Orchestrating APIs for LLMs - Agentic Workflows for GraphQL
summary
Demonstrates turning an OpenAPI spec into a working GraphQL supergraph in minutes by generating schemas, connectors and tests, then composing and validating them with Apollo connectors and the rover supergraph compose workflow. The pipeline hot-reloads MCP tools via an MCP server, iteratively fixes issues (e.g., unwraps a results array), runs connector tests, and exposes operations as agent-callable tools. Example output is a ~3,600-line enterprise graph built from many REST endpoints and executed by a Claude-driven agent.
AI-written summary · about
chapters (12)
- 0:00Agentic workflows for GraphQL
- 1:32API orchestration challenges
- 4:00The dream query and GraphQL
- 5:47Apollo connectors for REST
- 6:09Demo setup and MCP tool
- 9:20Agent-driven connector dev
- 13:08Testing connectors iteratively
- 15:38Exposing connectors as MCP
- 17:48Hot reloading MCP tools
- 22:08Automated GraphQL schema gen
- 23:56Apollo MCP in 15 min
- 25:01Resources and closing
related talks
43mFaster Together: How I Use AI and MCP to Build Better Connectors
01h 23mWorkshop: Apollo Router & MCP: A Modern Agentic Development Approach
01h 30mBuilding for the agentic future
21mSimplifying MCP Tool Sprawl With GraphQL
47mSchema Design in the Age of AI: Pragmatic Patterns to Evolve and Ship Fast
19mBringing GraphQL to the Enterprise: When IBM met Apollo
29mKeeping Agents in Check: Apollo MCP Authentication
26mJust Ask: Building a Natural Language GraphQL Agent from the Graph You Already…