Monday, February 16, 2026

OpenClaw: Install and Set Up Your Own Local AI Agent

What Is OpenClaw?

OpenClaw is an open-source AI agent that runs locally on your system and connects to messaging platforms like Telegram, WhatsApp, Discord, and more. Unlike most bots that require complex configuration files, OpenClaw is conversation-first — you interact with it through natural language, making it accessible even if you're not a seasoned developer.

Created by Peter Steinberger and maintained on GitHub, OpenClaw gives you a personal AI assistant that can chat, search the web, create files, and automate everyday tasks — all from your own machine.

Why Run an AI Agent Locally?

Running an AI agent on your own hardware has some compelling advantages:

  • Privacy — your conversations and data stay on your machine.
  • Control — you choose which AI model to use (Gemini, Claude, OpenAI, and others).
  • Extensibility — add skills, tools, and integrations that fit your workflow.
  • No vendor lock-in — OpenClaw is fully open source.

Prerequisites

Before you begin, make sure you have:

  • Node.js 22 or newer — check with node --version. Download it from nodejs.org if needed.
  • A terminal you're comfortable using (Terminal on macOS, PowerShell on Windows, or any Linux shell).
  • A Telegram account (optional, but recommended for chatting with your bot from any device).

Installation

Installing OpenClaw is a single command. Open your terminal and run:

macOS / Linux:

curl -fsSL https://openclaw.ai/install.sh | bash

Windows (PowerShell):

iwr -useb https://openclaw.ai/install.ps1 | iex

The installer auto-detects your operating system, verifies compatibility, and sets everything up. No repository cloning or manual configuration is needed at this stage.

Running the Onboarding Wizard

Once installed, launch the interactive onboarding wizard:

openclaw onboard --install-daemon

This starts a terminal UI (TUI) that walks you through the entire setup. When prompted, select QuickStart — it configures everything with sensible defaults so you can get your bot running as fast as possible.

Selecting Your AI Model

During onboarding, OpenClaw will ask you to choose a model provider. Supported providers include Google (Gemini), Anthropic (Claude), OpenAI, and more. For example, if you choose Google:

  1. Select the Google model/auth provider.
  2. Choose Google Antigravity OAuth and sign in with your Google account.
  3. Pick a model such as gemini-3-flash (or any model your subscription supports).

The flow is similar regardless of which provider you pick — authenticate, then choose a model.

Connecting to Telegram

OpenClaw supports multiple messaging platforms, but Telegram is the easiest to get started with:

  1. Open Telegram and search for @BotFather.
  2. Start a chat and run the /newbot command.
  3. Follow the prompts to name your bot and copy the bot token.
  4. Paste the token into OpenClaw's onboarding prompt.

That's it — your bot is now linked to Telegram.

Skills and API Keys

The wizard will also ask about skills (specialized tools your bot can use) and API keys for third-party services. For a first setup, it's fine to select "Skip for now" for skills and decline the API key prompts. You can always add these later.

Checking the Gateway

After onboarding completes, verify that the OpenClaw Gateway is running:

openclaw gateway status

The Gateway is the background service that keeps your bot active. You can also run it in the foreground for debugging:

openclaw gateway --port 18789

Accessing the Dashboard

OpenClaw includes a browser-based Control UI for managing your bot visually:

openclaw dashboard

This opens a web dashboard at http://127.0.0.1:18789/ where you can view conversations, manage settings, and monitor your bot's activity.

Hatching Your Bot

When the setup is complete, it's time to bring your bot to life. Choose "Hatch in TUI" to configure your bot's personality right in the terminal. OpenClaw will ask a few simple questions — give your bot a name, tell it what to call you, and define its personality. Once configured, your bot will confirm its setup with a personalized response.

Chatting Through Telegram

With your bot hatched, head to your Telegram bot and send /start. OpenClaw will provide connection details — follow the prompts, and you're chatting. Ask it questions, give it tasks, or just have a conversation. The real power is that you can interact with your bot from your phone, anywhere, while it runs on your local machine.

Enabling Web Search

Out of the box, OpenClaw handles conversational tasks well, but for real-time information it needs web search access. To enable this:

  1. Ask your bot (via Telegram or the TUI) how to enable web search.
  2. It will guide you to get a Brave Web Search API key.
  3. Send the API key to your bot through chat.

Your bot can now search the web, gather live information, and even create files based on its research.

Useful Environment Variables

OpenClaw supports several environment variables for advanced configuration:

VariablePurpose
OPENCLAW_HOMESets the home directory for internal path resolution
OPENCLAW_STATE_DIROverrides the state directory
OPENCLAW_CONFIG_PATHOverrides the config file path

These are optional but useful if you want to customize where OpenClaw stores its data.

Security Considerations

Since OpenClaw runs locally and has direct system access, it's important to use it responsibly:

  • Run as a non-privileged user — don't give your bot root/admin access.
  • Keep files in a dedicated directory — limit what the bot can read and write.
  • Avoid public chats initially — start with private conversations until you're comfortable.
  • Be explicit with instructions — vague commands can lead to unintended actions.
  • Test on a disposable system — if you're experimenting, use a VM or sandbox.

OpenClaw isn't malicious — it simply does exactly what it's told. Treat it like running scripts on your system: powerful and useful, but worth being careful with.

What You'll Have When You're Done

After following this guide, you'll have:

  • ✅ A running OpenClaw Gateway
  • ✅ Authentication configured with your preferred AI model
  • ✅ Access to the Control UI dashboard
  • ✅ A connected Telegram bot (or other messaging channel)
  • ✅ Optional web search capabilities

Next Steps

Once your bot is up and running, there's plenty more to explore:

  • Pairing — configure DM safety and approval flows.
  • Channels — connect additional messaging platforms.
  • Advanced Setup — build from source and configure advanced workflows.
  • Skills & Tools — extend your bot's capabilities with custom skills.

OpenClaw makes running your own AI agent surprisingly approachable. Whether you're looking for a personal assistant, a research helper, or just a fun experiment, it's a great project to try out on a weekend afternoon.