Getting Started with OpenClaw: Your Personal AI Assistant Framework
A beginner-friendly guide to OpenClaw — learn what it is, core concepts, and how to install and configure your personal AI assistant.
Build an AI assistant that actually knows you — running on your own machine, connected to your tools, and always available.
What is OpenClaw?
OpenClaw is an open-source framework that turns large language models (like Claude or GPT) into a persistent, personal AI assistant. Unlike chatbots that forget everything after each conversation, OpenClaw gives your AI:
- Memory — It remembers your preferences, past conversations, and important context
- Tools — It can read files, browse the web, control your calendar, send messages, and more
- Always-on presence — Connect via Telegram, Discord, Signal, or other messaging platforms
- Local-first — Runs on your machine, keeping your data private
Think of it as building your own Jarvis — an AI that lives in your digital ecosystem and grows smarter over time.
Core Concepts
🏠 The Gateway
The Gateway is OpenClaw's brain — a local server that:
- Manages conversations with AI models
- Handles message routing across platforms
- Stores session history and memory
- Coordinates tools and skills
You start it once, and it runs in the background.
🤖 Agents
An agent is a configured AI personality with specific capabilities. You can have multiple agents:
- A main assistant for daily tasks
- A coding expert for development work
- A research agent for deep dives
Each agent can have its own model, skills, and workspace.
🔧 Skills
Skills are plug-and-play capabilities. Some examples:
- Weather — Check forecasts
- GitHub — Manage issues and PRs
- Apple Notes — Create and search notes
- Web Search — Search the internet
Skills are defined in simple Markdown files and can be shared with the community.
💬 Channels
Channels connect OpenClaw to messaging platforms:
- Telegram — Chat via a custom bot
- Discord — Join servers and DMs
- Signal — Private messaging
- iMessage — macOS native integration
You configure which platforms to use and who can message your AI.
🧠 Memory
OpenClaw supports persistent memory through:
- File-based memory — Markdown files in your workspace
- Vector databases — Semantic search over past conversations
- Auto-capture — Automatically saves important information
Your AI wakes up each session knowing who you are.
Quick Start: Installation
Prerequisites
- Node.js 20+ (check with
node --version) - An AI API key (Anthropic, OpenAI, or compatible)
- macOS, Linux, or Windows with WSL
Step 1: Install OpenClaw
npm install -g openclaw
Step 2: Run the Setup Wizard
openclaw configure
This interactive wizard will:
- Ask for your AI provider and API key
- Set up basic security (who can message your AI)
- Configure your first channel (Telegram recommended for beginners)
- Create your workspace directory
Step 3: Start the Gateway
openclaw gateway start
Your AI is now running! 🎉
Step 4: Connect via Telegram (Optional)
If you configured Telegram:
- Create a bot via @BotFather
- Copy the bot token to your config
- Message your bot — your AI responds!
Your First Conversation
Once connected, try these starter prompts:
"What can you help me with?"
"Remember that I prefer concise answers"
"Check the weather in Salt Lake City"
"What's on my calendar today?"
Your AI will learn your preferences over time.
Project Structure
After setup, your workspace looks like:
~/.openclaw/
├── openclaw.json # Main configuration
├── workspace/ # Your AI's home directory
│ ├── SOUL.md # AI personality & guidelines
│ ├── MEMORY.md # Long-term memory
│ ├── TOOLS.md # Local tool notes
│ └── memory/ # Daily conversation logs
└── sessions/ # Conversation history
The SOUL.md file is particularly important — it defines who your AI is and how it behaves.
Next Steps
- Customize your AI — Edit
SOUL.mdto shape its personality - Add skills — Browse ClawhHub for community skills
- Connect more channels — Add Discord, Signal, or other platforms
- Set up memory — Configure vector search for better recall
Resources
OpenClaw is open source and community-driven. Join us in building the future of personal AI assistants.
Log in to leave a comment.