Technology•

Seven Weeks After DevDay, OpenAI Puts Its Always-On Dots Agents in Front of a Small Group of ChatGPT Plus Subscribers

OpenAI began testing Dots, agents that run continuously on their own cloud computers, with selected ChatGPT Plus users and eligible US Business Standard members on November 19, excluding Europe and the UK.

Written By :MINRK
Seven Weeks After DevDay, OpenAI Puts Its Always-On Dots Agents in Front of a Small Group of ChatGPT Plus Subscribers

Photo by Christina @ wocintechchat.com M on Unsplash

1. A limited test begins for ChatGPT Plus and US business accounts

OpenAI Dots, the company's always-on artificial intelligence agents, entered testing on November 19, 2026, with a selected group of ChatGPT Plus subscribers and eligible ChatGPT Business Standard members in the United States. Each eligible user receives one personal dot, an agent that runs on its own dedicated cloud computer and keeps working on projects without being prompted at every step.

The test marks the first time the feature has reached the company's mainstream paid tier. Plus is the subscription most individual ChatGPT customers hold, so even a restricted trial there signals how OpenAI intends to move agents from a showcase product toward everyday use.

The two groups in the trial represent different kinds of customer. Plus subscribers are individuals paying for a personal account, while Business Standard is a plan sold to organizations whose employees share a workspace. Running the test across both lets the company see how the same agent behaves in private use and in a professional setting.

Access is narrow by design. Only some Plus subscribers have been chosen, the business side of the test is confined to eligible Business Standard accounts in the US, and each participant is limited to a single dot. The company has not opened the feature to its full subscriber base.

2. What is OpenAI Dots?

OpenAI Dots are AI agents that stay active continuously, each one running on a dedicated cloud computer, so that they can carry a project forward on a user's behalf instead of waiting for a new instruction after every reply. OpenAI introduced them at its DevDay 2026 conference on September 29, 2026.

The difference from a conventional chatbot lies in how the work is organized. A standard ChatGPT exchange follows a simple rhythm: a person types a request, the model produces an answer, and the interaction stops until the person types again. The system does nothing in between. A dot is meant to break that pattern by continuing to act after the conversation has moved on.

The dedicated computer is what makes that possible. In an ordinary chat, computing capacity is assigned for the few seconds needed to generate a response and then released. A dot instead has a persistent workspace of its own in the cloud, a place where files, tools and work in progress remain available from one session to the next. That gives the agent somewhere to keep operating when its owner is not watching.

The name reflects the personal nature of the arrangement. A dot is assigned to one user and works for that user, which separates it from a shared tool that anyone in an organization can call on. In the present test, that relationship is one to one: each participant has a single agent.

Dots are powered by OpenAI's GPT-6 Astra model. The pairing of a capable model with a standing computer is the core of the product: the model supplies the reasoning, and the machine supplies the continuity that lets a task run for longer than a single exchange.

3. How an always-on agent differs from a chat assistant

Chat assistants are reactive tools. They are useful for drafting a message, summarizing a document or answering a question, but each of those jobs begins and ends with a prompt. Anything that unfolds over hours or days, such as following a project through several stages, has to be driven manually by the user returning to the conversation again and again.

Agents are an attempt to remove that burden. An agent is given an objective and some latitude to pursue it, deciding which steps to take and in what order. The always-on version of the idea goes a step further by keeping the agent running in the background, so that progress does not depend on the user being present.

For the person using it, the practical change is one of role. With a chat assistant, the user is an operator issuing a stream of commands. With a persistent agent, the user is closer to a manager who sets a goal, checks in periodically and reviews what has been produced.

That shift is also why companies developing these systems tend to release them gradually. An agent that acts on its own for extended periods can make mistakes that go unnoticed for longer than an error in a single chat reply, which raises the stakes for reliability and for the controls that keep a user in charge of consequential actions.

4. From stage demonstration to real users in about seven weeks

The gap between announcement and this test was short. Dots were presented at DevDay on September 29, and the Plus and Business Standard trial started on November 19, roughly seven weeks later.

The staged nature of the release is visible in its limits. One dot per user keeps demand on computing resources predictable, since every agent requires a cloud machine of its own. Selecting only a portion of Plus subscribers lets the company observe how people use the feature before committing to a wider launch. Restricting the business test to eligible US accounts narrows the range of conditions it has to support at the outset.

