OpenAI Dots Explained: What Are These New Always-On AI Agents?
What Is OpenAI Dots?
OpenAI has introduced a new type of AI assistant called Dots, marking another major step in the evolution of artificial intelligence.
Dots are designed to work differently from a traditional chatbot. Instead of simply waiting for a user to send a message and then responding, a Dot can be given an ongoing goal and continue working on that responsibility over time.
In simple terms, you can think of a Dot as an AI assistant that you can delegate work to.
You give it a goal, explain what you want it to accomplish, decide what it is allowed to do, and the Dot can work through multiple steps before bringing the results back to you.
OpenAI announced Dots during its DevDay 2026 event in San Francisco on September 29, 2026. The company describes them as proactive, always-on agents designed to handle complex projects and everyday tasks.
How Are OpenAI Dots Different From ChatGPT?
A normal ChatGPT conversation usually works like this:
You ask a question → ChatGPT responds → You ask another question → ChatGPT responds again.
Dots are designed around a different model:
You provide a goal → Dot plans the work → Dot performs multiple tasks → Dot continues working → Dot returns results or asks for a decision.
That difference is important.
A Dot isn't intended simply to give you information. It is designed to take responsibility for completing a larger piece of work.
For example, instead of repeatedly asking an AI to research customer feedback, identify problems and prepare improvements, a user could assign the broader responsibility to a Dot.
The Dot could then work through the individual steps and return the completed work for human review.
What Can OpenAI Dots Do?
Dots are designed for projects that involve several steps rather than a single question.
Some potential uses include:
Researching information
Monitoring ongoing projects
Analyzing customer feedback
Preparing documents
Working with connected applications
Assisting with software development
Creating and updating project materials
Tracking recurring tasks
Communicating through supported work platforms
Following changing requirements
Performing computer-based tasks
Coordinating information from multiple sources
The important idea is that the Dot can continue working toward the overall objective rather than requiring the user to manually guide every individual step.
Dots Can Work With Connected Apps
One of the major features of Dots is their ability to work with applications that the user chooses to connect.
This potentially allows an AI agent to work with information and workflows that already exist inside a person's digital environment.
For example, a professional could potentially use a Dot to help monitor information from workplace applications, organize updates and prepare follow-up work.
OpenAI says Dots can work across connected tools and communication platforms, including environments such as Slack and other supported applications.
This turns the AI from a standalone chatbot into something closer to a digital worker operating across multiple tools.
Dots Have Their Own Cloud Computer
Another important part of the technology is that a Dot can operate using its own cloud computer and browser.
This means the agent isn't limited to generating text inside a conversation.
It can use computer-based tools as part of completing a task.
That capability becomes particularly useful for complicated workflows where the AI needs to navigate websites, use software, gather information or perform several actions in sequence.
The result is a significant change in how people can interact with AI.
Instead of asking:
“How do I do this?”
Users can increasingly ask:
“Can you do this for me?”
OpenAI Dots Are Powered by GPT-6 Astra
OpenAI says Dots are powered by GPT-6 Astra, the company's model designed for advanced computer-use and agentic tasks.
The model provides the intelligence behind the agent while the Dot's environment gives it the ability to work with tools and applications.
This combination is important because an AI model by itself can generate an answer, while an agent can use that intelligence to perform a sequence of actions.
That is the fundamental idea behind Dots.
What Does "Always-On" Mean?
The phrase “always-on” is one of the most important parts of the Dots announcement.
A traditional AI conversation generally stops when the user stops interacting with it.
A Dot is designed to continue working on assigned responsibilities between conversations.
For example, imagine you ask a Dot to monitor a project.
Instead of having to return every morning and ask for an update, the Dot could continue following the project and bring important developments to your attention.
This creates a new relationship between people and AI.
The user becomes more like a manager, while the AI handles portions of the ongoing work.
Can Dots Work on Multiple Projects?
Yes.
OpenAI describes Dots as being able to take on ongoing projects and continue making progress while the user works on other things.
This means a person could potentially delegate different responsibilities to different Dots.
For example:
Dot 1: Research competitors
Dot 2: Monitor customer feedback
Dot 3: Assist with software development
Dot 4: Prepare weekly business information
Instead of maintaining several separate conversations and manually coordinating each task, the user could delegate different responsibilities to different AI agents.
An Example: Software Development
OpenAI has highlighted software development as one example of how Dots could be used.
Imagine a developer working on a mobile application.
The Dot could monitor customer feedback and identify repeated requests or smaller problems.
It could then help define potential improvements, work on fixes, test those changes and prepare the completed work for the developer to review.
The developer could continue focusing on major product development while the Dot handles smaller pieces of work.
This illustrates the bigger idea behind agentic AI:
AI doesn't just help you think about the work. It can help carry out the work.
Dots Can Adapt When a Project Changes
Real-world projects rarely remain unchanged.
Requirements change.
Customers provide new feedback.
Deadlines move.
New information appears.
A traditional automation may break when its original conditions change.
Dots are designed to deal with more dynamic situations.
For example, if a sales proposal changes because a customer requests different features, an agent can potentially update related material rather than requiring the entire process to be restarted manually.
This flexibility is one of the reasons agentic AI is attracting significant attention.
What Are Custom Rules for Dots?
OpenAI has also introduced controls that allow users to define how a Dot should behave.
Users can establish rules about what an agent is allowed to do independently and when it should ask for approval.
This is important because an AI that can access applications has much more power than an AI that simply generates text.
For example, a user may want a Dot to freely research information but require approval before making an important external change.
These controls create a boundary between autonomous work and human decision-making.
Are OpenAI Dots Safe?
