Artificial intelligence is moving beyond the simple chatbot.
Instead of waiting for users to type a question, the next generation of AI is being designed to take action, continue working, and return with results later.
OpenAI is now pushing this idea with its new Dots, a new type of “always-on” AI agent announced at OpenAI DevDay on September 29, 2026.
Unlike a normal ChatGPT conversation, Dots are designed to take responsibility for ongoing tasks. OpenAI says they can keep making progress between conversations, work with connected applications, and ask the user for input when an important decision requires human judgment.
This could represent an important change in how people think about AI assistants.
What Are OpenAI Dots?
OpenAI describes Dots as always-on AI agents that can work toward goals over time.
A traditional chatbot usually works like this:
You ask → AI answers → conversation ends.
An AI agent works differently:
You give it a goal → AI plans → AI performs tasks → AI checks progress → AI reports back.
Dots are designed around this second model.
According to OpenAI, each Dot has its own cloud computer and browser and can work with connected tools and applications. Dots can also maintain context for ongoing work rather than treating every interaction as a completely new conversation.
That means an AI assistant could potentially continue working even when you’re not actively chatting with it.
How Are Always-On AI Agents Different From Chatbots?
This is one of the most important differences.
A chatbot is primarily reactive.
If you ask:
“Summarize these reports.”
It summarizes the reports.
An agent is intended to be more goal-oriented.
You might instead give it a broader instruction such as:
“Track these reports, identify important changes, prepare a weekly summary, and tell me when something needs my attention.”
The agent can then work through multiple steps instead of simply producing one answer.
OpenAI’s broader agent documentation describes agents as systems that can independently accomplish tasks using models, tools, and defined guardrails.
Dots Can Work Across Apps
Another important feature is connectivity.
OpenAI says Dots can work across ChatGPT and connected services, including communication platforms such as Slack and Microsoft Teams. The company also says Dots can work with more than 4,000 supported apps through integrations.
This could make AI more useful for everyday work.
For example, imagine a marketing manager who needs to monitor several projects.
Instead of checking:
- Slack
- Documents
- Project management tools
- Research websites
one by one, an AI agent could potentially gather relevant information and provide an organized update.
The human still makes important decisions, but the AI handles more of the repetitive work.
OpenAI Dots Can Keep Working in the Background
The “always-on” part is what makes Dots particularly interesting.
OpenAI says Dots can continue working toward assigned goals and follow up when something requires the user’s attention.
Consider a simple example.
You could ask an AI agent to research a technology topic.
A traditional chatbot might give you an answer immediately.
An always-on agent could instead:
- Research the topic.
- Collect information from multiple sources.
- Organize the findings.
- Identify important developments.
- Prepare a draft.
- Return with the completed work.
The important change is that the AI is no longer limited to a single question-and-answer interaction.
Dots Are Powered by GPT-6 Astra
OpenAI says Dots are powered by GPT-6 Astra, its newer AI model.
The company has also designed Dots around their own cloud computing environment, giving the agents access to a browser and tools needed to perform tasks.
This combination is important because an AI agent needs more than a language model.
It needs the ability to:
- Understand instructions
- Plan tasks
- Use tools
- Access information
- Perform actions
- Remember relevant context
- Check its own progress
- Ask for human input
This is why the AI industry is increasingly focusing on agentic AI rather than only larger language models.
What Could People Use Dots For?
The potential applications are broad.
1. Research
An AI agent could monitor a topic over time and organize new information.
For example, a technology website could use an agent to track:
- AI announcements
- Smartphone launches
- Semiconductor news
- Software updates
- Startup funding
- Cybersecurity developments
The agent could then notify the user when something important happens.
2. Content Creation
Content creators could use AI agents to assist with the production process.
An agent could potentially help with:
- Topic research
- Keyword research
- Content outlines
- Draft preparation
- Fact organization
- Updating older articles
- Social media ideas
Human editing would still be important, especially for accuracy, originality, and editorial quality.
3. Software Development
Developers could give an AI agent a larger software task rather than asking for individual code snippets.
For example:
“Review this project, identify bugs, suggest fixes, and prepare the changes for review.”
That is closer to delegating a project than asking a chatbot a question.
OpenAI’s developer ecosystem is already heavily focused on agent development, including tools for building agents, evaluating them, and connecting them to applications.
4. Business Tasks
Companies could use agents for repetitive workflows.
Possible examples include:
- Preparing reports
- Monitoring customer requests
- Organizing information
- Checking project updates
- Preparing meeting summaries
- Tracking business tasks
This could be especially useful when work involves information spread across multiple applications.
