What if your AI assistant could keep working even after you close ChatGPT?
That is the idea behind OpenAI Dots, the company’s new generation of always-on AI agents. OpenAI introduced Dots at DevDay 2026 as agents designed to work toward your goals, handle multi-step tasks and continue working in the background.
Unlike a traditional chatbot, Dots are designed to take on ongoing work instead of waiting for every individual instruction. They can use a cloud computer, browser and connected apps while keeping context about the work you have assigned.
So, what exactly are OpenAI Dots, and how different are they from a normal AI assistant?
Overview
What are OpenAI Dots?
OpenAI Dots are always-on AI agents designed to handle tasks over longer periods.
Each Dot runs on its own cloud computer and browser. It can also connect with more than 4,000 apps through ChatGPT’s connected tools.
Instead of simply answering a question, a Dot can work toward a larger goal. For example, you could give it a project, provide feedback and let it continue working while you focus on something else.
Who are Dots for?
Dots could be useful for:
- Professionals
- Researchers
- Developers
- Content creators
- Business teams
- Students
- People managing repetitive digital tasks
Why does it matter?
The biggest change is persistence.
Traditional AI usually waits for you to send another message. Dots are designed to continue working in the background.
As a result, the role of AI can shift from answering questions to managing ongoing digital work.
OpenAI Dots Specifications
| Feature | Details |
|---|---|
| Product | OpenAI Dots |
| Type | Always-on AI agents |
| AI Model | GPT-6 Astra |
| Computer | Dedicated cloud computer |
| Browser | Built-in cloud browser |
| Connected Apps | 4,000+ |
| Background Work | Yes |
| Multi-step Tasks | Yes |
| Context | Persistent across work |
| Communication | ChatGPT, Slack and Microsoft Teams |
| Voice Interaction | Supported |
| Feedback | Learns from user feedback |
| Custom Rules | Supported |
| Primary Dot | Available |
| Team of Dots | Coming later |
OpenAI says Dots are powered by GPT-6 Astra and can use their own cloud computers and browsers to carry out tasks.
Performance
Always-On AI
The biggest feature is right in the name.
Dots are designed to remain active instead of ending when you leave the conversation. OpenAI says they can work toward users’ goals 24/7.
For example, you could ask a Dot to research a topic, monitor information and prepare an update. Instead of repeatedly giving instructions, you can provide feedback while the agent continues the task.
Multi-Step Tasks
Dots are designed for work that requires several steps.
A task could involve researching information, opening websites, working with connected apps and preparing a result.
Therefore, Dots move beyond the simple question-and-answer model that most people associate with chatbots.
Connected Apps
Another major feature is app connectivity.
OpenAI says Dots can connect to more than 4,000 apps. This gives the agents access to a much wider digital workspace.
For example, connected applications can provide the information or tools needed to complete a larger workflow.
However, access still depends on the permissions and apps connected to the Dot.
Cloud Computer
Every Dot gets its own cloud computer and browser.
This means the agent can work inside a computer environment rather than simply generating text inside a chat window.
Consequently, Dots can handle tasks that require browsing websites and interacting with digital tools.
Learning From Feedback
Dots are also designed to learn from the way users work.
OpenAI says a Dot can learn your preferences, how you think about tasks and what a successful result looks like over time.
That could make repeated workflows more personalized.
At the same time, it raises an important question: how much information should an always-on AI remember?
Dots Can Work Across Different Apps
OpenAI is also making Dots available beyond the ChatGPT interface.
Users can interact with their Dots through ChatGPT, Slack and Microsoft Teams. OpenAI has also described voice interaction as another way to communicate with a Dot.
This could make Dots more useful for teams.
Instead of opening a separate AI application, users could communicate with their agent through the tools they already use for work.
Dots and Human Approval
Always-on AI does not mean the agent should make every decision on its own.
OpenAI says Dots follow ChatGPT’s permissions and support custom rules that determine which tasks require approval.
For example, a user could allow an agent to research information automatically while requiring approval before it takes a more consequential action.
This creates an important balance between automation and human control.
Privacy and Security
The always-on design also creates new security questions.
A Dot can work with connected apps, websites and information over an extended period. Therefore, permissions become particularly important.