Safety is a major issue with any AI system that can take actions.
An ordinary chatbot might provide incorrect information.
An agent can potentially take an incorrect action.
That difference makes safeguards particularly important.
OpenAI says Dots include monitoring and safety mechanisms designed to detect potentially harmful behavior. The company has also tested Dots against attempts to manipulate agents through malicious instructions and other attacks.
However, no complex software system can be assumed to be completely risk-free.
Users should therefore pay attention to permissions and understand what applications and information an AI agent can access.
The more authority an agent receives, the more important oversight becomes.
What Happens If a Dot Needs a Human Decision?
Dots are not designed to eliminate humans from every workflow.
There are situations where the agent may need the user to make a decision.
For example, a Dot might complete research and prepare several possible actions but leave the final decision to the person.
This creates a useful model:
AI handles execution.
Human handles judgment.
That distinction could become increasingly important as AI agents become more capable.
Who Can Use OpenAI Dots?
OpenAI is rolling out Dots across supported versions of ChatGPT.
The initial availability includes eligible Pro and Business Premium users, while Enterprise users can access a beta depending on workspace settings and administrator controls.
The rollout is gradual, so availability can vary by account and region.
OpenAI has also stated that some regions are excluded from the initial Pro rollout.
Because the feature is new, access and usage conditions may change as the rollout expands.
Why Are People Searching for OpenAI Dots?
There are several reasons Dots have attracted so much attention.
1. AI Is Moving Beyond Chatbots
For years, the main AI experience was asking questions and receiving answers.
Dots represent a move toward AI systems that can perform tasks independently.
That is a major change.
2. AI Could Become a Digital Coworker
People are increasingly interested in AI that can perform practical work.
An always-on agent could potentially handle repetitive research, monitoring, documentation, software tasks and other responsibilities.
This makes Dots relevant to businesses as well as individual users.
3. OpenAI Is Entering the AI Agent Race
Dots arrive during a period of intense competition in agentic AI.
Technology companies are developing systems designed to perform tasks rather than simply generate content.
OpenAI's Dots therefore represent another step in the industry's broader transition from conversational AI toward autonomous or semi-autonomous digital assistants.
4. Dots Can Work in the Background
The ability to continue working after the user leaves the conversation is one of the most noticeable differences.
People are interested in what happens when AI doesn't need to wait for the next prompt.
5. The Technology Raises Important Questions
More capable agents also create new questions.
Who controls the agent?
What information can it access?
What happens if it makes a mistake?
When should it ask for permission?
How should businesses monitor autonomous AI?
These questions are becoming increasingly important as AI moves into real-world workflows.
OpenAI Dots vs Traditional AI Assistants
The difference can be understood simply:
| Traditional AI Assistant | OpenAI Dots |
|---|---|
| Responds to prompts | Works toward assigned goals |
| Usually operates within a conversation | Can handle ongoing responsibilities |
| User often guides each step | Can determine the next steps |
| Mainly generates information | Can perform actions using connected tools |
| Work generally stops after the response | Designed to continue working |
| Limited task autonomy | Greater task autonomy |
| Human manages most of the workflow | AI can manage portions of the workflow |
The biggest change is not simply intelligence.
It is agency.
What Could OpenAI Dots Mean for Jobs?
The growth of AI agents could change how many digital jobs are performed.
Tasks involving repetitive research, reporting, administration, software testing, monitoring and documentation could potentially become increasingly automated.
At the same time, people may spend more time reviewing AI-generated work, making decisions, setting goals and handling situations that require human judgment.
The impact will likely vary considerably by industry and occupation.
Rather than replacing every job, agentic AI could initially change the mix of tasks within many jobs.
What Could Dots Mean for Businesses?
Businesses are particularly interested in AI agents because employees spend significant amounts of time moving information between different systems.
An agent capable of working across applications could potentially reduce some of that manual coordination.
For example, a business could eventually use AI agents to help monitor sales information, organize customer feedback, prepare reports and coordinate project updates.
The key challenge will be ensuring that these systems are reliable and that humans remain able to supervise important decisions.
What Makes Dots Different From Simple Automation?
Traditional automation normally follows predefined instructions.
For example:
If X happens → perform Y.
AI agents can operate more flexibly.
They can interpret information, decide what step may be appropriate next and adapt when circumstances change.
That makes them potentially more useful for complicated projects.
It also makes them harder to predict.
This is why agentic AI requires stronger safeguards than simple rule-based automation.
The Bigger Story Behind OpenAI Dots
The launch of Dots is about more than another ChatGPT feature.
It represents a broader shift in artificial intelligence.
The first major wave of generative AI focused on creating content.
The next wave is increasingly focused on performing work.
AI systems are moving from:
Answering → Reasoning → Using tools → Taking actions → Managing ongoing tasks
Dots fit directly into this transition.
If agentic AI continues to improve, users may eventually spend less time telling AI exactly what to do and more time explaining the outcome they want.
Final Thoughts
OpenAI Dots introduce a different vision of what an AI assistant can be.
Instead of waiting for every individual instruction, a Dot can be given an ongoing responsibility and allowed to work through multiple steps.
That could make AI considerably more useful for research, business operations, software development, project management and everyday digital work.
But greater autonomy also means greater responsibility.
The most important questions will not only be about how intelligent these agents become. They will also be about how reliably they operate, what permissions they receive, how their actions are monitored and where humans remain in control.
OpenAI Dots are therefore an important development in the move from conversational AI toward agentic AI.
The future of AI may not simply be about having a smarter chatbot.
It may be about having an AI system that can take a goal, work on it continuously and return with the finished result.