Your MCP Server is Bad - Agent-Native Product Design for Better AI Outcomes
AI Engineer via YouTube
Get 20% off all career paths from fullstack to AI
Learn AI, Data Science & Business — Earn Certificates That Get You Hired
Overview
AI, Data Science & Cloud Certificates from Google, IBM & Meta — 40% Off
One plan covers every Professional Certificate on Coursera. 40% off Coursera Plus Annual.
Unlock All Certificates
Learn to build effective Model Context Protocol (MCP) servers through this 55-minute conference talk that critiques common implementation mistakes and provides actionable solutions. Discover why most MCP servers fail by simply wrapping REST APIs without considering agent-specific needs, leading to confused language models and poor performance. Explore the three pillars of agent-native product design—Discovery, Iteration, and Context—through insights gained from developing FastMCP, a leading framework for MCP server creation. Master the transition from endpoint-focused thinking to product-focused design, understanding how to curate context into small, highly effective surface areas that optimize AI outcomes. Gain practical knowledge for avoiding the common pitfall of auto-generating servers from OpenAPI specifications without proper consideration for agent interaction patterns.
Syllabus
Your MCP Server is Bad (and you should feel bad) - Jeremiah Lowin, Prefect
Taught by
AI Engineer