OpenAI launches Dots, claims AI agents can work while you sleep and complete tasks based on how you want them done

Synopsis

The new agents are designed to handle longer-running tasks instead of waiting for every instruction, with access to connected apps, a cloud computer and user preferences, according to OpenAI.

OpenAI has launched Dots, a new type of AI agent designed to keep working on a user’s goals even after the conversation ends.

The company announced Dots at DevDay 2026 on September 29 in San Francisco. Dots are powered by GPT-6 Astra and are designed to handle longer-running tasks instead of waiting for users to give instructions at every step.

What are OpenAI Dots?

Dots are described by OpenAI as always-on agents and also different from a normal ChatGPT conversation.

A user can give a Dot a responsibility or a goal. The Dot can then work on the task, decide the next steps and continue making progress between conversations.

It has its own cloud computer and browser. It can also use connected tools and apps to complete work. For example, a developer could ask a Dot to monitor customer feedback.

Live Events

The Dot could identify repeated requests, work on smaller improvements, build and test changes, and return completed pull requests for the developer to review.

According to OpenAI, it has the ability to work with more than 4000 connected apps. Dots can also work with services such as Slack and Microsoft Teams.

It can keep working after the user leaves ChatGPT

A user can give the Dot a goal and leave. The Dot can continue working in the background and return with progress, results or a request for the user to make a decision.

OpenAI said that users can review the work and give the Dot a new direction whenever they want. Dots can reach out through supported communication channels when a decision needs the user's input.

OpenAI said Dots can use context from ChatGPT memory and learn from the user's feedback.

The more a user works with a Dot, the more it can understand their preferences, working style and what they consider a good result.

Users can ask questions, explore ideas and provide feedback while working with their Dot. The company also says users can jump on a voice call when they want to talk through something.

Each Dot gets its own cloud computer and browser and allows it to perform actions rather than simply provide instructions to the user.

OpenAI said that if the users wish they can also connect their own computer while setting permissions and rules for what the Dot can do. There are also controls for actions that require approval.

The company also described an Auto-review feature for actions that could affect accounts or involve sharing information. OpenAI said that they check these actions against the user’s instructions, custom rules and safety requirements before they are carried out.

Where are Dots available?

Dots are currently being rolled out to ChatGPT Pro and Business Premium users in supported regions.

For ChatGPT Pro users, OpenAI says Dots are available in supported markets except the European Economic Area (EEA), Switzerland and the UK.

For Business Premium users, OpenAI says Dots are available across supported ChatGPT regions.

Enterprise users can access a beta version when their workspace administrator enables it. OpenAI says the Enterprise setting is initially turned off by default.

Dots can currently be created through the ChatGPT desktop app or desktop web. Users can interact with their Dot through supported mobile access, but Dots cannot currently be created on mobile, and they are not supported on mobile web.

OpenAI also says a limited texting beta is available to Pro users in the U.S. The texting feature is not currently available for Business or Enterprise workspaces.

Eligible users can start with one Dot at no additional cost. OpenAI says users can create and name their primary Dot, while additional Dots are planned as the service expands.

OpenAI also introduced GPT-6.1 Sol

Alongside Dots, OpenAI introduced GPT-6.1 Sol and described it as a newer model in the GPT-6 family that offers capabilities comparable to GPT-6 Astra while focusing on speed and affordability.

OpenAI said GPT-6.1 Sol improves areas including agentic coding, computer use and professional work.

It is being rolled out in ChatGPT Work and Codex, starting with Pro users and expanding to Plus, Business, Enterprise and Edu, with Enterprise and Edu administrators required to enable access.

Earlier in September, OpenAI introduced the Agents API, which lets developers build and run cloud-based agents using the infrastructure behind Codex. The API is designed for agents that can work for longer periods, use tools, work with files and code, and coordinate subagents.

Gold paradox: Why India, a refining giant, is not a big bullion exporter

A refinery is not a laptop: Why America's diesel problem can't be switched off

How Irdai’s big reset can change insurance landscape, customer choice

Everyone asks who will fund AI. Few ask who pays if it burns

Can Fly91’s flightpath for India’s smallest airline defy the doomy past?

Smugglonomics: Geckos, dry dates, human hair – what makes something worth smuggling?