OpenAI’s workspace controls allow administrators to manage Dots access, connected capabilities and computer permissions. For Enterprise workspaces, Dots are off by default and administrators can decide which members can use them.
In addition, organizations can control whether Dots can use local computer access and whether users can create custom rules.
These controls matter because an AI agent with more access can also have more opportunities to make mistakes.
OpenAI Dots vs Traditional Chatbots
| Feature | Traditional Chatbot | OpenAI Dots |
|---|---|---|
| Interaction | User prompts it | User assigns ongoing work |
| Background Work | Limited | Yes |
| Persistent Tasks | Limited | Yes |
| Cloud Computer | Not typical | Yes |
| Browser Access | Limited/varies | Yes |
| Connected Apps | Some | 4,000+ |
| Multi-Step Work | Yes | Designed for ongoing workflows |
| Feedback | Per conversation | Can shape future work |
| Human Approval | Varies | Custom rules supported |
The biggest difference is therefore not simply intelligence. It is how the AI works over time.
What Can OpenAI Dots Actually Do?
The possibilities depend on the connected apps and permissions.
For example, a Dot could potentially help with:
- Research projects
- Market research
- Monitoring information
- Preparing reports
- Organizing digital work
- Repetitive business tasks
- Managing connected applications
- Following up on ongoing projects
- Researching information in the background
However, users still need to define the goal and decide how much autonomy to give the agent.
Dots Are More Than Another Chatbot
ChatGPT changed how people interact with AI by making conversations the main interface.
Dots take that concept further.
Instead of asking, “What can you tell me?”, the user can increasingly ask, “Can you handle this for me?”
That difference could become important as AI agents become better at using software and completing long-running tasks.
OpenAI Dots vs AI Assistants
Traditional digital assistants generally wait for commands.
Dots are designed to work more proactively.
For instance, a normal assistant might answer a question about a project. A Dot could instead keep researching that project, monitor new information and return with an update.
Therefore, the relationship between the user and AI becomes more like delegating work to a digital coworker.
Availability
OpenAI introduced Dots at its September 29, 2026 DevDay event. The company is rolling out the feature gradually across eligible plans and markets.
Current information indicates Dots are available to selected Pro and Business Premium users, while Enterprise access is being introduced through a beta with workspace controls. Availability can vary by region and account.
OpenAI has also indicated that more access will come later.
Pricing
OpenAI has included an initial Dot for eligible users, while additional capacity is tied to plan allowances. Reports also indicate that OpenAI introduced a higher-priced option for users who need more agent capacity.
Because access and plan details are still rolling out, users should check the current ChatGPT plan information rather than relying on a single launch-day price.
FAQ
OpenAI Dots are always-on AI agents designed to work on longer-running tasks instead of simply answering individual prompts.
Dots are powered by GPT-6 Astra, according to OpenAI.
Yes. Their main purpose is to continue working toward assigned goals in the background rather than stopping when the user leaves the conversation.
Yes. OpenAI says Dots can connect to more than 4,000 apps, depending on available connections and permissions.
Yes. Each Dot has access to its own cloud computer and browser for supported tasks.
Yes. OpenAI supports permissions and custom rules that can determine when an agent needs approval. Enterprise administrators can also control access and capabilities.
Availability depends on the ChatGPT plan, account and rollout region. OpenAI is gradually expanding access, so users in Nepal should check their ChatGPT account for availability.
Final Verdict
Should you use OpenAI Dots?
OpenAI Dots represent a major shift from AI that answers to AI that works.
Their biggest feature is not simply a smarter chatbot. Instead, Dots are designed to take a goal, use connected tools and continue working over time.
For researchers, professionals and businesses, that could make repetitive digital work much easier.
At the same time, always-on agents require careful permission management. More access means more responsibility for both the user and the platform.
Best for:
- Long-running research
- Business workflows
- Repetitive digital tasks
- Content research
- Professional work
- Team collaboration
- Users who want more proactive AI assistance
Final Thought
OpenAI Dots show where AI assistants could be heading next.
The future may not be about constantly opening an AI chatbot and asking another question. Instead, you could assign a goal, let the agent work in the background and return when there is something useful to review.
That makes Dots one of OpenAI’s more significant moves toward always-on, agentic AI.
The bigger question now is not whether AI can do more work.
It is how much work people will be comfortable handing over to it.
