OpenAI starts always-on ChatGPT helpers that work between chats
Reviewed by Helen Jones
On 29 September 2026 OpenAI documented Dots, always-on agents that take a goal and keep working after you leave the chat. A Dot can use connected apps on its own computer, continue between conversations, bring results back for review, and ping you when a decision needs a person. TechCrunch said the product started that day for ChatGPT Pro and Business Premium users in eligible markets. WIRED put initial consumer access on the $100-a-month Pro plan, with messages through ChatGPT, Slack, and Microsoft Teams, plus a waitlist to text Dots on iMessage or Android RCS. Official notes also list ChatGPT in Slack and Teams, Team Tasks that run on a schedule or a trigger through a team service account, workspace connections that follow connected-account permissions, and Agent Security controls for tools, files, and network access. Rollout is gradual and limited to eligible accounts.
Until now, most workplace AI still meant a blank chat. You restated context, waited for a draft, then did the checking yourself. Background work existed as a demo or a one-off, not as a named helper with a handoff when judgement was required. Dots change the default from a single reply to ongoing responsibility. That is useful only if the goal, allowed apps, and review step are written down first. Without those, an always-on agent will keep producing work you still have to rescue, only now it can also touch more tools while you are in another meeting.
Analysis
This is a trend to watch and a trap if you grant access before you define the job. Take one recurring task you already know well and write a one-page brief: the goal, which apps it may use, what it can do alone, and what must wait for a human review step. Do not turn a Dot on until that brief exists.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "OpenAI starts always-on ChatGPT helpers that work between chats", Collab365 Spaces. 3 sources referenced.