Integrating OpenClaw with B.AI
From Zero to Your Private AI Agent: Deploy OpenClaw with B.AI in 15 Minutes
OpenClaw (formerly ClawdBot or Moltbot) is an open-source personal AI assistant. Unlike cloud-based SaaS tools, OpenClaw runs locally on your own machine, giving you full control over your data, workflows, and operating environment.
You can interact with OpenClaw through familiar messaging platforms such as Telegram, WhatsApp, Lark, and DingTalk to handle tasks like managing email, organizing calendars, writing code, and automating everyday workflows.
OpenClaw is more than a chatbot. It is a functional AI agent designed for real-world execution. It supports persistent memory, access to your local file system and the internet, and extensibility through installable skills.
Because it is open-source and self-hosted, OpenClaw has attracted a vibrant community of developers and tech enthusiasts. Users have already built creative use cases ranging from business automation to personal productivity, demonstrating the potential of a truly personal AI assistant.
This guide walks you through the full setup process, including installation, initialization, and connecting OpenClaw to the B.AI API. By the end, you will have your own private AI agent running locally.
Step 1: Get Your B.AI API Key
Before configuring OpenClaw, you need a B.AI API key.
- Log in to B.AI Chat
- Go to the API key management page
- Create or copy your
api_key
Keep this key safe. You will use it in the configuration step below.
Step 2: Prepare Your System
Before installation, make sure your system meets the following requirements.
| Requirement | Details |
|---|---|
| Node.js | Node.js 24 is recommended. Node.js 22.14+ is also supported. |
| Operating System | macOS, Linux, or Windows. Native Windows and WSL2 are both supported, but WSL2 is recommended for better compatibility. |
| Package Manager | npm is supported for standard installation. pnpm is commonly used when building from source. |
To check your Node.js version, open a terminal and run:
node --version
If Node.js is not installed, or your version is too low, please install or upgrade it from the official website:
Step 3: Install OpenClaw
OpenClaw supports multiple installation methods. For most users, the easiest option is to install the latest CLI globally.
Option 1: Install with npm
npm install -g openclaw@latest
Option 2: Use the Official Install Script
macOS / Linux
curl -fsSL https://openclaw.ai/install.sh | bash
Windows PowerShell
iwr -useb https://openclaw.ai/install.ps1 | iex
Choose either method above. After installation, continue to the onboarding step.
Step 4: Complete the Initialization Wizard
After installing OpenClaw, run the onboarding command:
openclaw onboard --install-daemon
This command starts the initialization wizard and installs the local Gateway service.
During onboarding, OpenClaw will guide you through several setup sections, such as:
- AI model provider setup
- Communication channels
- Skills and optional features
- Local background services
If you plan to configure B.AI manually, you can skip the default model provider setup during onboarding and continue with the manual configuration below.
Once onboarding is complete, OpenClaw will be ready for custom model configuration.
Step 5: Configure the B.AI Model
After onboarding, you need to manually add the B.AI model provider to the OpenClaw configuration file and set it as the default model.
There are two ways to do this:
- One-Click Script: Follow the official one-click tutorial if available
- Manual Configuration: Follow the steps below
Configuration note: The manual configuration below uses the current
b.aiprovider namespace andapi.b.aihost so it stays aligned with the latest B.AI branding and API endpoint.
5.1 Open the Configuration File
Open the configuration file:
vim ~/.openclaw/openclaw.json
OpenClaw reads this file at startup to load model providers and agent settings.
5.2 Add the B.AI Provider Configuration
Locate the "models" section and merge in the following configuration.
Replace {B.AI_API_KEY} with your actual API key.
{
"models": {
"mode": "merge",
"providers": {
"b.ai": {
"baseUrl": "https://api.b.ai/v1/",
"apiKey": "{B.AI_API_KEY}",
"api": "openai-completions",
"models": [
{
"id": "gpt-5.2",
"name": "gpt-5.2"
},
{
"id": "gpt-5-mini",
"name": "gpt-5-mini"
},
{
"id": "gpt-5-nano",
"name": "gpt-5-nano"
},
{
"id": "claude-opus-4.6",
"name": "claude-opus-4.6"
},
{
"id": "claude-sonnet-4.6",
"name": "claude-sonnet-4.6"
},
{
"id": "claude-haiku-4.5",
"name": "claude-haiku-4.5"
}
]
}
}
}
}
5.3 Set the Default Model
In the same openclaw.json file, set the default model using the current OpenClaw structure:
{
"agents": {
"defaults": {
"model": {
"primary": "b.ai/gpt-5-nano"
},
"models": {
"b.ai/gpt-5-nano": {
"alias": "gpt-5-nano"
}
}
}
}
}
This tells OpenClaw to use the current official b.ai provider namespace for B.AI by default. If you expose more than one B.AI model, add them under agents.defaults.models as aliases so commands like openclaw models status and openclaw models set ... can resolve them correctly.
5.4 Restart the Gateway
After saving the configuration file, restart the Gateway so the changes take effect:
openclaw gateway restart
If the CLI tells you the Gateway service is not installed yet, run this first:
openclaw gateway install
openclaw gateway restart
Then verify the service state:
openclaw gateway status
5.5 Test the Connection
Send a simple test message from your terminal:
openclaw agent --agent main --message "How are you doing today?"
If OpenClaw returns a valid response, the connection to B.AI is working correctly. If you instead see a billing or insufficient-balance error, the provider is already connected but your API key does not currently have enough balance to complete the request.
Step 6: Understand Gateway and Diagnostics
If you run into issues during setup, it is helpful to understand the Gateway and the built-in diagnostic commands.
What Is the Gateway?
The Gateway is the local service layer that powers OpenClaw’s runtime features. It manages model access, local services, and background processes.
Common Gateway commands:
| Action | Command |
|---|---|
| Install the Gateway | openclaw gateway install |
| Start the Gateway | openclaw gateway start |
| Stop the Gateway | openclaw gateway stop |
| Restart the Gateway | openclaw gateway restart |
| Uninstall the Gateway | openclaw gateway uninstall |
| Check Gateway Status | openclaw gateway status |
Diagnostic Commands
After onboarding and configuration, run the following command to check your environment:
openclaw doctor
You can also check the Gateway status directly:
openclaw gateway status
If everything is working properly, the Gateway should show a healthy or running status.
Step 7: Launch OpenClaw
Once setup is complete, you can interact with your AI agent using either the web dashboard or the terminal interface.
Option 1: Web Dashboard
Start the dashboard:
openclaw dashboard
Then open the following address in your browser:
From the dashboard, you can:
- Chat with your AI
- View conversation history
- Configure models
- Monitor system status
Option 2: Terminal UI (TUI)
Start the terminal interface:
openclaw tui
Useful TUI commands:
| Command | Description |
|---|---|
/status | View the current system status |
/session <key> | Switch to a specific chat session |
/model <name> | Switch the current LLM |
/help | View available commands |
Step 8: Learn Essential Commands
1. Check Model Status
openclaw models status
2. Manage Channels
openclaw channels list
3. Search Memory
openclaw memory search "keyword"
4. View Documentation
openclaw docs
Done
You now have a working private AI agent powered by OpenClaw + B.AI.
You can now:
- Build automation workflows
- Connect Telegram bots
- Extend capabilities with skills
- Create your own AI agent product
Welcome to your personal AI infrastructure powered by B.AI.