Overview
When you create your MCP add-on, you declare the ride booking category by choosingride-booking as the category option in the Alexa AI CLI. Everything else about building, submitting, and certifying your add-on is the same as any other MCP add-on.
Declaring the ride booking category gives your add-on two benefits:
- Smarter arbitration. The platform can route ride-related requests (for example, “I want a ride”) directly to your add-on without requiring the user to say your provider’s name. Additional smarter arbitration based on the customer’s account linking, preferences, and usage history is coming soon.
- Visual experiences on headed devices. When your tool responses contain the required data described below, Alexa can render rich cards (maps, ride options, checkout flows) on headed devices.
How visual rendering works
Your tool responses can be structured or unstructured, but we recommend using structured responses to aid Alexa’s comprehension and accuracy. Because Alexa needs to understand your response to extract the right data, use clear, descriptive field names for the required information described below. Key rules:- All image URLs must be publicly reachable HTTP or HTTPS URLs. The device fetches images directly, so URLs that require authentication or resolve to internal/private hosts may fail silently and the visual may not render.
- If a tool response is missing any of the required data for a given capability, Alexa may skip that visual. Your spoken or text response still reaches the customer normally.
- Optional fields may be omitted freely. The visual can still render without them.
- You don’t build or call anything extra. Alexa renders the visual automatically from the data you return in your tool response.
Worked example
Suppose yourget_ride_estimates tool returns the following (JSON or plain text both work):

