OpenAI’s Dots are persistent agents with their own cloud computer and browser. They can work toward a goal across conversations using apps you connect, while you give feedback and check progress in ChatGPT, Slack, or Teams. OpenAI’s examples include tracking customer feedback, revising launch materials, updating research, and preparing interview content. Read OpenAI’s announcement.
How this differs from a chat reply
A normal chat usually returns an answer after a prompt. A Dot can retain a goal and work context, break work into steps, continue inside authorized apps, and bring drafts or proposed actions back for review. It runs on its own cloud computer, separate from your local machine unless you choose to connect that machine. It is not an unrestricted background script: app access and independent actions are governed by permissions and rules.
Start with a reviewable task
- Create a Dot in ChatGPT desktop or desktop web. State the goal, deadline, and deliverable.
- Connect only the apps it needs. Name the source material, data it must avoid, and steps that require your approval.
- Begin with a small task, such as summarizing unresolved items in a project channel with links to the original messages. Do not start by granting permission to send external messages or edit production data.
- Use Activity View to inspect progress and evidence. Check the result before assigning follow-up work.
Permissions and action boundaries
Custom Rules can allow actions, require approval, or block them. Actions that affect accounts or share information go through action review; sensitive actions such as changing a password remain with the user. OpenAI describes proactive research as read-only: it can inspect connected apps and prepare suggestions, but it cannot send messages, change app content, or control a browser or computer through that background research. A task you explicitly start can act under its specific permissions. Dots can still make mistakes, so consequential work needs review.
Availability and plan limits
OpenAI says the first Dot is included with Pro and Business Premium. Enterprise, Edu, and Healthcare workspaces can join a beta when an administrator enables it, with rollout limited to eligible markets. Plans include an allowance for deeper work and extended limits during the first month. Conversations with a Dot do not count toward ChatGPT usage limits, but tasks it runs in Codex or ChatGPT Work count against those products’ allowances.