The short answer
ChatGPT Work is OpenAI’s agent for multi-step work across apps and files, launched in July 2026. Per The Next Web’s report, it is built on Codex technology, supports scheduled tasks and plugins for tools such as Slack, Teams, Drive, and SharePoint, and comes with admin controls on Enterprise and Edu. Prepare with data rules, permissions, and training.
What is ChatGPT Work?
According to The Next Web’s report on the launch, OpenAI introduced ChatGPT Work on July 10, 2026 alongside a new model release. It acts as an agent across your apps and files, and it is built on Codex technology. ChatGPT now has three modes, Chat, Work, and Codex, available on every plan including Free, with usage metered the way Codex is.
The same report says the Codex desktop app became the new ChatGPT desktop app, with the old app renamed ChatGPT Classic. Naming and packaging change quickly, so check OpenAI’s current documentation before you brief your organization.
What can ChatGPT Work do?
Based on that reporting, Work can carry out multi-step tasks across apps and files rather than only answering questions. Reported capabilities include Scheduled Tasks, a public beta of Sites, desktop Computer Use, and plugins for tools such as Slack, Teams, Drive, SharePoint, email, calendars, and CRMs.
Typical knowledge-work uses are the same as for any work agent: preparing briefings from several sources, extracting data from many documents into one table, drafting recurring reports, and organizing files. See our guide to ChatGPT Work scheduled tasks and plugins.
What admin controls exist?
The report says Enterprise and Edu plans have admin controls, a Compliance API, and spend controls. Because Work is metered like Codex, spend controls matter: agentic tasks can use far more than a chat message.
Before enabling Work broadly, ask OpenAI in writing about data handling, retention, audit access, and how plugin access is scoped. Do not rely on a news report for contractual facts.
How should you prepare your organization?
Prepare before you enable it for everyone.
- Data rules. Define what may enter Work and which accounts are approved.
- Plugins. Decide which are approved, who owns each, and what access they get.
- Permissions. Choose conservative defaults and document exceptions.
- Scheduled tasks. Require an owner, purpose, and review for each.
- Spend. Set caps and name someone to review usage weekly.
- Training. Teach delegation and review, not just features.
Use our AI acceptable use policy guide to turn these into policy.
How should you train people on it?
Train on tasks and review. Have each participant delegate one real, low-risk job, then audit the result against the sources. Ask: where did it assume something, what did it miss, and what would I have to fix before using this?
Then compare with the other major work agent so people understand the shared concepts. See ChatGPT Work vs Claude Cowork, and our ChatGPT training guide.
What do we know about ChatGPT Work, and what should you verify?
Reporting from The Next Web in July 2026 described ChatGPT Work as built on Codex technology, with scheduled tasks, plugins, a Sites beta, and Computer Use, and said the Codex desktop app became the new ChatGPT desktop app. That is a news report, not OpenAI documentation, and the product is new and changing.
Before you train anyone on it, verify with OpenAI’s own pages: which plans include it, what admins can control, which actions need approval, and how data is handled. Then run a small pilot on one recurring task, the same way you would for any agentic tool.
Search volume for "ChatGPT Work" and "chat gpt work" is high, so expect colleagues to ask. A one-page internal explainer with your approved status, data rules, and who to ask will prevent shadow use. See ChatGPT Work vs Claude Cowork.
How should you evaluate ChatGPT Work for your organization?
Use the same checklist as for any agent: what can it read, what can it change or send, where does work run, who reviews it, and what controls do administrators have? Document the answers, pilot on a low-risk task, and decide whether to approve, restrict, or wait. Re-check quarterly because the product is evolving.
Frequently asked questions
When did ChatGPT Work launch?
According to The Next Web, OpenAI launched ChatGPT Work on July 10, 2026.
Is ChatGPT Work the same as Codex?
No, but it is built on Codex technology according to the same report. Codex targets software work and Work targets knowledge work.
Is ChatGPT Work available on free plans?
The report says Chat, Work, and Codex are available on every plan including Free, with metered usage. Confirm current limits with OpenAI.
Is it safe for sensitive data?
That depends on your plan, configuration, and policy. Set data rules first and start with low-risk tasks.
What is ChatGPT Work?
According to a July 2026 report by The Next Web, it is a work-oriented ChatGPT experience built on Codex technology. Verify current details in OpenAI’s documentation.
Is ChatGPT Work the same as Codex?
Reporting says it is built on Codex technology, but they are presented as different surfaces. Confirm current naming and eligibility with OpenAI.
Sources and further reading
Product features, plans, and limits change often. This guide reflects the vendor documentation listed above as of November 2026. Check the vendor’s current documentation before you make purchasing or policy decisions. TechStack Consulting is an independent training firm and is not affiliated with Anthropic or OpenAI.