Run Clawdbot Locally With Ollama — Complete Guide [2026]

What You Are Setting Up (Quick Context)

You are installing Clawdbot, an open-source AI assistant framework, and connecting it to Ollama so it runs locally using your own models.

This setup:

• Runs fully on your machine
• Does not require a Mac Mini or cloud AI subscription
• Uses Ollama models like GLM 4.7
• Gives you a web dashboard, sessions, logs, and automation features

Prerequisites (Do This First)

Before starting, make sure you have:

1. A computer running Windows, macOS, or Linux
2. Node.js version 22 or higher
3. npm installed
4. Terminal or Command Prompt access

Step 1: Install Ollama

1. Open your browser
2. Go to ollama.com
3. Download Ollama for your operating system
4. Install it normally
5. After installation, keep Ollama running in the background

To confirm Ollama is installed, open a terminal and run:

ollama --version

If you see a version number, you are good.

Step 2: Install Clawdbot Globally

1. Open Command Prompt / Terminal
2. Run this command:

npm install -g Clawdbot@latest

1. Wait for installation to finish

If you see npm warnings about updates, ignore them.

If you get errors

• Update Node.js to version 22 or higher
• Restart the terminal
• Run the command again

Step 3: Run Clawdbot Onboarding

This creates the base configuration.

1. In the terminal, run:

Clawdbot onboard

1. When prompted:
   • Select Yes for security warning
   • Choose Quick Start
2. Model provider:
   • Select Skip for now
3. Default model:
   • Keep the default (Cloud Opus 4.5)
4. Channel selection:
   • Select Skip for now
5. Skills:
   • Select No
6. Hooks:
   • Select Skip
7. Gateway service:
   • Allow it to start

At the end, Clawdbot is installed and configured locally.

You may see a browser open with an error. This is expected at this stage.

Step 4: Pull an Ollama Model

Clawdbot needs a local model from Ollama.

Recommended models from the raw content:

• GLM 4.7
• Qwen Coder
• GPT-OSS 20B or 120B

Example using GLM 4.7:

ollama pull glm4.7

Wait for the download to complete.

Step 5: Launch Clawdbot with Ollama

Now connect Ollama to Clawdbot.

1. Open a new terminal window
2. Run:

ollama run glm4.7

1. In another terminal window, run:

Clawdbot launch

1. When prompted:
   • Select GLM 4.7
   • Confirm with Y

Clawdbot now registers the Ollama model.

Step 6: Verify Configuration File

Clawdbot updates its config automatically.

Check this file:

C:\\Users\\<your-username>\\Clawdbot\\Clawdbot.json

Inside, you should see:

• The Ollama provider
• GLM 4.7 listed as an available model

This confirms successful integration.

Step 7: Start the Clawdbot Gateway

The gateway runs the web dashboard.

Check status:

Clawdbot gateway status

If not running, start it manually:

Clawdbot gateway --port 18789 --verbose

Step 8: Open the Clawdbot Web Interface

1. Open your browser
2. Go to:

<http://localhost:18789>

You should see:

• Clawdbot dashboard
• Health status showing OK
• Left sidebar with Overview, Sessions, Channels, Skills, Cron Jobs

Step 9: Start Your First Session

1. Click New Session
2. Type:

hi

Clawdbot will respond with an onboarding message asking:

• Who you are
• What it should be called

This confirms everything is working.

Step 10: Understanding the Interface

Sessions

• Every chat runs as a session
• Sessions keep memory and logs

Channels

• Optional integrations for Telegram, WhatsApp, Discord
• Can be added later

Cron Jobs

• Schedule tasks like:
  • Email summaries
  • Daily checks
  • Automated workflows

Skills

• Extend Clawdbot abilities
• Install later when needed

Logs

• See everything Clawdbot executes
• One terminal shows all activity

Important Security Notes (Do Not Skip)

Clawdbot:

• Executes commands
• Reads files
• Browses the web
• Acts autonomously

Best practices:

• Do not connect personal email
• Do not give password vault access
• Use separate accounts
• Prefer read-only permissions
• Treat it like a virtual assistant with system access

Running locally or on a VPS limits damage if something goes wrong.

Finally

You now have:

• Clawdbot running locally
• Ollama powering the AI
• No cloud dependency
• Full control and visibility
• Zero cost setup

If you want next guides, like:

• Telegram integration
• Cron automations
• Skills setup
• VPS deployment

Say the word and I’ll do it properly, step by step.