To understand how Clawdbot works, it helps to think of it not as a single application, but as a system with two distinct parts working together. These are the Gateway and the Control Interface.

Think of the Gateway as the engine of a car. It’s the powerful, core component that does all the heavy lifting behind the scenes. The Control Interface, on the other hand, is the dashboard. It’s the user-friendly web portal you use to tell the engine what to do, manage its settings, and connect it to other tools.

Together, these two components allow Clawdbot to be both powerful and easy to manage.

The Gateway is the heart of the Clawdbot system. It's a service that runs continuously on your hardware, waiting for instructions. It’s responsible for four primary jobs.

First, it handles message routing. When you give Clawdbot a command, the Gateway analyzes it to understand your intent. It then determines which tool or sequence of tools is needed to fulfill the request.

Next, it makes LLM inference calls. The Gateway is what communicates with the large language model you've chosen (like a model from OpenAI or Anthropic). It packages your request, sends it to the LLM, and processes the model's response.

It also takes care of credential management. To interact with different services like Google Calendar or Spotify, Clawdbot needs API keys and other credentials. The Gateway stores these securely and uses them when needed to authorize its actions.

Finally, the Gateway is responsible for tool execution. Once the LLM decides on a course of action, the Gateway runs the necessary code. This could mean calling an external API, running a local script, or modifying a file on your computer. It turns the AI's plan into a real-world action.

Because the Gateway handles these core functions, it requires certain permissions to operate effectively on your system. It needs access to the file system to read and write files, and network access to communicate with LLMs and other web services.

If the Gateway is the engine, the Control Interface is your command center. It’s a web-based platform that lets you manage your Clawdbot installation without having to touch configuration files directly.

Through this interface, you can add, remove, and configure tools. This is where you tell Clawdbot about the services you want it to connect with. For each tool, you provide the necessary API keys and credentials, which are then passed to the Gateway for secure storage.

The Control Interface is also where you manage your primary settings, such as selecting which LLM provider you want to use and inputting your main API key for the model. This makes it simple to switch between different models or update your credentials as needed.

Essentially, the Control Interface provides a user-friendly way to configure the Gateway. Any changes you make in the web UI are communicated to the core service, which then adjusts its behavior accordingly. This separation allows you to manage a complex system through a simple, intuitive dashboard.

Now, let's test your understanding of Clawdbot's architecture.

Understanding this two-part structure is the first step in mastering how Clawdbot operates and interacts with your digital world.