Text Driven Actions

What are Actions?

AI Actions are custom functions that give your Foldspace agent the ability to take actions in your platform. Think of them as creating API endpoints for your AI, allowing it to connect directly to your backend services and third-party APIs.

This is the core mechanism that allows your agent to act, not just talk. Instead of only answering questions based on a knowledge base, it can perform tasks, fetch live data, and automate workflows that are unique to your platform. The primary benefit is giving users instant access to information and functionality, saving them from having to navigate complex menus and dashboards.

Types of Actions

You can empower your agent with two primary types of actions, depending on your use case.

Execute-Only Actions

Use these when you need the agent to perform a task or fetch data and then use that information to formulate a natural language response. This is perfect for GET requests or simple POST operations.

Example: A user asks, “What’s the status of my support ticket?” The action fetches the status, and the agent replies, “Your ticket is currently ‘In Progress’.”

Execute + Render Actions

Use this when you want to display a custom, interactive UI component directly in the conversation — a form, card, or confirmation the user can act on without leaving the chat.

Example: A user asks, “Show me the details for task #42.” The action fetches the task data and renders an editable card where the user can update the title or description and save their changes.

Next Steps

Ready to build? An action’s logic and its UI are configured separately. Dive into the implementation guides below.

• Execute actions — run logic and return data to the agent.
• UI components in chat — render an interactive component in the chat.