# Build actions

Actions are the capabilities the agent runs in your product — defined in Agent Studio, implemented in your code.

Actions are what the agent can *do*: call your APIs, submit forms, navigate your UI, or render a component. Each action has two halves — you **define** it in Agent Studio (name, description, input schema) and **implement** it in your code (a handler registered under the action's key). The agent decides *when* to call an action from the user's request; your handler runs it and returns data or renders UI.

  
  
  

## Where to start

Read [What are Actions](/guides/text-driven-actions/) for the model, then [Connect actions](/guides/connecting-actions/) to wire a handler, then [Execute actions](/guides/executing-actions/) for the runtime.

## Related

- [Authoring actions](/user-guides/authoring-actions/): define actions in the Agent Studio dashboard.
- [Agentic UI](/guides/recipes/): render UI components from an action.