ChatGPT Work
OpenAI's cloud agent mode for multi-step tasks across files and apps
ChatGPT Work extends ChatGPT from answering questions to doing jobs. You describe an outcome, such as preparing a weekly status report, building a slide deck from a folder of notes, or reconciling figures between two spreadsheets, and the agent works through it step by step. It gathers context from connected tools, uses a browser and code when needed, and produces finished files. Tasks can run for a long time without you watching, and you can return to review the result.
Connections run through plugins for email, calendars, Slack, Microsoft Teams, Google Drive, SharePoint, CRMs, and project trackers. Scheduled tasks let it keep working in the background, for example checking a channel each morning and updating a document. It builds on the same technology as OpenAI’s coding agent, Codex, applied to general knowledge work. Administrators on business and enterprise plans can control which plugins and browsers it can use, pre-approve or block sensitive actions, and set usage limits.
Availability and limits depend on your ChatGPT plan and change as OpenAI rolls the feature out, so check the current plan page. As with any agent, the output can contain mistakes, and actions that send messages or change records deserve review before approval. Claude Cowork, Microsoft Copilot’s agents, and Google’s Gemini agents are the closest competitors. ChatGPT Work is most useful for people who already keep their documents and conversations in tools the agent can reach.