OpenClaw + Ollama Setup Guide: Build Your Self-Hosted AI Agent (Zero API Costs)

Imagine having an AI assistant that doesn't just chat—but actually does things for you. It can read your files, execute commands, send messages on your behalf, and integrate with your favorite messaging apps like WhatsApp and Telegram. All running locally on your machine, with zero API fees, and complete data privacy.

That's exactly what OpenClaw delivers. Unlike ChatGPT or other cloud AI tools that simply respond to prompts, OpenClaw is a full-fledged AI agent framework that can take real actions on your computer. Paired with Ollama for local LLM execution, you get a powerful automation system that works entirely offline.

In this guide, you'll learn how to build this system using FlyEnv—the fastest way to get OpenClaw and Ollama running without complex Docker setups or manual configuration.

What You'll Build

By the end of this guide, you'll have:

• A self-hosted AI agent gateway running on your machine
• Integration with messaging apps (WhatsApp, Telegram, Discord, iMessage, and more)
• Local LLM processing through Ollama—no API keys, no usage limits
• An agent that can read/write files, execute commands, and automate tasks
• Complete data privacy—your messages and files never leave your hardware

What is OpenClaw?

OpenClaw is an open-source, self-hosted gateway that connects your messaging apps to AI coding agents. Think of it as a bridge between where you communicate (WhatsApp, Telegram, etc.) and an AI that can actually perform tasks.

Key Capabilities

Feature	Description
Multi-Channel	Connect WhatsApp, Telegram, Discord, iMessage, Slack simultaneously
Tool Use	Read/write files, execute shell commands, make HTTP requests
Session Management	Persistent conversations with context across sessions
Memory	Agent remembers past interactions and preferences
Multi-Agent Routing	Route different tasks to specialized agents

OpenClaw vs ChatGPT

Capability	ChatGPT	OpenClaw + Ollama
Action Execution	Text only	Files, commands, messages
Messaging Integration	None	WhatsApp, Telegram, Discord, etc.
Data Privacy	Cloud processed	100% local
API Costs	Per-token fees	Zero (after setup)
Offline Use	No	Yes

Prerequisites

Before starting, ensure you have:

• FlyEnv installed (Download here)
• 8GB+ RAM (16GB recommended for larger models)
• ~10GB free disk space for model storage
• A messaging app account (WhatsApp, Telegram, or Discord)

Step 1: Install Node.js 24

OpenClaw requires Node.js 24 (or Node.js 22 LTS 22.16+ for compatibility).

1. Open FlyEnv and navigate to the Node.js section
2. Select Node.js 24 from the version list (recommended)
3. Click Install and wait for completion
4. Set Node.js 24 as your active version

FlyEnv Advantage: Unlike manual Node.js installation, FlyEnv automatically handles PATH configuration and version switching. No more command not found errors.

Step 2: Install Ollama

Ollama powers the AI brain of your OpenClaw agent.

1. In FlyEnv, open the Library/Tools section
2. Find Ollama in the available applications
3. Click Install—FlyEnv handles all dependencies automatically
4. Start the Ollama service

Step 3: Download an AI Model

Choose a model based on your hardware capabilities:

For 8-16GB RAM Systems

Model	Size	Best For
qwen2.5-coder:7b	~4GB	Code generation, technical tasks
llama3.2	~4GB	General purpose, reasoning
deepseek-r1:7b	~4GB	Step-by-step reasoning

For 16GB+ RAM or GPU Systems

Model	Size	Best For
qwen2.5-coder:32b	~20GB	Complex coding, large context
gpt-oss:20b	~12GB	Tool use, agent tasks
glm-4.7-flash	~25GB	Advanced reasoning

Download via FlyEnv

1. Navigate to Ollama Models in FlyEnv
2. Select your preferred model from the list
3. Click Download and wait for completion

Recommendation: Start with qwen2.5-coder:7b for testing. It's fast, capable, and works well on most systems.

Step 4: Install OpenClaw

Now install the gateway that connects everything:

1. In FlyEnv's Library, locate OpenClaw
2. Click Install and follow the prompts
3. Once installed, click Start to launch the gateway

OpenClaw will start a local gateway (default: http://127.0.0.1:18789/) and open the Control UI in your browser.

Step 5: Configure OpenClaw to Use Ollama

Initial Setup Wizard

When you first open OpenClaw, you'll see a security warning—read it carefully. Unlike simple chatbots, OpenClaw can:

• Read and write files on your computer
• Execute system commands
• Send messages on your behalf
• Make web requests

This power requires responsible configuration.

