Developing AI Agents with OpenAI AgentKit

Ship more than prompts – build AI coworkers (ship agents). This course will teach you how to design, build, and embed functional AI agents using OpenAI AgentKit.

What you’ll learn

Building an AI agent that actually ships inside a product takes much more than crafting clever prompts. In this course, Developing AI Agents with OpenAI AgentKit, you’ll learn to design, implement, and operate agents and apps on the OpenAI platform end-to-end.

First, you’ll explore the AgentKit ecosystem – agents, tools, sessions, workflows, connectors, and Evals – and how they connect through the Model Context Protocol (MCP) and the modern Responses API. Next, you’ll discover how to embed agents into real user experiences by designing visual workflows in Agent Builder. Then, you’ll wire them into your own applications with ChatKit, while also seeing how the Apps SDK lets you surface similar capabilities as apps that run inside ChatGPT. Finally, you’ll learn how to trace, debug, evaluate, and secure your agents with guardrails, performance optimization, and governance workflows so they remain safe and sustainable in production.

When you’re finished with this course, you’ll have the skills and patterns needed to take an idea from prototype to a functional, production-ready AI agent built on AgentKit.

Homepage