GUIDE
OpenAI Dots for business
Dots are always-on agents from OpenAI. Each one runs on its own cloud computer, connects to thousands of apps, and keeps working on a goal while you are away.
Last updated October 2026. Dots launched at DevDay on September 29, 2026. Verify availability and capabilities before deciding.
What it does
A Dot runs on a separate cloud computer with a browser, connects to a large library of apps through plugins, and works toward a goal rather than a single request. You reach it through ChatGPT, Slack, or Teams. When you are not working with it, it looks for ways to help in the background. That background work runs read-only until you allow more.
Why a business uses it
It fits teams that already work inside ChatGPT and want an agent that reaches the rest of their tools without building a workflow first. The permission model is explicit, which matters when an agent can sign into real systems.
Where it needs care
The more access a dot has, the more it can do, and the more damage a mistake can cause. Sensitive actions such as changing passwords or deleting data should require approval.
- Decide up front which apps a dot can touch.
- Keep sending, spending, and deleting behind approval.
- Review what it did when it runs unattended.
- Be aware that availability and features are still rolling out.
How I deploy it
I set up the connected apps, define what each dot owns, set the approval boundaries, and check that the background work is useful rather than noisy. Where a dot cannot reach a system cleanly, I connect it through an API, MCP, or a small piece of custom code.
Want a Dot doing real work?
Tell me the goal and the systems involved. I will tell you what it should be allowed to do.