OpenAI has launched Dots, a new type of always-on AI agent designed to keep working toward user-defined goals even when the user is not actively interacting with ChatGPT.
Announced at OpenAI’s DevDay 2026, Dots are designed to move AI beyond the traditional question-and-answer chatbot model. They can maintain context, work across connected tools, perform multi-step tasks and return to users when a decision or approval is needed.
OpenAI says Dots are powered by GPT-6 Astra and operate using their own cloud computer and browser. Users can control which apps a Dot can access, what actions it can perform and when human approval is required.
What Are OpenAI Dots?
Dots are persistent AI agents designed to work continuously on projects rather than responding only to individual prompts.
A user can assign a Dot a responsibility, provide instructions and allow it to determine the next steps required to make progress.
For example, a Dot could:
- Research information.
- Analyze data.
- Prepare documents.
- Work on software.
- Monitor an ongoing project.
- Track changes across connected tools.
- Prepare updates for human review.
OpenAI describes the system as an extension of the user that can continue working between conversations.
Dots Run on GPT-6 Astra
OpenAI says Dots are powered by GPT-6 Astra, which the company describes as its most aligned model for the product.
The agents also use their own cloud computer and browser. This allows a Dot to perform work in a persistent environment rather than requiring the user’s personal computer to remain active.
The architecture is significant because it allows the agent to continue working after a user steps away from the conversation.
Dots Can Work Across Connected Apps
One of the key features of Dots is their ability to work with connected applications.
OpenAI’s product documentation shows integrations involving services such as:
- Google Calendar
- Google Drive
- GitHub
- Slack
- Microsoft Teams
- Other connected tools
The company says users decide which applications their Dot can access and can manage its permissions.
This allows a Dot to combine information from different tools instead of treating every task as an isolated conversation.
Always-On Does Not Mean Unlimited Access
Although OpenAI calls Dots “always-on,” users remain in control of their permissions and actions.
OpenAI says Dots include built-in safeguards and automatic review systems. Actions that could affect accounts or share information may require additional checks or user approval.
Users can also create custom rules specifying which actions should be allowed automatically, which should require approval and which should be blocked.
Some sensitive actions remain under direct user control. OpenAI gives changing a password as an example of an action that stays with the user.
Dots Can Continue Working After You Leave
A major difference between Dots and conventional chatbots is persistence.
Once a user assigns work, the Dot can track progress, determine what needs to happen next and continue working between conversations.
OpenAI says Dots can also use background agents to work on multiple tasks in parallel.
If a task reaches a point where human judgment is required, the Dot can return to the user with the relevant information or decision.
This approach changes the interaction model from:
User asks → AI answers
to:
User assigns responsibility → AI works → AI reports back → User reviews or decides
Dots Can Use Memory and User Context
OpenAI says Dots can use relevant ChatGPT memory, conversation context and their own saved notes about user preferences and ongoing work.
This allows an agent to maintain continuity across conversations and supported communication channels.
The feature could be useful for long-running projects where the user does not want to repeatedly explain the same background information.
However, the ability to retain context also makes permission and privacy controls particularly important.
Users Can Contact Dots Through Multiple Channels
OpenAI says users can communicate with the same Dot through ChatGPT and supported work communication platforms.
The product documentation lists:
- ChatGPT
- Slack
- Microsoft Teams
- Voice calls
The same Dot can continue its work even when users change where they communicate with it.
OpenAI says users can also configure where different types of updates should be delivered.
For example, routine progress could remain in ChatGPT while decisions requiring attention could be sent through another supported channel.
Dots Are Designed for Work and Personal Tasks
OpenAI is positioning Dots for both professional workflows and everyday tasks.
For work, examples include:
- Updating an investor presentation.
- Reviewing customer requirements.
- Tracking software migrations.
- Investigating bugs.
- Turning customer feedback into product changes.
- Preparing content from transcripts.
For personal use, OpenAI demonstrates scenarios such as coordinating schedules and preparing food-delivery options.
These examples show that OpenAI sees persistent agents as a broader computing interface rather than simply another chatbot feature.
Safety and Privacy Are Central to the Design
Always-on agents introduce additional safety considerations because they can access connected applications and potentially take actions on a user’s behalf.
