ChatGPT no longer just wants to answer your questions. With Dots, OpenAI wants to enable it to work for you, even when you're offline. Presented at DevDay in San Francisco on September 29, 2026, Dots can track objectives over time, use connected applications, and report their results to the user. With the new collaborative environment Space, OpenAI also aims to transform ChatGPT into a true workspace. However, these new features are not yet equally accessible in all countries, particularly in France.
Key points to remember
- A Dowry is a personal agent integrated into ChatGPT that can pursue a mission between two conversations.
- It has its own computer in the cloud and can work with the applications that the user chooses to connect.
- ChatGPT Space brings together pages, files, and work in a shareable environment.
- OpenAI is getting closer to the territory of office suites like Microsoft 365 and Google Workspace.
- At launch, Dots are not available to individual Pro subscribers in the European Economic Area.
- Important actions must remain controlled, as OpenAI recognizes that its agents can make mistakes.
ChatGPT moves from conversation to action
Until now, the traditional use of ChatGPT relied primarily on a series of exchanges. The user would ask a question, receive an answer, and then formulate a new request. The announcements at DevDay 2026 introduce a different logic: the artificial intelligence receives a goal, breaks down the work, and continues to progress over time.
It is this passage of the conversation to the’execution This is the real change. OpenAI no longer presents ChatGPT simply as an assistant to be consulted occasionally. The company wants to make it an environment capable of tracking responsibilities, organizing information, and preparing actions.
What is a Dot?
A Dot is a constantly active agent in ChatGPT, powered by GPT-6 Astra. It has its own computer in the cloud, maintains the context necessary for its work, and can progress between user interactions.
You can give it a goal and specify what it is allowed to accomplish on its own. It can then determine the necessary steps, conduct research, examine related information, and present its results for verification. When a decision requires human judgment, it must consult its owner.
In practical terms, a Dot can receive several types of missions.
- monitor a topic and report any significant developments; ;
- prepare a daily watch; ;
- review a schedule each morning; ;
- follow a to-do list or a project; ;
- search for information in connected applications; ;
- perform recurring checks; ;
- prepare results that will then be validated by a person.
How can this be used in everyday life?
For an individual, a Dot could be used to track deadlines, regularly search for specific information, or plan a trip. For an entrepreneur, it could be used to monitor a market, summarize competitor announcements, and provide reminders of decisions to be made.
In a media outlet, the agent could conduct initial monitoring, identify emerging topics, and gather official sources. In a communications department, they could monitor media coverage or prepare content proposals.
The Dot doesn't necessarily replace the professional. It can primarily absorb some of the repetitive work of research, classification, and preparation. Editorial selection, analysis, decision-making, and responsibility remain human.
Can the Dot control your computer?
Each Dot initially works on its own computer in the cloud. Access to the user's personal computer is optional and disabled by default.
If this access is voluntarily granted from the desktop application, the agent can interact with certain files and create tasks using tools such as Work or Codex. This capability increases automation possibilities, but it requires very specific permissions.
OpenAI provides several levels of control. Depending on the rules chosen, the Dot can act independently, act after an explicit request, solicit confirmation, or leave the decision entirely to the user. The company also indicates that an agent can make mistakes, even when following rules defined by its owner.
ChatGPT Space organizes collective work
The second important announcement for the general public and the teams is called ChatGPT Space. This environment brings together pages, files, and various works related to the same subject.
A page can serve as a project sheet, report, research document, tracking table, or summary. It can also contain graphs and interactive tools generated with ChatGPT.
Multiple people can view or edit the same page. Each collaborator can ask their own ChatGPT for help analyzing or improving the content. Spaces can contain folders, pages, and subpages to structure a project sustainably.
Space does not replace existing ChatGPT projects. The projects continue to organize conversations, files, and instructions, while Space is designed more for producing and sharing structured content.
Is OpenAI preparing a competitor to Microsoft Office?
With collaborative documents, shared files, graphics and soon presentations and spreadsheets, ChatGPT is gradually approaching office suites.
OpenAI is not announcing the demise of Word, PowerPoint, Excel, Google Docs, or Google Sheets. Its objective seems rather to be to reduce the back-and-forth between artificial intelligence and the software into which the result must then be copied, corrected, and shared.
The difference lies in the role of AI. In a traditional office suite, it assists the software. In OpenAI's vision, the user could first express their objective to ChatGPT, which would then organize the information, documents, and some of the necessary actions.
Are these new products available in France?
Availability should be distinguished from the demonstrations performed during the keynote. At launch, Dots are deployed to ChatGPT Pro subscribers located outside the European Economic Area, Switzerland and the United Kingdom.
An individual Pro user based in France cannot yet necessarily create their own Dot. However, OpenAI is announcing Dots for Business Premium users in supported regions. A beta version also exists for Enterprise, Education, and Healthcare organizations when their administrator activates it.
ChatGPT Space is being rolled out gradually to Pro, Business, and Enterprise plans. Pages can be created and edited on the web and in the desktop application. On mobile, they can be viewed and shared, but not yet edited. Presentations and spreadsheets are planned for a later stage.
Risks that should not be underestimated
A permanent agent can save time, but they can also repeat an error, use outdated information, or misinterpret an instruction. The more access they have, the greater the potential consequences.
Before connecting an email, file storage, or business tool, the user should check the permissions granted, the information visible, and the actions the agent can perform without confirmation.
Shared spaces require the same vigilance. Information copied to a page can become visible to everyone with access to that page, even if they could not view the original source.
Our reading
Dots and Space illustrate a profound transformation of ChatGPT. OpenAI no longer simply wants to provide the best possible answer to a question. The company wants its artificial intelligence to pursue objectives, organize knowledge, and participate directly in daily work.
The promise is appealing, with less repetition, fewer mechanical tasks, and greater continuity. But this autonomy also imposes new control rules. The challenge will not only be to determine what the agent can do, but what it should be allowed to do without human intervention.
Also read on Wazup-InTech, GPT-6.1 Ground and autonomous agents, how OpenAI wants to transform the work of companies.
