OpenAI Dots is a persistent assistant in ChatGPT whose background research uses read-only tools; separate permissions and checks govern follow-up actions. OpenAI announced Dots on September 29, 2026. An interview account published by TechCrunch on October 27 focused on how that boundary works in practice.
What OpenAI Dots is designed to do
OpenAI describes each Dot as an agent with its own cloud computer and browser, powered by GPT-6 Astra. Dots can work on ongoing tasks through connected apps; OpenAI says its plugin ecosystem reaches more than 4,000 apps. The company has cited examples such as research analysis, software fixes, sales proposals, and content production.
The persistent-assistant format is the point: a Dot can review permitted information in the background and suggest a useful next step. That research does not itself carry out the suggested action.
How Dots handles background research and actions
OpenAI says proactive research uses read-only tools on permitted connected sources. It cannot itself send messages, change content in connected apps, or control the browser or computer. Follow-up actions have separate rules: app permissions determine access, Custom Rules can allow, require approval for, or block actions, and Auto-review checks actions that could affect accounts or share information.
OpenAI says purchases made with a payment card already saved with a merchant require user approval. Transfers between financial accounts are handed back to the user. OpenAI also warns that Dots can make mistakes and advises reviewing consequential work.
What happens to connected data and local access
OpenAI says Business, Enterprise, and Edu workspace content is not used to improve its models by default. On personal plans, users control whether Dot conversations and work can be used for model improvement. OpenAI says it does not train directly on proactive-research threads or their notes; information brought from that research into an eligible task may be used depending on the user’s settings.
Connecting a personal computer is optional, and local-computer access starts turned off. OpenAI says users can revoke that access. Disconnecting an app does not erase information the Dot has already obtained, while resetting a Dot deletes its conversations, saved memories, and scheduled tasks.
How Dots relates to Codex in practice
TechCrunch reported that much of Dots’ functionality could also be achieved with Codex, but with more setup and hand-holding. OpenAI describes Dots as a persistent assistant with its own cloud computer; local-computer access is optional. Its Help Center says a Dot can create tasks in Codex cloud environments, provided the user creates the environment in Codex first. Local-computer work requires the ChatGPT desktop app on that computer.
Matt Maher’s personal workflow shows one way he uses Dots: voice-directed coding work and email alerts. Some tasks in his demonstration required manual corrections.
What one early calendar alert tells you
TechCrunch reported that OpenAI enterprise product lead Alexander Embiricos recalled receiving a calendar-conflict alert three minutes before an event during early development. It is one reported example, not a general measure of Dots’ alert timing.
How OpenAI describes the rollout
OpenAI says the first Dot is included at no additional charge in Pro and Business Premium plans. Enterprise, Edu, and Healthcare users can try a beta when a workspace administrator enables it. OpenAI’s Help Center describes Pro access as gradually rolling out to users aged 18 and older in markets excluding the European Economic Area, Switzerland, and the United Kingdom.