Seven weeks is a brief interval for a product of this type. Features shown at developer conferences often remain in preview for months, available only to a company's highest-priced plans or to hand-picked partners. Here, standard Plus subscribers are being included, even if only in small numbers, within two months of the announcement.

A phased approach of this kind is common for products that are expensive to run. Unlike a chat reply, which consumes computing power briefly, a continuously running agent occupies infrastructure for as long as it is active. Expanding access to a large subscriber base in one step would multiply that cost immediately.

5. Is OpenAI Dots available in Europe?

No. The current test excludes users in the European Economic Area, the United Kingdom and Switzerland, so subscribers in those regions cannot take part even if they hold an eligible plan.

The exclusion means a large share of OpenAI's potential international customers will have to wait. No date has been given for when the feature might reach those markets.

Staggered regional launches have become a familiar pattern for new AI features, which frequently arrive in the United States first and in European markets later. For subscribers in the excluded regions, the practical effect is that they pay for the same plan as participants elsewhere but cannot yet be considered for the trial.

On the business side the geographic limit is tighter still, because the Business Standard portion of the test applies to eligible members in the United States. A company with teams spread across several countries would, for now, see the feature only for its US-based staff on that plan.

6. A redesigned mobile app is being tested alongside OpenAI Dots

Alongside the agent trial, OpenAI is beta-testing a refreshed layout for its mobile app. The redesign is intended to make working with dots smoother and to improve access to the tools inside the app.

The connection between the two tests is straightforward. An interface built around a single conversation window suits a product in which the user and the model take turns. It is less suited to an agent that keeps working in the background, where the user needs a way to see what is in progress and to step in when required. A phone is also the device on which many people would check on a long-running task during the day.

Testing the interface and the agent together gives the company feedback on both at once: whether the agents are useful, and whether people can follow and direct them comfortably from a mobile screen.

7. Why the business tier is part of the test

Including Business Standard accounts points to the workplace as a target for the product. Much office work consists of tasks that continue over time and involve repeated follow-up, the kind of activity a persistent agent is designed for.

Tracking the status of a project, keeping on top of follow-ups with clients and handling routine coordination are examples of work that an always-on agent could take over. If dots perform those jobs reliably, the time saved would come from the unglamorous administrative layer of a working day.

The value of that kind of help depends on trust. An employee will hand recurring coordination work to an agent only if it can be relied on to do the work correctly and to keep going without supervision, which is the capability a dedicated, continuously running computer is meant to provide.

Business customers also provide a different kind of evidence from individual subscribers. Their usage shows whether an agent can fit into established routines and shared responsibilities, which is a harder test than helping one person with a personal task.

8. What comes next for the rollout

OpenAI has not announced when dots will be offered to all Plus subscribers, or when the regional restrictions will be lifted. The present phase is a test, and its scope remains limited to the selected Plus users and eligible US Business Standard members who received access from November 19.

The next signals to watch are an expansion of the Plus group beyond the initial selection, any change to the limit of one dot per user, and an extension of availability to the European Economic Area, the United Kingdom and Switzerland. The outcome of the mobile app beta will indicate how the company plans to present agents to a mass audience.

For now, the test establishes one point clearly. A feature that was a conference demonstration at the end of September is being used by paying customers on the company's standard consumer plan before the end of November, with the wider release still to be scheduled.

Related Articles

NEWSLETTERS

Don't miss another story.

Subscribe to the MINRK Newsletter today.

Tick the newsletters you want. No selection? You'll get MINRK Headlines.

By signing up, you will receive emails about MINRK products and you agree to our terms of use and privacy policy.

MINRK Daybook

Market analysis for crypto traders and investors.

EVERY WEEKDAY

MINRK for Advisors

Defining crypto, digital assets and the future of finance for financial advisors.

EVERY THURSDAY

MINRK Protocol

Exploring the tech behind crypto one block at a time.

WEEKLY

MINRK Long & Short

A must read for institutions. Insights, news and analysis delivered weekly.

EVERY WEDNESDAY

MINRK Headlines

The biggest crypto news and ideas of the day.

EVERY WEEKDAY

MINRK Policy

Examining the intersection of cryptocurrency and government.

WEEKLY

MINRK Research

Join thousands of readers who rely on MINRK for data-driven insights on the latest digital asset trends.

MONTHLY