OpenAI has introduced Open AI Dots, a new class of AI agents designed to do more than answer questions inside a chat window. The idea is to give users an AI assistant that can keep working on projects, use connected apps, operate its own cloud computer and return with completed work instead of waiting for instructions at every step.
OpenAI describes Dots as “always-on agents” built to learn what matters to a user and work towards their goals around the clock. They are powered by GPT-6 Astra, have their own cloud computer and can connect with more than 4,000 apps through OpenAI’s plugin ecosystem.
That makes Dots notably different from the familiar question-and-answer experience associated with AI assistants. A Dot is intended to behave more like a continuing digital collaborator: someone you can hand work to, check in with later and gradually teach how you prefer things done.
What exactly is an OpenAI Dot?
A Dot is a persistent AI agent connected to a user’s ChatGPT environment.
Instead of opening a separate conversation every time a task comes up, a user can give a Dot projects, ideas and follow-up work while it continues handling other assignments. OpenAI says users do not have to manage separate threads or direct every individual step.
The important distinction is continuity.
A Dot can learn a user’s preferences, working style and expectations through repeated interaction and feedback. OpenAI says users can reach their Dot through ChatGPT, Slack or Microsoft Teams, ask questions, discuss ideas, provide feedback or speak with it through a voice call.
Over time, the goal is for the Dot to understand not just what a person is asking for, but also how that person likes the work to be done.
How are Dots different from regular ChatGPT?
The clearest difference is that Dots are designed to keep working even when the user is not actively chatting with them.
Traditional chatbot interactions generally begin when the user sends a prompt. Dots are built around a more persistent model of work.
Each Dot has access to its own cloud computer and browser, along with the applications a user has chosen to connect. Users can open that computer and inspect what the Dot is doing. A Dot can also be given permission to connect to a user’s laptop, allowing it to work alongside the user directly on that machine.
OpenAI also says Dots can work on several projects while continuing to accept new tasks.
That changes the interaction from a sequence of isolated prompts into something closer to an ongoing working relationship.
What can OpenAI Dots actually do?
OpenAI is positioning Dots as general-purpose agents rather than tools built for one narrow job.
One example involves software development. A Dot could watch customer feedback, identify recurring requests, scope smaller improvements and bug fixes, build and test the changes, then prepare pull requests for a developer to review. Introducing dots _ OpenAI
OpenAI also gives examples from its own internal use. According to the company, Dots have been used to investigate bugs appearing in Slack, turn new designs into working applications and help teams stay coordinated during planning cycles. An early tester’s Dot also noticed a forgotten invoice, prepared it and sent it after receiving the user’s approval.
Those examples illustrate the broader ambition behind the product. The Dot is not simply being asked to generate a paragraph, analyse a document or answer a question. It can potentially observe a workflow, perform several steps and bring the result back for review.
Dots can follow users across different apps
OpenAI is also trying to reduce the need for people to return to one particular interface whenever they want to communicate with their agent.
Users can message or call a Dot through ChatGPT on desktop, web and mobile. Dots can also send updates when they need a decision, have a question or want to report progress.
They can be reached through Slack and Microsoft Teams as well, while texting is planned for later. OpenAI says Dots retain context across these channels, meaning a project started in ChatGPT can continue through a conversation with colleagues in Slack without requiring the user to rebuild the background from scratch.
That cross-platform memory is central to the product’s design.
Rather than treating each app as a separate AI session, OpenAI wants the Dot to remain the same assistant wherever the user communicates with it.
What does “proactive research” mean?
One of the more significant features introduced with Dots is what OpenAI calls proactive research.
When a user is not actively working with a Dot, the agent can look for useful things to do in the background using applications that have already been connected.
There is an important restriction, however. During this proactive mode, OpenAI says the tools available to the Dot are read-only. They cannot send messages, change content inside apps or take control of the user’s browser or computer.
In practical terms, this means a Dot may be able to notice something that deserves attention before the user specifically asks about it, but its ability to take consequential action remains constrained.
How much control does the user have?
OpenAI says users decide which applications a Dot can access and can manage those permissions using existing ChatGPT app controls.
Dots also have built-in rules governing when they can act independently and when they must ask the user first.
Users can create Custom Rules that allow certain actions, require approval for others or block particular actions altogether. Progress, including background activity, can be followed through an Activity View, and users can redirect the Dot while it is working.
