Building Production AI Agents

For engineers who have built LLM applications and RAG systems and now need to ship agents that work. An agent is a model plus a harness — the tools, permissions, memory, loop and checks around it — and this course teaches harness engineering: the agent loop, tool schema and error design, MCP and agent skills, persistent memory and context compaction, when a multi-agent design is worth its cost, handoffs, trajectory evaluation and execution-based verification, tool-layer security against indirect prompt injection, loop and cost budgets, and a capstone audit of an AI-built agent.

Lessons are AI-assisted and human-reviewed. Learn more.