Course contents
What Is ChatGPT Work
OpenAI's agent mode: give it a goal, and it returns a finished spreadsheet, deck, document, or web app — not a chat reply.
In July 2026 OpenAI flipped ChatGPT from an app that answers into an app that works. ChatGPT Work is a toggle at the top of the chat — web, desktop, and mobile — that hands your request to an agent running on a cloud computer. It accepts a goal, connects to your apps and files, breaks the job into steps, runs independently for minutes or hours, and comes back with a finished artifact: a completed spreadsheet, a slide deck, a formatted document, or a working web app.
The lineage matters for your mental model. OpenAI first built Codex, its agentic tool for programmers, then merged it with ChatGPT — so Work is best understood as Codex made accessible for everyone. It's not a new subscription: Work is bundled into Plus, Pro, Business, and Enterprise plans, metered by usage the way Codex is — longer tasks consume more of your included allowance.
One expectation to reset on day one: Work tasks take minutes, not seconds. A research deck might run for thirteen minutes while the agent searches dozens of sites and assembles slides. That's not lag — that's the difference between an answer and finished work. This course teaches you to fill those minutes with parallel tasks instead of watching the spinner.
Riley's full walkthrough of all 14 Work capabilities — the single best companion to this course. Watch in chunks as you go.