Class 7 taught ReAct. Today we go further — Reflection and Planning are two powerful patterns every agent builder must know.
Two LLMs collaborate — Generator and Reflector — to improve an answer in a single cycle.
Generator and Reflector iterate N times. Watch quality improve and discover diminishing returns.
Give the agent a task. Watch it decompose into a clear numbered plan before acting.
The plan is ready. Each step executes and gets ticked off one by one, then synthesized.
One task. Three agents. Three completely different ways of thinking. Compare outputs side by side.
Learn the anatomy of a LangGraph: nodes, edges, conditional edges, shared state, START and END.
All labs make real OpenAI API calls · Requires OPENAI_API_KEY in .env.local
OPENAI_API_KEY
.env.local