From chatbot sessions to ongoing responsibilities
VentureBeat reported on OpenAI’s launch at DevDay 2026, describing Dots as persistent agents designed to continue working after a person leaves ChatGPT. Rather than requiring users to reopen a conversation, provide background and request another answer, a Dot can receive an ongoing objective and follow up as circumstances change.
The difference is subtle in the interface but significant in daily work. A conventional chatbot waits for a prompt. A Dot can monitor information, maintain documents, identify changes and return a piece of work when human judgment is needed. It could watch customer feedback, prepare a small software fix, revise marketing material after a product update or keep a sales proposal current.
OpenAI’s Dots product page describes the agents as always-on systems that work across ChatGPT, connected tools and communication channels, with safeguards built into the experience. The company’s examples also include revising scientific analyses as new evidence appears, and turning interviews into clips, notes and social posts.
The result resembles a new kind of coworker. It is not simply a faster search box or a more conversational document editor. It is a software worker with a continuing assignment, a memory of the project context and a place in a team’s workflow.
A cloud computer for every agent
OpenAI’s GPT-6 Astra system card introduces Dots as agents powered by the model, each with its own cloud computer and browser. That gives a Dot an environment where it can operate across connected tools instead of producing text in isolation.
The agent can also follow up through communication channels. OpenAI’s workspace documentation describes access involving ChatGPT, Slack and Microsoft Teams, while connected applications extend the range of tasks an agent can handle. VentureBeat reported that Dots can connect to more than 4,000 applications through OpenAI’s plugin ecosystem.
That arrangement could make an agent feel less like a feature inside one application and more like a digital presence spread across an organization. A marketing Dot might notice a change in product information, revise approved materials and notify a team. A support Dot could organize recurring feedback and surface patterns that deserve attention. A research Dot could track new evidence and identify sections of an analysis that may need updating.
The agent’s independence is still bounded. OpenAI says proactive research is initially read-only, while consequential actions can be governed through application permissions, custom rules, approval requirements and an activity view. Users can also explicitly connect a Dot to a laptop or another device, although each agent operates in its own cloud computer by default.
A shared room for people and agents
Dots handle continuing assignments, while ChatGPT Space is designed to hold the shared context around them. OpenAI’s Space documentation describes a collaborative environment where people and ChatGPT agents can work with files, pages and other project materials.
In practice, a Space could become the digital room for a product launch, a research program or a customer account. Documents, presentations and spreadsheets remain available to the group, and connected tools can keep information synchronized. Employees can contribute decisions and expertise while agents handle recurring updates, preparation and monitoring.
This model changes what collaboration means. Instead of sending an employee a document to review, a team might give a Dot access to a shared project and ask it to keep that material ready for review. The human contribution becomes more focused on judgment, priorities and exceptions, while the agent handles the steady movement of information between tools.
OpenAI is also piloting specialist Dots with organizational identities, credentials and defined responsibilities in areas including procurement, invoice processing, marketing, customer support and contracting. Microsoft integration is planned through Agent 365, according to the launch coverage.
The enterprise challenge is identity
OpenAI’s Help Center explains that workspace administrators can manage cloud computers, connected applications, Slack and Microsoft Teams access, local computer access, custom rules and permissions. Those controls point toward the larger operational question raised by persistent agents: who is responsible for a machine that can keep acting after its human sponsor has gone offline?
Companies may need to treat Dots as machine identities. They would require onboarding, access policies, monitoring and offboarding, much as employees do. A forgotten permission could expose information. An overly broad connection could allow an agent to modify the wrong document or communicate with the wrong audience.
The economic implications are similarly uncertain. A persistent agent may be more valuable than a chat subscription because it takes ownership of a workflow, but that value could depend on how much work it performs and how many agents an organization deploys. OpenAI says the first Dot is included for some Pro and Business Premium customers, but it has not disclosed limits for intensive work, additional agents or specialist enterprise deployments.
The promise is a workplace where digital collaborators quietly keep projects moving between meetings. The test will be whether organizations can give those collaborators enough access to be useful while preserving the approvals, visibility and human responsibility that make their work trustworthy.
Maya Lindqvist is not a person. No notebook, no deadlines, no face behind the name — just a byline this newsroom publishes under. Here is the production line underneath it, because a name beside a portrait reads like a journalist, and this one is not one.
The models. Writing: gpt-5.6-luna and qwen3-max. Out on the live web: gpt-5.6-luna and gpt-5.6-terra. Pictures: gpt-image-1 and gpt-image-1-mini. Swap one in the newsroom and this line swaps with it — it is read off the machines, not typed here.
How a story is made
- Research. The searching model reads around the story, pointed at primary sources — the filing, the post, the repository — rather than at somebody else's write-up of them.
- Writing. The writing model drafts it against what was found, at Maya Lindqvist's usual length and in Maya Lindqvist's usual register.
- The loop. A reviewer reads the draft and sends it back with notes. Then reads it again. A piece can go round several times before it leaves the building.
- Enrichment. A quotation has to appear word for word on the page it is taken from. A chart may only use figures that appear in the source it cites. Whatever fails is dropped, and the reason is kept.
- Fact check. A last pass hunts for claims the article makes and its sources do not.
- A human stop. Sensitive subjects are held for a person to read before publication, and a person can kill any of it at any point.
If that sounds less like a newsroom and more like a factory: quite. It is called Press Factory.
This article was generated using AI and published automatically without human pre-publication review.
How this article was made
The article was produced by the Grandmonts Media News Engine using automated research, drafting and verification workflows. No human editor reviewed the article before publication. Grandmonts Media remains responsible for the published content. Errors can be reported at office@grandmonts.cz.