OpenAI lets whole teams hand recurring work to ChatGPT, on a schedule or a trigger

OpenAI has published a video, “Put ChatGPT to work on shared team tasks”, on its channel. It shows recurring team work handed to ChatGPT. Coverage of the feature says Team Tasks can run on a schedule or when an event happens.

Team Tasks live inside the new ChatGPT Space

The video page carries only a title, and the text we could read gave no feature list, so the details below come from reporting on the underlying product, not from OpenAI’s own wording. Team Tasks sit inside ChatGPT Space, the shared workspace OpenAI introduced at its DevDay event on September 29, 2026. According to coverage from Android Authority and others, Space replaces the older Library for Pro, Business and Enterprise users. It is a home for editable Pages, files, presentations and spreadsheets that teammates can work on together in real time.

Team Tasks are aimed at recurring work, with weekly project updates as the usual example. Reports say a task can run at set times or react to an event, such as a new email or a Slack message, and that teammates can refine its instructions together. One report lists the feature for Business and Enterprise plans, not Pro, so the exact eligibility should be checked against OpenAI’s documentation.

The tasks are tied to dots, the always-on agents OpenAI announced alongside Space. VentureBeat reported that each dot runs on GPT-6 Astra, has its own cloud computer and browser, connects to more than 4,000 apps and can be reached from ChatGPT, Slack and Microsoft Teams. It also said the first dot is included at no extra charge for Pro and Business Premium customers, while Enterprise, Edu and Healthcare customers can try the beta once an administrator turns it on.

Existing sharing rules still apply. Coverage says workspace administrators’ permissions carry over, and teams can give view or edit access to a Page. Space works on desktop and web; on mobile, people can read and share Pages but not create or edit them.

How OpenAI is showing dots at work

A separate OpenAI post from October 16 gives a concrete picture. An OpenAI engineer, Ali, used his dot to triage an on-call incident at 3am. According to the post, the dot investigated the issue and started a Codex thread, so he could steer the work from his phone in the ChatGPT app. It is a single-user case, but it shows the pattern the team version extends: an agent that keeps working, reports back, and can be redirected by a person.

Usage limits are the open question. One report says Codex and ChatGPT Work tasks started or managed by a dot still count toward those products’ normal plan limits. Another says tasks assigned to dots on shared pages do not count toward ChatGPT usage limits. The two accounts do not match, and OpenAI’s own figures were not available to us.

Team Tasks follow a run of enterprise releases

The announcement fits a run of enterprise-focused releases from OpenAI in recent weeks. On October 8 the company said Ultrafast mode for GPT-6.1 Sol was rolling out in the API, Codex and ChatGPT Work, at up to 8x the speed of Sol Standard. A day later the same models were in the news through Microsoft, which put GPT-6.1 Sol into Foundry as Routines and Tool Search went generally available. OpenAI also opened a free GPT-Rosalind plan for university labs. Team Tasks add a third front: not a bigger model, but a place where a group’s recurring chores live.

The shift also exposes a reliability question. OpenAI’s status page logged a degraded-performance incident in Codex and ChatGPT Work on October 14, and an earlier one on October 8 with errors for some APAC users. A task that fires at 3am or on an incoming email only helps if the platform is up and the run is checkable. Reviewers of the feature have suggested teams decide who may edit a task and test the next run before leaving it unattended, which is a sensible rule for any agent that acts without a person watching.

Pricing is the other gap. VentureBeat said OpenAI has not disclosed the allowance for heavy work, the cost of extra dots or pricing for specialist dots now in enterprise pilots. Until that is public, teams cannot judge whether a shared task is cheaper than the hours it replaces.