dotsgpt
September 29, 2026 · Launch

Dots arrive: OpenAI agents that keep working after you log off

Unveiled at DevDay, each Dot gets a private cloud computer, runs on GPT-6 Astra and hunts for work before you assign it.

Agents involved:Dots

OpenAI unveiled Dots at DevDay on September 29, 2026. A Dot is a persistent agent built on GPT-6 Astra that operates from a cloud computer with a web browser, terminal and files of its own. It carries on when your devices are off and reaches your own computer only with your permission.

Conversations happen in group chats inside ChatGPT, Microsoft Teams and Slack, or by voice, while iMessage and RCS are on a waitlist. OpenAI's catalog gives it more than 4,000 apps to work with, QuickBooks, GitHub, Gmail, Google Docs and Slack among them.

The feature OpenAI leads with is proactive research. Unprompted, a Dot scans your connected apps for work it could handle, such as an unpaid invoice that needs a reminder, and brings the suggestion to you for approval. In that mode its access is read-only, so it cannot send or change anything.

Guardrails come in three layers: built-in policies, Custom Rules you write, and an automatic review that flags actions touching customers or sharing data. Stored passwords live in a vault that the Dot draws on but cannot read, and an Activity View records everything it does.

Dots come with every ChatGPT Pro plan and the new Business Premium plan, while Enterprise, Edu and Healthcare can request a beta. Pro itself is unavailable in the UK, Switzerland and the European Economic Area.

Why it matters to you

ChatGPT Pro subscribers outside Europe can set up a Dot today. Write your Custom Rules and give it read-only tasks before you let it send anything.

Sources

More from the Wire