OpenAI dots: always-on agents are here
AI agents with their own computer that work in the background on your behalf are rolling out gradually.
On September 29, 2026, at DevDay in San Francisco, OpenAI introduced dots: "always-on" agents with their own cloud computer that work for the user around the clock. They run on the GPT-6 Astra model.
The idea: instead of asking ChatGPT a question, you hand a dot a task, and it keeps going in the background while you are away.
What do they offer?
- Their own computer and browser: each dot has a separate cloud computer and browser.
- How you reach them: through ChatGPT, Slack and Microsoft Teams; you can ask questions, discuss ideas, give feedback and even talk by voice.
- More than 4,000 apps: connected through OpenAI's plugin ecosystem.
- Learning: they learn from your feedback and can work toward your goals around the clock.
- Proactive research: even when you have not given a task, they research through read-only connections; examples include investigating a bug, building an app from a design and noticing a missing invoice.
Who can use them?
The rollout is gradual and limited to eligible markets:
| Plan | Access |
|---|---|
| ChatGPT Pro | One dot included |
| ChatGPT Business Premium | One dot included |
| Enterprise | If a workspace admin approves (beta) |
| Additional dots | Pricing not stated |
You create dots in the ChatGPT desktop app or the browser; once created, you can also talk to them from the mobile app. We could not find a clear statement on whether Türkiye is among the eligible markets.
Are you in control?
Dots work with the same permissions as ChatGPT. You can set custom rules that decide which tasks they may do on their own and which need your approval.
What it means for a small business
For now, not much directly: access is gradual and limited to higher tiers. But the direction matters: AI is moving from "ask and get an answer" to "hand over a task and let it run in the background". An assistant that follows orders, email, the calendar and reports could become one of the places where small teams save the most time.
One thing to watch
Connecting your accounts to an always-on agent means giving it broad permissions. Limit up front what it can reach and which actions it can take without your approval, and review what it does regularly during the first weeks.
Source: 9to5Google
- #OpenAI
- #agents
- #ChatGPT
Wondering what this means for your business?
Let's talk it through — free, no obligation.
Comments0
No comments yet — start the conversation.
Weekly notebook
The week's notes, in your inbox
AI, software and digital marketing news worth your time — once a week, no spam.
