How Generative AI Apps Use APIs

on YouTube ↗24m

" summary

Argues that mastering APIs beats generic ML fluff and demonstrates four concrete API patterns: serving LLMs via standard SDKs (Anthropic-style), RAG with embeddings and vector stores pulling enterprise docs into prompts, agentic tool-calls via LangChain-like frameworks, and MCP servers. Shows a live fastMCP server with a get_counts tool that calls a public sample API for fantasy football, returns results in ~2 seconds, and formats JSON for the LLM. Notes billing is token-based — an example call uses 40 input tokens, 24 output tokens.

AI-written summary · about

" chapters (15)

1. 0:00Generative AI and APIs
2. 1:25Mastering APIs for AI success
3. 2:29APIs and generative AI overlap
4. 3:13Serving LLM models via APIs
5. 4:36RAG apps with APIs
6. 6:21AI agents and real-time APIs
7. 7:49Model Context Protocol
8. 8:52AI API skills
9. 11:21API to serve LLM
10. 12:44Billing for LLM APIs
11. 14:09Consuming LLM APIs
12. 15:12RAG with Copilot
13. 16:57Fast MCP demo
14. 21:17AI app API design
15. 22:42Q&A

" related talks

• 26mAPIs in the world of Generative AIClaudio Tag · apidays Singapore 2024
• 28mAPIs and GenAI : Your Perfect Combo with LLM-Powered Frameworks.Rafik Ferroukh · apidays Paris 2024
• 24mAgentic API - the only API you will ever needDan Erez · apidays Helsinki & North 2025
• 01h 10mPractical Applications of Generative AI: How to Sprinkle a Little AI in Your AppPhil Haack · NDC Porto 2023
• 26mAPIs in Agentic worldRonak Shah · apidays New York 2025
• 01h 10mPractical Applications of Generative AI: How to Sprinkle a Little AI in Your AppPhil Haack · NDC Sydney 2024
• 29mAPI Integration augmented by GenAIPierre-Olivier Patin · apidays Paris 2024
• 31mThe Impact of AI on APIs: Can AI replace the API Architect?Saheed Abiola Lasisi · apidays London 2024