- OpenAI Dots are always-on AI agents that keep working on tasks even when you log off.
- Each Dot runs on GPT-6 Astra with its own cloud computer and browser.
- Dots support 4,000+ plugins and can reach you with questions or progress updates.
- You control access with per-app rules, approvals, and a full activity view.
- First rollout targets Pro and Business Premium users; one Dot per account for now.
OpenAI Dots What Is: Core Concept
OpenAI Dots is a new agent product announced by OpenAI, turning ChatGPT from a chat window you babysit into a persistent assistant that works in the background. A Dot is not a single-use agent you monitor line by line. You hand it a project, close your laptop, and it keeps going on its own cloud computer.
Video Highlights:
- Dots keep working when you leave, powered by GPT-6 Astra
- Each Dot has its own cloud computer and browser
- Dots can notice tasks like a forgotten invoice and act after approval
- Access rolls out to Pro and Business Premium plans first
Under the hood, a Dot is powered by GPT-6 Astra and runs on dedicated cloud infrastructure. Because the workload lives in the cloud rather than your laptop, progress continues while your machine is asleep, locked, or switched off. You can check in on progress at any time, but repeated supervision is not the intended workflow.
Think of a Dot as a teammate, not a chatbot. You assign outcomes, and the Dot reports back, asks questions, or requests a decision when it genuinely needs one.
Key Features Breakdown
Dots ship with several headline capabilities that separate them from standard ChatGPT conversations. The table below summarizes the core pillars.
| Feature | What It Does | Why It Matters |
|---|---|---|
| Always-on work | Continues projects in the background | No babysitting or repeated corrections |
| Multi-project handling | Runs several projects without separate chats | One Dot, many parallel outcomes |
| Dot-to-Dot teamwork | Your Dot collaborates with a teammate's Dot | Shared goals across teams |
| Plugin access | Over 4,000 plugins available | Extends work and personal use cases |
| Preference learning | Learns goals, feedback, and definition of "good" | Fewer repeated instructions over time |
| Two-way contact | Message or call your Dot; it can reach you | Progress updates and decision requests |
Work Assistant
- Background projects
- Invoice preparation and follow-up
- Local file interaction
Personal Assistant
- 4,000+ plugins
- Notices forgotten tasks
- Personal life and work life support
Team Collaborator
- Dot-to-Dot coordination
- Works toward shared goals
- Future: multiple Dots per user
A standout example from the announcement: a Dot noticed a forgotten invoice, prepared it, and sent it after the user approved the action. That proactive pattern, spotting something you missed and handling it with a checkpoint, is the core promise of the product.
Once your Dot is live, assign one small recurring project and let it run overnight. Review the activity log the next morning to see how it handles autonomy before scaling up.
Security, Control, and Transparency
OpenAI put a strong emphasis on control with Dots, allowing you to define exactly what your Dot can touch. Security settings are granular, and every action is visible in an activity view.
Set per-app access rules
Choose which apps each Dot may access. Rules fall into three buckets: allow, require approval, or block.
Review the activity view
Inspect what your Dot is doing at any time. Spot unauthorized access attempts or off-track behavior early.
Rely on sensitive-action review
If an action could compromise your account or expose sensitive information, OpenAI's review system determines whether your explicit approval is required first.
Approve or redirect
When your Dot asks a question or requests a decision, respond in ChatGPT and it resumes work.
| Control Layer | Function | When It Triggers |
|---|---|---|
| App access rules | Allow / approve / block per app | Configured upfront |
| Activity view | Full transparency of Dot actions | Continuous |
| Review system | Flags risky or sensitive actions | Before execution |
| Approval prompts | You confirm or decline | On request |
For small and medium businesses without large IT departments, the activity view and approval rules are your main safety net. Audit them regularly rather than assuming defaults are right for your workflows.
Access, Plans, and Limitations
Availability is phased, and there are firm limits at launch worth knowing before you plan workflows around Dots.
| Plan / Workspace | Access Status |
|---|---|
| Pro | Rolling out, first Dot included |
| Business Premium | Rolling out, first Dot included |
| Enterprise | Beta available if workspace admin enables it |
| Education | Beta available if workspace admin enables it |
| Healthcare | Beta available if workspace admin enables it |
The first Dot is included with eligible plans alongside an allowance for deeper work. Two limitations stand out: you can only run one Dot per account today, and multi-Dot collaboration is a stated future direction rather than a current feature. Dot-to-Dot teamwork between different people's Dots, however, was demonstrated at launch.
Even Pro users reported waiting for the feature to appear on their accounts after announcement day. If you do not see Dots yet, it is a staged rollout, not an eligibility problem.
Getting Ready for Dots:
- Confirm your plan is Pro, Business Premium, or an admin-enabled workspace
- List apps and data your Dot should and should not access
- Define one low-risk starter project for day one
- Learn the activity view and approval flow
- Track announcements for multi-Dot support
Comparisons and FAQ
Dots enter a competitive field. Mark Zuckerberg has Meta's Muse, Elon Musk has Grok bots, and Sam Altman now has Dots. In regions like the UK, Muse is unavailable and Grok bots are the closest comparison point reviewers use, while lighter assistants are not considered a fair match.
| Product | Vendor | Closest Analogy |
|---|---|---|
| OpenAI Dots | OpenAI | Always-on cloud agent with plugins |
| Grok bots | xAI | Conversational agent counterpart |
| Muse | Meta | Region-limited assistant agent |
The differentiator is the always-on cloud model plus deep preference learning. Whether Dots outpace rivals depends on reliability in long-running tasks, so test with projects you can verify.
Q: What is OpenAI Dots exactly?
OpenAI Dots are always-on AI agents inside ChatGPT, powered by GPT-6 Astra. Each Dot has its own cloud computer and browser, so it keeps working on your projects in the background even when your laptop is off.
Q: Who can use Dots right now?
Rollout targets Pro and Business Premium users first, with the first Dot included alongside an allowance for deeper work. Enterprise, Education, and Healthcare workspaces can try a beta if their workspace administrator enables it.
Q: Can I run multiple Dots at once?
Not yet. At launch you can only have one Dot per account, and OpenAI describes multiple Dots working together as a future direction rather than a feature available today. Your Dot can, however, collaborate with teammates' Dots.
Q: How do I control what my Dot can access?
You set per-app rules that allow access, require approval, or block it. An activity view shows what your Dot is doing, and OpenAI's review system can require your approval before sensitive or account-risky actions execute.