OpenAI has introduced “dots,” a new kind of always-on AI agent that it says works on a person’s behalf around the clock. According to the announcement, each dot runs on GPT-6 Astra and has its own cloud computer and browser.
It learns from feedback over time and can pursue a user’s goals continuously, including in the background.
Through a plugin ecosystem, dots can connect to more than 4,000 apps. Users can reach their dot through ChatGPT on desktop, web and mobile, through Slack and Teams, or by voice call, and the company says texting is coming soon. Dots carry context across all of these channels.
OpenAI describes dots as an extension of the user, taking on work in the way the user would do it and sometimes before being asked. It offers examples from inside the company, such as dots investigating a bug as soon as it appears in Slack and turning a new design into a working app. It also cites an early outside tester whose dot noticed an unsent invoice to a publication, prepared it, and sent it after his approval.
The announcement sets out five use cases: a developer whose dot turns customer feedback into tested fixes and pull requests with videos of the changes, a product lead whose dot revises launch materials when scope changes, a scientist whose dot reruns analyses as new data arrives, a sales lead whose dot updates proposals and builds a proof of concept as requirements shift, and a content creator whose dot drafts clips, show notes and social posts from interview transcripts.
The company puts considerable weight on control and safety. Each dot works on its own cloud computer, kept separate from the user’s own machine unless the user chooses to connect it. Dots can use saved passwords to sign in to supported websites without exposing them to the model. When a dot works proactively in the background, it is restricted to read-only tools, so it cannot send messages, change app content or control the browser or computer.
Users can set Custom Rules that allow, require approval for, or block specific actions, and can follow all activity, including background work, in an Activity View. An “auto-review” system checks actions that could affect accounts or share information, and certain sensitive tasks, such as changing a password, always stay with the user. Monitoring can pause or stop a dot’s work if it detects a safety concern. OpenAI also cautions that dots can still make mistakes and that consequential work should be reviewed.
On data use, OpenAI says it does not use content from ChatGPT Business, Enterprise or Edu workspaces to improve its models by default, and that personal-plan users can control whether their dots’ conversations and work are used for this. It says it does not train directly on proactive research or a dot’s notes to itself, though information from them may be used if it informs an eligible conversation or task, depending on the user’s settings.
For organisations, OpenAI is previewing “specialist dots,” which take on dedicated responsibilities with their own identity, credentials, and access to a company’s systems of record, including IT-provisioned hardware. They build on early internal testing across procurement, invoice processing, email marketing, customer support and commercial contracting. The company is starting with focused enterprise pilots, with its engineers working directly with customers to define each dot’s responsibilities, tools and approval process. It is also working with Microsoft to integrate specialist dots with the governance and security controls in Agent 365. Looking further ahead, it envisions teams of dots working together for a user.
Dots are rolling out now in ChatGPT to Pro and Business Premium users in eligible markets, with Enterprise, Edu and Healthcare users able to try a beta once their workspace admin enables it. A user’s first dot is included in the plan at no extra cost, with extended limits for deeper work during the first month after launch. Conversations with a dot do not count toward ChatGPT usage limits, but tasks it starts in Codex or ChatGPT Work do. OpenAI says users will later be able to add more dots and scale each one’s speed or monthly workload. Setup begins in the ChatGPT desktop app or a desktop browser, after which the dot can be messaged on mobile.
Everything above comes from the announcement text you pasted, so the capabilities, model name and availability are OpenAI’s own claims. I haven’t checked them independently, and the document doesn’t include pricing beyond the first dot being included in the plans. If you plan to publish this, it’s worth confirming the details against OpenAI’s official page and its safety materials. I can also shorten this into a news brief or turn it into headlines and slugs.
