MCP-UI: Where Intent Becomes Interface
summary
Claim: AI agents threaten to flatten interfaces into black‑and‑white chat but 55% US and 62% UK developers prefer AI to search. Demonstration uses MCP (Model Context Protocol) with a Goose MCP client and a Chrome DevTools MCP server, showing how MCPUI overlays interactive components (HTML, iframe, React/web components). A concrete build renders a GitHub project calendar, offers click‑through workload analysis and buttons, and visibly works in the demo, with MCPUI still early stage and open source for contribution.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:48AI over traditional search
- 3:34Agentic interfaces concerns
- 3:55Model Context Protocol
- 6:22MCP standardizes AI agent communication
- 8:08MCP workflow and demo
- 8:51Live MCP with Chrome DevTools
- 9:52Introducing MCP-UI and components
- 10:34Agentic commerce with MCP-UI
- 12:43GitHub projects tracker
- 14:48Live GitHub projects calendar
- 16:57MCP-UI impact and future
- 17:39Q&A
related talks
33mFrontend after AI: The new UX
22mThe Critical Advantage With AI
27mMisplAIced: Good UI Needs Good UX
32mFigma deep dive: Agentic workflows and the MCP
25mAgentic User Interfaces in Copilot and Power Platform
50mKEYNOTE: The agentification of software has a UX problem
56mBeyond Chat: Bringing Models to the Canvas
19mFrom concept to confidence with Dscout AI