01 What a Dot Is
A Dot is a single agent you create, name and make your own. It is an "extension of you": it learns your preferences, how you think, and what good looks like, then applies that to the work you assign. It runs continuously — pursuing your defined goals in the background with minimal oversight — and can juggle several tasks at once. OpenAI's vision is that over time teams of Dots work together on your behalf; today you start with a primary dot.
Each Dot has its own cloud computer and its own browser, so it can use tools and apps without touching your machine. It can also connect to other devices, and you can grant it permission to use your laptop, where it works directly alongside you. Through OpenAI's plugin ecosystem a Dot can connect to more than 4,000 apps. When it is not actively on a task, it does "proactive research" through read-only connections to the tools you have linked — for example, noticing that an invoice was never sent and drafting one for your approval.
02 How You Work With One
You message a Dot through ChatGPT, Slack or Microsoft Teams, and you can hop on a voice call to talk something through. You can open its cloud computer at any time to inspect what it is doing. The design assumes you are not in the loop for every step: you give it the goal, it does the work, and you review the result.
- Developer: the Dot watches customer feedback, scopes smaller fixes, builds and tests them, and hands back complete pull requests with attached videos.
- Scientist: it reruns analysis as new data lands and flags what needs review.
- Sales lead: it builds a proof of concept and updates the proposal as requirements shift.
- Content creator: it learns your voice and clips new material as it arrives.
03 Availability and Control
Dots rolled out to Pro and Business Premium users in eligible markets on September 29, with Enterprise able to try them once a workspace admin approves. One Dot is included with those plans; OpenAI says you can add more later. You create a Dot in the ChatGPT desktop app or browser, then reach it from the mobile app.
Control is the headline feature. Dots follow the same permissions and controls as ChatGPT and support custom rules around when they can act on their own versus when they require your approval. Every action is subject to review and approval gates, and you can open the Dot's computer to see the work in progress. OpenAI is already working with Microsoft to bring specialist Dots — with their own identity, IT-provisioned hardware, and deep integrations into company systems — into Microsoft Agent 365's security controls.
04 The Context: Why Now
Dots land at a specific moment. They are the same category as Meta's Muse, the avatar agent that hit the AI world days earlier, and both are part of a push to make agentic AI feel like a coworker rather than a chatbox. The branding is deliberately soft — little floating blobs, because Dots look like, well, dots.
05 The Bottom Line
Dots turn ChatGPT's model into a worker: you assign a goal, it runs in the background on its own cloud computer, and you review the output. It is the clearest signal yet that the frontier labs are moving from answering questions to doing the work. Right now it is a Pro/Business/Enterprise product with one Dot, gated by approval and a new permissions layer.
What to watch is whether the safety controls hold up as Dots get more autonomy — the cancellation of GPT-6.1 Astra that same week is the cautionary headline attached to every "always-on" claim OpenAI makes.
This article was drafted with AI assistance and reviewed by a human editor.