Modern Day Mashups: AI Agents are Reviving the Programmable Web
summary
MCP (Model Context Protocol) claims to let AI agents stitch APIs together with little maintenance: over 4,000 MCPs exist now, under 500 from official vendors. An open-source agent Goose demonstrates a real mashup—Google Maps MCP + OpenStreetMap tiles + developer and computer-controller MCPs—to produce an interactive restaurant map in about 30 seconds. A GitHub MCP clones and configures repos, handles Docker and generates passwords, completing complex setup in under two minutes. SDKs include TypeScript, Python, Rust, Java, Kotlin, and C.
AI-written summary · about
chapters (17)
- 0:00Web mashups golden age
- 3:14Why mashups faded
- 4:20Introducing Model Context Protocol
- 6:25MCP’s open ecosystem growth
- 8:13Live demo modern mashups
- 9:58Figma to website automation
- 11:23AI project setup automation
- 13:29How MCP servers work
- 16:48Tool selection and orchestration
- 18:13Finding MCP servers
- 19:41MCP revives creative dev
- 20:23Getting started with MCP clients
- 21:08Q&A: AI vs traditional devs
- 23:18Q&A: MCP security
- 24:24Q&A: API limits and reliability
- 25:29Q&A: Uncommon agent use cases
- 26:54Q&A: Future of agentic web
related talks
23mMCP Mashups: How AI Agents are Reviving the Programmable Web
30mComposable AI: Building Next-Gen AI Agents with MCP
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
32mFrom Monolith to AI Agent Modernizing Java Systems with MCP
46mMCP and A2A Revolutionizing Interconnection and Collaboration of AI Agents
20mLetting AI Interface with your App with MCP
11mDon't let AI agents push your buttons - use webMCP instead!
33mThe Model Context Protocol: Connecting AI to Everything