Interacting with LLMs via GraphQL
summary
Built and released Gato GraphQL in 2023 and demonstrates replacing the JavaScript glue layer so one single GraphQL query can fetch a post, call OpenAI Chat completions to translate it, and store it back in the DB in one round trip. Shows concrete query patterns that chain operations using an export directive (RFC), and explains the need for powerful custom directives to compose prompts, call arbitrary LLM endpoints and extract JSON paths.
AI-written summary · about
chapters (13)
- 0:00GraphQL and LLMs intro
- 1:06Integrating AI with GraphQL
- 3:13Single-request GraphQL workflows
- 4:35Exposing AI in GraphQL
- 6:23Directives for data and AI
- 7:48Exporting GraphQL variables
- 9:30Server support for directives
- 9:50Unrestrained LLM via GraphQL
- 12:56Prompt construction directives
- 13:18Multiple LLM APIs with directives
- 15:30Navigating JSON with directives
- 16:33GraphQL server flexibility
- 16:55Conclusion
related talks
31mWhy AI needs GraphQL
31mGraphQL at Any Level of Your Stack
15mTraining and Using Generative AI Models with GraphQL
23mAn Introduction to GraphQL
55mGraphQL - APIs the New Way
29mAI, SQL, and GraphQL Walk into a Fertility Clinic… LLM-based Medical feature development
27mAI, SQL, and GraphQL Walk into a Fertility Clinic
48mUnlocking LLM Potential: From Exploration to Integration