OpenAI also says its action-review system checks proposed actions against the user’s instructions, Custom Rules and safety requirements to decide whether an action can proceed automatically or requires human approval.
Some sensitive actions remain reserved for the user. OpenAI specifically cites changing a password as an example of a task that the Dot cannot take over. The company also cautions that Dots can still make mistakes and says consequential work should be reviewed.
How does OpenAI handle passwords and computer access?
Dots operate on separate cloud computers.
OpenAI says a user’s own computer and its contents remain separate unless the user explicitly chooses to connect that machine.
For supported websites, Dots can use saved passwords to sign in without exposing those passwords directly to the model. OpenAI also says security systems monitor for malicious instructions and potentially harmful behaviour and can pause or stop a Dot if a safety concern is detected.
These controls matter because Dots are designed to move beyond generating text and into environments where the AI may interact with real accounts, software and company systems.
What happens to data used by Dots?
OpenAI says content from ChatGPT Business, Enterprise and Edu workspaces is not used to improve its models by default.
For personal ChatGPT plans, users can control whether conversations and work involving Dots can be used for model improvement.
OpenAI also says it does not train directly on a Dot’s proactive research or its internal notes to itself. Information from those activities may, however, contribute when it informs an eligible conversation or task, depending on the user’s settings.
What are specialist Dots?
OpenAI is developing a second category called specialist Dots for companies.
While a personal Dot works primarily on behalf of an individual, specialist Dots are designed around defined responsibilities inside an organisation.
A company can give each specialist Dot its own identity, credentials and access to the systems required for its job. OpenAI says it has been testing this approach internally in areas including procurement, invoice processing, email marketing, customer support and commercial contracting.
OpenAI is initially working with organisations through focused enterprise pilots, with its engineering teams helping companies define what each Dot can do, which tools it can use and where human approval is required.
OpenAI is also bringing specialist Dots to Microsoft Agent 365
The company is working with Microsoft to integrate specialist Dots with governance and security controls in Microsoft Agent 365.
The goal, according to OpenAI, is to let businesses manage these agents using Microsoft systems they already use.
For larger organisations, this could become an important part of the proposition. Giving an AI agent access to internal systems is not simply a productivity decision. It also involves identity management, permissions, security controls and oversight.
Who can use OpenAI Dots?
OpenAI is beginning the rollout with ChatGPT Pro and Business Premium users in eligible markets.
Enterprise users, including those using Edu and Healthcare plans, can try the beta when their workspace administrator enables access.
The first Dot is included with Pro or Business Premium at no additional cost.
OpenAI says conversations with a Dot do not count against normal ChatGPT usage limits. However, when a Dot launches or manages work through products such as Codex or ChatGPT Work, those tasks continue to count against the relevant usage limits.
The company says users will eventually be able to add additional Dots and increase either the speed of a Dot or the amount of work it can handle each month.
How do you create an OpenAI Dot?
The initial setup starts through the ChatGPT desktop application or a desktop browser.
Users create their first Dot, connect the applications they want it to access and allow the Dot to introduce itself. Once setup is complete, it can also be messaged through the ChatGPT mobile app.
OpenAI currently describes one primary personal Dot, although its longer-term vision includes multiple Dots working together on a user’s behalf. The company is also previewing specialist agents with distinct identities for organisational roles.
Significance of OpenAI Dots
Dots represent a shift in how OpenAI wants people to interact with artificial intelligence.
The chatbot model starts with a request: a person asks for something and the AI responds.
Dots introduce a different pattern. The user can assign responsibility, allow the agent to stay aware of ongoing work and return when attention or approval is required.
That does not mean the human disappears from the process. OpenAI’s own design repeatedly brings the user back into consequential decisions through permissions, approvals, Custom Rules and activity monitoring.
The larger change is in the amount of work that can happen between those moments.
If the model works as intended, a user would spend less time telling an AI exactly what to do next and more time reviewing completed work, making decisions and setting direction.
That is the central idea behind OpenAI Dots: not another chatbot waiting for the next prompt, but an AI agent designed to remain present, carry context and keep working after the conversation has moved on.
Also Read: YouTube India Report: Regional Creators Turn Local Stories Into Global Businesses
