- Announcement
- Coverage
- Global
- Primary source
- AWS Machine Learning Blog ↗
What changed?
The pattern combines an MCP server, AgentCore Runtime and AgentCore Gateway so an AI host can display useful controls and visual results inside a conversation. AWS illustrates the approach with a rental application that can show inventory, create bookings and return structured confirmations.
The bigger signal is that conversational products are moving beyond plain text. A portable MCP interface can let one service deliver the same tools and widgets across compatible AI hosts, while the application keeps its business logic and data layer behind a thin protocol adapter.
What should you keep in mind?
This is a technical walkthrough rather than a universal compatibility guarantee. Teams still need to review authentication, authorization, session isolation, accessibility and host-specific support before exposing real customer actions.
Go to the source
Source announcement: . This is an original summary, not independent on-the-ground reporting.
Last updated: . Editorial policy & corrections