Agentic UI
Agentic UI is how the agent operates your interface: the components it renders and drives inside the conversation, plus worked end-to-end use cases you can adapt.
Interaction modalities
Section titled “Interaction modalities”Users interact with the agent in more than one way, and an action can respond in whichever fits the moment — or a combination:
- Text — the default chat conversation.
- Voice — spoken interaction; configure it in Voice.
- Visual — interactive UI components (Chatterblocks) rendered in the chat; see UI components in chat.
- Combination — mix them, for example a spoken request that returns a visual component to act on.
Agentic UI components
Section titled “Agentic UI components”The building blocks the agent renders and drives inside your product.
UI components in chat Render forms, cards, and dashboards (Chatterblocks) inline in the conversation.
Shared State (Tandem) Keep the agent and your live page in sync so responses match what the user sees.
Shared State examples
Section titled “Shared State examples” React meeting form Two-way sync between the agent and a React form.
React filter panel The agent drives a filter panel as the user browses.
Backbone Shared State wired into a Backbone app.
Vanilla JS Shared State with no framework.
Use cases
Section titled “Use cases”End-to-end walkthroughs that put the pieces together.
Analyze report data Run a task agent over report data and render the result inline.
Embed the agent Run the agent as an embedded, always-on surface in your app.