OpenAI's dots run agents on their own computer

OpenAI introduced dots, always-on personal agents that run on their own cloud computer rather than inside a chat window. Each dot has its own browser and stays available continuously, and it can also be granted access to the user’s own laptop. OpenAI says the plugin ecosystem connects a dot to more than 4,000 apps.

The channels are ChatGPT on desktop, web and mobile, plus Slack and Teams, with texting described as coming soon. Rollout starts with Pro and Business Premium in eligible markets, while Enterprise, Edu and Healthcare get an admin-enabled beta. The first dot is included at no extra cost, with extended deeper-work limits for the first month.

How the isolation works

The design point that matters for anyone worried about handing a model a machine is that the dot’s cloud computer is separate from the user’s own. OpenAI describes saved passwords that can be used for sign-in without being exposed to the model, and background “proactive research” restricted to read-only tools. Monitoring can pause or stop work, and an auto-review step gates actions against the user’s instructions, custom rules and safety requirements.

Dots can also be given separate identities, their own credentials, IT-provisioned hardware and systems-of-record integrations. Those “specialist dots” are described as focused enterprise pilots rather than a general release, and Microsoft Agent 365 governance integration is in progress.

The credit detail is unusually specific: dot conversations do not count against ChatGPT usage limits, but work started in Codex or ChatGPT Work does.

OpenAI’s own caveat is the one to keep in view: “Dots can still make mistakes, so always review consequential work.” Sensitive tasks such as changing a password always stay with the user. Workspace content is not used to improve models by default on Business, Enterprise and Edu, and proactive research and dot notes are never trained on directly.

(Original source: OpenAI)