Schema Design in the Age of AI: Pragmatic Patterns to Evolve and Ship Fast
summary
Shows how Apollo MCP server plugs a Graph OS operation collections workflow into LLM agents so new GraphQL operations become tools seconds after saving — no commits, no deployments, no restarts. Demonstrates with Roverdev + MCP flag and Claude calling tools like search for product titles and prices and get product reviews by product ID against a dataset of five products (three Air Jordans). Argues GraphQL’s deterministic schema acts as a tool catalog to cut token orchestration costs and speed iteration.
AI-written summary · about
chapters (10)
- 0:00AI application layer era
- 4:35AI fundamentally different
- 7:03MCP as client-server protocol
- 13:45Non-deterministic orchestration
- 17:13GraphQL fits AI tooling
- 20:43Apollo MCP server in action
- 29:50GraphQL for AI agents
- 32:16Schema design best practices
- 37:54Embracing chaos mitigating risk
- 41:24Key takeaways for AI
related talks
21mBuilt to Evolve: 13 Years of GraphQL
29mBuilding MCP Apps With GraphQL Patterns You Already Know
30mSchema Design Patterns: Leveraging Existing REST APIs for Rapid GraphQL
14mKeynote: GraphQL in the AI Era
01h 28mWorkshop: Demand-Driven Schema Design
23mMCP: Is GraphQL the Right “Tool” For the Job?
26mKeynote: Apollo's Journey with GraphQL: Transforming Enterprise APIs for the Future
26mAsk, Don't Fetch: Why GraphQL + MCP Is the Native Language of AI Agents