OpenAI Dots are persistent AI agents inside ChatGPT that can keep working on an ongoing responsibility instead of ending when one conversation ends.
OpenAI introduced Dots at DevDay on 29 September 2026. Each Dot is powered by GPT 6 Astra, has its own cloud computer and can use the apps a user chooses to connect.
How is a Dot different from normal ChatGPT?
A normal ChatGPT conversation mainly responds when you send a message. A Dot is designed to keep responsibility for a goal over time.
OpenAI says you can give a Dot an outcome, define what it is allowed to do and let it continue making progress between conversations. It can decide what step comes next and return to you when it needs judgment or approval.
What can a Dot do?
The useful part is not one specific feature. It is the combination of a persistent computer, connected tools and memory of the responsibility you gave it.
Examples include:
- keeping a project moving and following up on unfinished tasks;
- researching something over time and bringing back changes;
- working across connected apps instead of staying inside one chat window;
- preparing drafts or updates and asking for review before an important action;
- handling recurring work where the goal stays the same but the information changes.
Each Dot gets its own cloud computer
OpenAI says every Dot has a dedicated cloud computer and browser. That gives the agent a workspace where it can navigate websites and software instead of only generating text.
This is one of the biggest differences between a persistent agent and a normal chatbot. The agent has somewhere to continue work even when the user is not actively typing in the same conversation.
Which model powers Dots?
Dots are powered by GPT 6 Astra. OpenAI describes Astra as a model built for complex multi step work, coding, research and computer use.
The model matters, but the agent layer matters just as much. A capable model without access to tools, memory and a working environment still cannot complete many real tasks on its own.
What apps can a Dot use?
A Dot can work with apps that the user chooses to connect. OpenAI has shown Dots working across ChatGPT and workplace tools, with integrations such as Slack and Microsoft Teams forming part of the broader rollout.
The exact apps available depend on the user’s plan, connected services and current rollout. Access can also change as OpenAI adds more integrations.
Does a Dot act without permission?
Users define what a Dot is allowed to do and which connected apps it can access. OpenAI says the agent should return to the user for decisions that need human judgment.
That does not remove the need to review permissions carefully. A persistent agent can potentially see or act on more information than a single chat, depending on what the user connects.
Which ChatGPT plans have Dots?
As of early October 2026, OpenAI says Dots are rolling out to Pro users in supported markets, excluding the European Economic Area, Switzerland and the UK during the initial rollout.
Business Premium users can access Dots across supported ChatGPT regions. Enterprise, Edu and Healthcare workspaces can try the beta when an administrator enables it.
Rollouts are gradual, so two users on the same general plan may not see the feature at exactly the same time.
Why are AI companies building persistent agents?
Chatbots are useful for questions. Agents are trying to own outcomes.
If you ask a chatbot to prepare for a meeting, it can write a checklist. A persistent agent can potentially keep watching the relevant information, update the preparation as things change and return when there is something important to review.
OpenAI is not alone here. Meta Muse and startups such as Instinct are also competing to become the personal agent people trust with ongoing tasks.
What are the risks?
The more capable an agent becomes, the more important its permissions and review points become. Users should pay attention to what accounts are connected, what actions require approval and what information is stored in the agent’s context.
Persistent agents can save time, but they also create a larger surface for mistakes because they can perform more steps than a simple answer generator.
Should you use a Dot?
Dots make the most sense for responsibilities that continue over time. A one off question does not need a persistent agent.
If the work involves repeated research, follow ups, connected apps or a project that changes daily, the format is more interesting. The real test will be whether the agent stays reliable after days or weeks of work, not whether it can complete one impressive demo.
For more explainers on new apps and online tools, see Holyprofweb’s Apps & Websites section.