On September 29, 2026, OpenAI announced Dots, a new kind of ChatGPT agent. Instead of answering one question and stopping, a Dot keeps working in the background, even after you close ChatGPT. Think of it as a digital assistant that has its own computer in the cloud.
Key takeaways
- Dots run in their own isolated virtual machine and can browse the web and use connected apps.
- They keep working on long tasks, like watching your inbox or researching a topic, while you’re away.
- They ask for your approval before sending messages or buying anything.
- At launch they’re for ChatGPT Pro and Business Premium subscribers in eligible markets.
What is a Dot?
Regular ChatGPT works like a conversation: you ask, it answers, and it stops. A Dot is “persistent.” You give it a job, and it keeps going. According to Tom’s Guide, each Dot runs inside its own isolated cloud computer, so it can open websites, use more than 4,000 connected apps through plugins, and finish tasks that take hours instead of seconds.
You set a Dot up in ChatGPT on the web or the desktop app. After that, you can talk to it in other places: by voice call, in Slack, in Microsoft Teams, and OpenAI says SMS is planned.
What can it actually do?
- Watch your email and flag what needs a reply.
- Run background research in a read-only mode, then report back.
- Investigate problems, like a bug in a website or app.
- Handle errands that involve several websites, with your sign-off at the end.
The important safety detail: Dots are designed to ask you before taking actions like sending a message or making a purchase.
Who can use Dots, and what does it cost?
At launch, Dots are available to ChatGPT Pro and Business Premium subscribers in “eligible markets,” plus Enterprise customers in beta when their administrator turns it on. Your first Dot is included with a Pro subscription, and Pro users get extended usage limits during the first month. OpenAI hasn’t listed every eligible country in the coverage we reviewed, so Canadian users should check inside their ChatGPT account.
Why it matters
Dots are part of a bigger shift in 2026: AI is moving from chatting to doing. Meta launched its own agent, Muse, last month, and finance apps like Robinhood are adding agents that can analyze markets and place trades with optional manual approval. For small businesses, an always-on assistant that watches the inbox or tracks competitors could save hours a week. For everyone, it raises the same question: how much access should an AI have?
Station take
This is the most useful kind of AI for busy people, and the riskiest. Start with read-only jobs like research and summaries, keep approvals turned on, and don’t connect accounts the task doesn’t need. Read our 6 rules for using AI agents safely before you hand one your inbox.
Sources: Tom’s Guide · TechCrunch · VentureBeat. This article was prepared with AI assistance and reviewed by Media Station. Disclosure.