Configuration Steps

1. Choose Quick Start (recommended for beginners)
2. Select AI Provider: Choose Ollama or Local / Self-Hosted
3. Enter API Endpoint: http://localhost:11434
4. Select Your Model: Choose the model you downloaded (e.g., qwen2.5-coder:7b)
5. Complete Setup

Verify Configuration

Your configuration file is stored at ~/.openclaw/openclaw.json. A basic setup looks like:

{
   "models": {
      "providers": {
         "ollama": {
            "baseUrl": "http://127.0.0.1:11434",
            "api": "ollama",
            "models": [
               {
                  "id": "qwen3.5:0.8b",
                  "name": "qwen3.5:0.8b",
                  "api": "ollama",
                  "reasoning": false,
                  "input": [
                     "text"
                  ],
                  "cost": {
                     "input": 0,
                     "output": 0,
                     "cacheRead": 0,
                     "cacheWrite": 0
                  },
                  "contextWindow": 32000,
                  "maxTokens": 32000
               }
            ]
         }
      }
   }
}

Step 6: Connect Messaging Apps

Now connect your favorite messaging platforms to control the agent remotely:

Option A: WhatsApp

openclaw configure --section channels

Select WhatsApp and scan the QR code with your phone.

Option B: Telegram

1. Create a bot via @BotFather
2. Add the bot token to OpenClaw configuration
3. Start chatting with your bot

Option C: Discord

1. Create a Discord application
2. Add the bot token to OpenClaw
3. Invite the bot to your server

Security Tip: Start with allowFrom whitelist to limit who can interact with your agent:

{
  "channels": {
    "whatsapp": {
      "allowFrom": ["+15555550123"],
      "groups": {
        "*": { "requireMention": true }
      }
    }
  },
  "messages": {
    "groupChat": {
      "mentionPatterns": ["@openclaw"]
    }
  }
}

Step 7: Test Your AI Agent

Send your first command via your connected messaging app:

Example Commands

Command	What It Does
Read the file ~/Documents/notes.txt	Reads and summarizes the file
Create a Python script that calculates fibonacci numbers	Writes code to a file
List files in the current directory	Executes shell command
Send an email to [email protected] about the project update	Drafts and sends email

From Terminal

You can also interact via the built-in TUI (Terminal User Interface):

openclaw

Or use the web dashboard at http://127.0.0.1:18789/

![Agent Interaction ]

Advanced: Hybrid Cloud + Local Setup

For complex tasks, combine local and cloud models:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "ollama/qwen2.5-coder:32b",
        "thinking": "anthropic/claude-sonnet-4-20250514"
      }
    }
  }
}

• Local model: Handles file reads, simple edits, routine tasks (saves API costs)
• Cloud model: Handles debugging, architecture, complex reasoning (used sparingly)

This hybrid approach can reduce API bills from $20-50/day to $2-5/day.

Troubleshooting

Issue	Solution
Model loads slowly or crashes	Out of memory. Use a smaller quantization: qwen2.5-coder:7b-q4_K_M
Tool calls fail	Set "reasoning": false in config. Some local models struggle with tool format
Context window errors	Local models have smaller context. Set accurate contextWindow in config
Cannot connect to Ollama	Verify Ollama is running: curl http://localhost:11434/api/tags
Messages not received	Check channel configuration and permissions

Security Best Practices

⚠️ OpenClaw is intentionally powerful and broadly permissioned.

1. Run in isolated environment when possible
2. Use allowFrom whitelist to limit message sources
3. Review third-party skills before enabling
4. Require mentions in group chats to prevent accidental triggers
5. Regularly audit file system permissions

Video Walkthrough

Prefer watching? See Build a 100% Local AI Agent: OpenClaw + Ollama Setup via FlyEnv In 5 Minutes in action:

Frequently Asked Questions (FAQ)

Q: Is OpenClaw free to use after setup?

A: Yes. OpenClaw is open source (MIT license), and Ollama runs models locally. Once configured, there are no ongoing API fees or usage limits. Your only cost is the electricity to run your computer.

Q: Does my data leave my machine?

A: No. When using OpenClaw + Ollama, all processing happens locally. Your messages, files, and data never leave your hardware unless you explicitly enable cloud integrations.

Q: What hardware do I need?

A: Minimum 8GB RAM for smaller models (7B parameters). For comfortable usage:

• Basic: 8GB RAM + CPU (7B models)
• Recommended: 16GB RAM + GPU (13B models)
• Power user: 32GB+ RAM + high-end GPU (32B+ models)

Q: Can I use cloud models if needed?

A: Yes. OpenClaw supports hybrid setups. You can configure local models as default and switch to cloud providers (Anthropic, OpenAI, etc.) for specific tasks that require more capability.

Q: How does this compare to ChatGPT with plugins?

A: ChatGPT plugins run in OpenAI's cloud environment. OpenClaw runs on your hardware with direct file system access. It's faster for local operations, works offline, and maintains complete privacy. However, it requires more technical setup.

Q: Is this suitable for business/enterprise use?

A: OpenClaw is currently experimental software. While powerful, it lacks enterprise-grade safeguards. It's best suited for personal use, development environments, or carefully controlled scenarios where you understand and accept the security implications.

Next Steps

Now that you have a self-hosted AI agent running:

• Set Up Local Email Testing with Mailpit - Test email automation workflows
• Expose Localhost with Cloudflare Tunnel - Access your agent remotely from anywhere
• Build More AI Agents - Explore other AI tools in the FlyEnv ecosystem

---

Ready to build your personal AI agent? Download FlyEnv today and join the self-hosted AI revolution—zero API costs, complete privacy, unlimited possibilities.