Course catalog
Drive OpenAI's building surfaces — chat, Canvas, custom GPTs, and the Codex agent.
OpenAI ships more than one place to build: a chat window, a side-by-side Canvas editor, packaged custom GPTs, a Codex agent in your terminal, and a Codex that runs tasks in its own cloud container. Each one is good at a different job, and picking wrong costs you an afternoon. This course walks the literal surfaces — the model picker labels, Canvas revisions, the custom GPT configuration view, the codex install and login, the sandbox and approval settings, config.toml, AGENTS.md, cloud tasks, and code review on a pull request. You finish by comparing terminal, IDE, cloud, and chat honestly, moving one task between them without re-explaining it, and deciding when a subscription seat beats paying per token.
Section 1
Choose the right current ChatGPT mode and entry point, edit durable artifacts in Canvas or blocks, set reasoning deliberately, and package reusable work when your workspace allows it.
Section 2
Use Codex in the terminal with an explicit working directory, project instructions, configuration, sandbox boundary, approval policy, and verification loop.
Eligible first-time members can get 7 free days · Cancel anytime
Get startedSection 3
Delegate bounded repository work to Codex cloud, prepare the environment, review evidence against the right baseline, and carry the result through a pull request.
Section 4
Choose among Chat, Work, Codex, IDE, terminal, cloud, Sites, plugins, and automations by task shape; hand off state and separate subscription seats from API metering.