Why “Always-On” AI Matters
The biggest change may not be intelligence.
It may be continuity.
People already use AI for writing, research, coding, planning, and analysis.
But users usually have to start the interaction every time.
Always-on agents could change that relationship.
Instead of:
“AI, do this.”
The experience could become:
“Here is my goal. Keep working on it and let me know when something important happens.”
That is a much more powerful concept.
The Privacy Question
Always-on AI also creates an important challenge: access.
An agent that can work across multiple applications potentially has access to more information than a normal chatbot.
That makes permissions extremely important.
OpenAI says Dots include controls that allow users to manage what the agents can do, and the company has emphasized safeguards around agent actions.
Users should still think carefully before connecting an AI agent to sensitive accounts.
The more powerful the agent becomes, the more important it is to understand:
- What data can it access?
- Which apps can it control?
- What actions can it perform?
- When does it need approval?
- How can access be removed?
Convenience and control need to develop together.
Could AI Agents Replace Human Workers?
This is one of the biggest questions surrounding agentic AI.
AI agents can automate certain repetitive tasks, but that does not mean every job can simply be handed to an AI.
Human skills remain important for:
- Judgment
- Creativity
- Leadership
- Relationship building
- Strategy
- Accountability
- Complex decision-making
In many workplaces, the more realistic near-term change may be AI working alongside people rather than replacing every person.
An employee might delegate repetitive research to an agent and spend more time reviewing results and making decisions.
OpenAI Dots vs Traditional AI Assistants
The difference can be summarized simply:
| Traditional AI Assistant | Always-On AI Agent |
|---|---|
| Waits for a prompt | Can continue working toward a goal |
| Usually handles individual requests | Can manage multi-step tasks |
| Primarily responds | Can take actions using tools |
| Limited continuity | Designed for ongoing work |
| User starts most tasks | Agent can follow up when needed |
The technology is still developing, but the direction is clear: AI is moving from answering questions toward completing workflows.
OpenAI Is Entering a Bigger AI Agent Race
OpenAI is not alone in pursuing this idea.
Other technology companies are also developing AI systems that can perform tasks on behalf of users.
The competition is shifting from:
“Who has the best chatbot?”
to:
“Who can build the most useful AI agent?”
That is a much bigger challenge.
A successful agent needs reliable reasoning, useful integrations, security, permissions, memory, and the ability to recover when something goes wrong.
Simply having a powerful language model is not enough.
What Does This Mean for the Future of AI?
The introduction of Dots suggests that AI assistants could become more like digital coworkers.
Instead of opening an AI application whenever you need help, you could have an agent continuously working on selected responsibilities.
Imagine starting your day and receiving a message:
“I reviewed your project updates, identified three issues, prepared a report, and found one decision that needs your approval.”
That is very different from today’s typical chatbot experience.
It could save time, but it also means people will need to become comfortable managing AI permissions and reviewing AI-generated work.
Final Thoughts
OpenAI’s new Dots are an example of how quickly AI assistants are evolving.
The next stage of AI may not simply be about generating better answers.
It may be about taking responsibility for tasks.
OpenAI’s Dots are designed to work continuously toward user goals, use connected tools, and return when human input is needed.
The technology is still new, and questions around reliability, privacy, security, and human oversight will remain important.
But the direction is significant.
AI is moving from something you talk to into something you can delegate work to.
And if always-on AI agents become reliable enough for everyday use, the way people work with computers could look very different over the next few years.
Frequently Asked Questions
What are OpenAI Dots?
Dots are OpenAI’s new always-on AI agents designed to work toward ongoing goals rather than simply responding to individual prompts.
How are Dots different from ChatGPT?
A regular ChatGPT conversation generally responds to your requests. Dots are designed to continue working toward assigned goals between conversations and can use connected tools and applications.
What model powers OpenAI Dots?
OpenAI says Dots are powered by GPT-6 Astra.
Can AI agents work while I am offline?
Dots are designed as always-on agents that can continue making progress between conversations, although specific capabilities depend on the available integrations, permissions, and account access.
Can OpenAI Dots use other applications?
Yes. OpenAI says Dots can connect with supported applications and services, including tools such as Slack and Teams.
Are AI agents safe?
AI agents introduce additional security and privacy considerations because they can access tools and perform actions. OpenAI says Dots include permission and safety controls, but users should still carefully review what an agent is allowed to access and do.