Building Agentic AI: A 5-Week Hands-On Bootcamp with the OpenAI Agents SDK
Details
Welcome to Session 2 of our 5-week bootcamp.
This session will focus on Pydantic fundamentals and structured outputs to help us define and validate the data our agents work with. Then we’ll explore dynamic instructions, allowing an agent’s behavior to adapt based on runtime information.
By the end of the session, you’ll have a stronger foundation for building effective and dependable agents.
What We'll Cover
Session 1: Agent Foundations — September 10
- Effective Prompting
- Understanding Tokens
- Models and Instructions
- Building Your First Agent
Session 2: Controlling Agent Behavior — September 17
- Pydantic Fundamentals
- Dynamic Instructions
- Structured Outputs
- Context and Runtime Data
Session 3: Memory and Tools — September 24
- Memory and Sessions
- Creating Tools
- Tool Calling
- Using Multiple Tools
- Tracing and Debugging
Session 4: Building More Capable Agents — October 1
- Single-Agent Capstone
- Model Context Protocol (MCP)
- Connecting Agents to MCP Tools
- Guardrails
Session 5: Multi-Agent Systems — October 8
- Multi-Agent Design Patterns
- Handoffs and Agents as Tools
- Multi-Agent Development
- Multi-Agent Capstone
Prerequisites: A basic understanding of Python is recommended. You do not need prior experience building AI agents or working with the OpenAI Agents SDK. Familiarity with Python classes, functions, type hints, and basic API usage will help you get the most out of the bootcamp.
You should also create and securely save an OpenAI API key that you'll use throughout the bootcamp to run examples, build agents, and interact with OpenAI models from your Python code. Make sure you can access your key before the session begins.
We hope you join us!
AGENDA
- 6:00 p.m. – Welcome
- 6:05 p.m. – Session 1 Review
- 6:10 p.m. – Introduction to Pydantic
- 7:20 p.m. – BREAK
- 7:25 p.m. – Dynamic Instructions
- 8:00 p.m. – BREAK
- 8:10 p.m. – Structured Output
- 8:45 p.m. – Questions/Discussion
- 9:00 p.m. – End