm's global bin to your PATH:

npm config get prefix
export PATH="$(npm config get prefix)/bin:$PATH"

Initialize the workspace

openclaw init

This scaffolds your workspace at ~/.openclaw/workspace with default config and memory files.

Basic Configuration

nano ~/.openclaw/config.yaml

Set your AI model

model:
  provider: openai
  name: gpt-4o
  apiKey: sk-your-openai-api-key-here

Connecting to Telegram

Step 1: Create a Telegram Bot

1. Open Telegram and message @BotFather
2. Send /newbot and follow the prompts
3. Copy the bot token BotFather gives you

Step 2: Add Telegram plugin to config

plugins:
  telegram:
    enabled: true
    token: "7123456789:AAF_your_telegram_bot_token_here"
    allowedUsers:
      - your_telegram_username

Step 3: Start the gateway

openclaw gateway start

To run it persistently as a systemd service:

# /etc/systemd/system/openclaw.service
[Unit]
Description=OpenClaw Gateway
After=network.target

[Service]
ExecStart=/usr/bin/openclaw gateway start
Restart=always
User=youruser
Environment=NODE_ENV=production

[Install]
WantedBy=multi-user.target

sudo systemctl daemon-reload
sudo systemctl enable openclaw
sudo systemctl start openclaw

First Commands — Verifying It Works

Open Telegram, find your bot, and send it a message:

• Send Hello — should get an AI response
• Send /status — shows session info, model, and runtime

Or test directly from CLI:

openclaw chat "What's in my workspace directory?"

Check gateway status and logs:

openclaw gateway status
openclaw gateway logs --tail 50

Tips and Next Steps

1. Write a SOUL.md

Drop a SOUL.md in your workspace to give your assistant a personality, tone, and priorities. OpenClaw reads it at every session start — treat it like a version-controlled system prompt.

2. Add custom Skills

Skills are SKILL.md files that give the assistant specialized instructions for specific tasks — Notion integration, GitHub workflows, weather lookups, or any repeatable automation.

3. Use MEMORY.md for persistent context

OpenClaw starts fresh each session but reads your workspace files as memory. Keep a MEMORY.md with ongoing tasks, key facts, and important context.

4. Lock it down

Always set allowedUsers in your plugin config. Treat your bot token like a password — rotate it immediately if ever exposed.

Need Help With Enterprise Deployment?

Setting up OpenClaw for a team, integrating it with your existing infrastructure, or building custom skills and automations at scale? We can help.