Global Tech News For people figuring out what to build with AI.
AI

OpenAI Launches Dots, Agents That Keep Working Between Chats

A woman works at a laptop in a home office.

OpenAI began rolling out Dots on Sept. 29, a ChatGPT feature that lets users set up AI agents to keep working on tasks between conversations. Each Dot uses its own cloud computer and browser, connects to apps a user selects, and runs on GPT-6 Astra, the company says. Access starts with Pro and Business Premium users in eligible markets; Enterprise customers can try a beta when their workspace administrator enables it.

Users set up a Dot on desktop, give it a goal and context, then check its progress. At launch, they cannot create one on mobile, although an existing Dot can be reached from the mobile app where access is available. Users can also reach a Dot in Slack or Teams. OpenAI says its plugin ecosystem connects Dots to more than 4,000 apps, though supported apps can vary.

The main change is continuity: a Dot can retain context, run scheduled tasks and review connected information in the background rather than waiting for another prompt. OpenAI calls that background work “proactive research.” It says those tools are read-only, so they cannot send messages or change app content. For tasks a user assigns, custom rules can let the Dot act, require prior approval, ask before an action or hand the decision back to the user.

Each Dot’s cloud computer is separate from a user’s own computer, and local access starts turned off. If a user connects a computer, the Dot can work with local files and tools there; access can be revoked. OpenAI also says Business, Enterprise and Edu workspace content is not used to improve its models by default, while personal-plan users have separate data controls. Disconnecting an app does not erase information the Dot has already learned from it.

OpenAI’s launch examples come from its own teams and an early tester; the release includes no independent evaluation of how reliably Dots complete those jobs. The Help Center warns that Dots can make mistakes and says users should review consequential work.

Sources