OpenAI says Dots use built-in safeguards designed to defend against malicious instructions and monitor potentially harmful behavior.
The company also says Dots operate with permission controls and automated checks before sensitive actions.
These controls are particularly important when agents can interact with email, documents, calendars, code repositories or other systems containing private information.
Dots Are Rolling Out Gradually
OpenAI says Dots are rolling out gradually to eligible users.
According to OpenAI’s product documentation, access is available across Pro, Business Premium and Enterprise plans in eligible markets, with Enterprise access controlled by workspace administrators.
OpenAI also says Dots can be created using the desktop app or ChatGPT on a desktop browser. Mobile access is available after setup where the supporting update is available, while mobile web is not supported for creation.
Availability may therefore differ by plan, region and workspace settings.
Dots Push AI Toward Proactive Work
The launch represents a shift in how AI assistants are designed.
Traditional chatbots generally wait for users to send a prompt. Agentic systems can instead receive a broader goal and determine a sequence of actions.
Dots are designed around this latter model.
The user provides the objective and boundaries, while the agent handles intermediate steps and returns for human input when necessary.
This could make AI useful for longer-running workflows where completing the task involves research, analysis, software interaction and follow-up actions.
OpenAI Enters a Growing Always-On Agent Market
OpenAI’s launch comes as other technology companies are also developing persistent AI agents.
Reuters reported that Dots are part of OpenAI’s push toward autonomous AI systems and enterprise applications.
The wider industry is moving toward AI systems that can operate across multiple applications rather than remaining inside a single chatbot interface.
Competition is increasingly focused on:
- Agent autonomy.
- Model capability.
- App integrations.
- Security.
- User control.
- Long-running task execution.
What Dots Mean for the Future of AI Assistants
Dots demonstrate how AI assistants are evolving from conversational tools into persistent digital workers.
Instead of asking an AI system to complete every individual step, users can increasingly give it responsibility for an entire workflow.
The success of this model will depend on several factors, including reliability, security, permissions, accuracy and the ability to keep humans in control of important decisions.
For OpenAI, Dots also represent a significant expansion of ChatGPT’s role from an interactive assistant toward an environment where AI can continuously work on projects.
Final Takeaway
OpenAI has launched Dots, always-on AI agents designed to work continuously toward user-defined goals.
Powered by GPT-6 Astra, Dots can operate through their own cloud computer and browser, connect to selected applications, maintain relevant context and continue working between conversations.
OpenAI is also emphasizing permissions, automatic reviews and user approvals for sensitive actions.
The launch marks another step toward persistent AI agents that do more than answer prompts—while also making reliability, privacy and human control increasingly important parts of the AI assistant experience.
What are OpenAI Dots?
Dots are always-on AI agents designed to work continuously on user-assigned goals, including research, analysis, documents and software tasks.
What model powers OpenAI Dots?
OpenAI says Dots are powered by GPT-6 Astra.
Can Dots work when my computer is turned off?
Yes. OpenAI says a Dot operates in the cloud with its own computer and browser, allowing it to continue working when the user’s computer is off.
Can Dots access my apps?
Dots can access connected applications according to the permissions granted by the user. OpenAI lists integrations including Google Drive, Google Calendar, GitHub, Slack and Microsoft Teams.
Are OpenAI Dots fully autonomous?
Dots are designed to work independently toward assigned goals, but they remain subject to user permissions, safety controls and approval requirements. Some sensitive actions require user involvement.
Which ChatGPT plans have access to Dots?
OpenAI says Dots are rolling out to Pro, Business Premium and Enterprise users in eligible markets. Enterprise administrators must enable the feature for their workspaces.
Can I use Dots on mobile?
A Dot is created on desktop through the ChatGPT desktop app or desktop browser. After setup, users can interact with the same Dot through the mobile app where the supporting update is available.
Can Dots send messages or make changes without approval?
Some actions can be handled according to permissions and custom rules, while actions affecting accounts or sharing information may require review or user approval. Sensitive steps such as changing a password remain with the user.
Can Dots remember my preferences?
OpenAI says Dots can use relevant ChatGPT memory and their own saved notes about preferences, decisions and ongoing work.
Are Dots available to everyone?
No. OpenAI says the rollout is gradual and availability depends on plan, region and, for Enterprise users, workspace administration.
