Crafting Intelligent Agents with Context Engineering
summary
Defines context engineering as assembling the exact data you pass into an LLM and shows a working travel agent called Sorly travel bot built with TypeScript AI SDK that combines RAG + embeddings (HNSW, cosine similarity) with tool/function calling for flights, weather and FCDO lookups. Demonstrates memory (short/long term), summarization (one message shrinks from 27,024 tokens to ~13/12), pruning, and tool-selection via RAG MCP which cuts prompt tokens >50% and triples useful tool picks. Demo outputs structured JSON and persists itineraries.
AI-written summary · about
chapters (10)
- 0:00Context engineering intro
- 1:34Why context engineering matters
- 8:14Travel chatbot demo
- 13:48Crafting system prompts
- 26:04Retrieval augmented generation
- 35:59Memory strategies
- 45:07Tools and tool calling
- 53:33Structured output
- 56:20Future of agent design
- 57:03Q&A
related talks
31mBuild Intelligent AI Agents With Context Engineering
15mUnderstanding Context Engineering for Agentic Systems
31mDesigning AI Agents with the Model Context Protocol: From Answers to Actions
01h 30mHow to build AI Agents: A step by step guide to creating AI agents
46mFullstack AI: The Building Blocks for Production Ready AI-Powered Software
31mTame Your Agents: AGENTS.md and Context Engineering for Deterministic AI
31mBuilding the Context Engine AI Agents Need
29mThe AI Agent Memory Landscape