Agenda — AI Agents¶
Session length: 3 hours | Difficulty: Advanced | Prerequisites: OpenAI Tools, Python dataclasses, basic async
What you will build today¶
A ReAct agent that loops through Thought → Action → Observation cycles to answer multi-step research questions, with three different memory strategies.
Schedule¶
| Time | Topic | File |
|---|---|---|
| 0:00–0:20 | What agents are, when they're needed vs chains | 01-what-are-agents |
| 0:20–0:50 | ReAct loop: Thought, Action, Observation | 02-react-loop |
| 0:50–1:15 | Planning strategies: sequential, parallel, tree-of-thought | 03-planning-strategies |
| 1:15–1:45 | Tool-calling agents: building a real agent from scratch | 04-tool-calling-agents |
| 1:45–2:05 | Memory strategies: in-context, summarized, vector | 05-memory-strategies |
| 2:05–2:45 | Practice exercises | 06-practice-exercises |
| 2:45–3:00 | Interview questions review | 07-interview-questions |