OpenAI Dots: ChatGPT AI Agent Setup Guide & Tips - ChatGPT

OpenAI Dots: ChatGPT AI Agent Setup Guide & Tips

Learn what OpenAI Dots is, how the always-on ChatGPT agent works, key features, access plans, and a step-by-step setup guide.

2026-09-30
OpenAI Dots Wiki Team
Quick Guide
  • OpenAI Dots is an always-on AI agent for ChatGPT that keeps working in the background.
  • Each dot runs on its own cloud computer and browser, even when your laptop is asleep.
  • Dots support 4,000+ plugins and learn your preferences over time.
  • Access rolls out to Pro and Business Premium users first; one dot per account for now.
  • Strong permission controls and activity view let you audit everything your dot does.

What Is OpenAI Dots? The New ChatGPT AI Agent

OpenAI Dots is the company's newly announced always-on assistant built into the ChatGPT ecosystem. Unlike a standard chat session that stops the moment you close the tab, a dot is a persistent agent: you hand it a project, and it continues working in the background without constant monitoring or repeated corrections.

Video Highlights:

  • Dots keep working after you leave, powered by GPT-6 Astra under the hood
  • Each dot has its own cloud computer and browser
  • Multiple projects can run through one dot, no separate chats required
  • Teammate dots can collaborate to achieve shared goals
  • Rollout begins with Pro and Business Premium users

The headline concept is simple: an agent you delegate to rather than one you babysit. In OpenAI's demo, one person's dot and a teammate's dot worked together toward a shared goal, which points toward a future of agent-to-agent collaboration in the workplace.

Key Distinction

A dot is not a chatbot you prompt repeatedly. It is a persistent worker that accepts projects, shares progress proactively, and asks you questions only when it needs a decision.

Core Features Explained

Dots arrive with a focused set of capabilities designed around autonomy, integration, and personalization.

Always-On Operation

  • Runs on its own cloud computer
  • Works while your laptop sleeps or is off
  • Handles multiple projects at once
  • Can interact with local files

Plugin Ecosystem

  • Access to 4,000+ plugins
  • Covers personal and work use cases
  • Example: noticing a forgotten invoice, preparing it, sending after approval

Preference Learning

  • Learns from your feedback over time
  • Remembers goals and what a good result looks like
  • Reduces back-and-forth corrections

Communication flows both ways. You can message or call your dot inside ChatGPT, and the dot can reach out to you to share progress, ask a question, or request a decision.

Feature Overview Table

FeatureWhat It DoesWhy It Matters
Background workContinues projects when you log offTrue delegation, not babysitting
Cloud computer + browserRuns independently of your deviceNo uptime dependency on your laptop
Multi-project handlingOne dot, several projects at onceNo need to manage separate chats
Plugin access4,000+ integrationsExtends usefulness across work and life
Two-way messagingDot can contact you firstProactive updates and decision requests
Preference learningAdapts to your feedback and goalsBetter results with less prompting
Under the Hood

Dots are powered by GPT-6 Astra, each with a dedicated cloud environment. This is what separates them from standard ChatGPT conversations that end when you close the window.

Security, Permissions, and Control

OpenAI placed transparency at the center of the Dots announcement, which matters for small and medium businesses without large IT departments monitoring every tool.

Permission Levels

SettingBehaviorBest Used For
AllowDot accesses the app freelyTrusted, low-risk tools
Require approvalDot asks before each actionSensitive apps like email or payments
BlockAccess denied entirelyAnything off-limits

The activity view shows what your dot is doing in real time, including attempts to access apps you did not authorize. On top of that, a review system checks whether potentially risky actions, such as sharing sensitive information or actions that could compromise your account, require your explicit approval first.

Set Permissions Early

Review the default app permissions the moment your dot is enabled. Lock down anything sensitive before delegating projects, and check the activity view during your first few sessions.

Step-by-Step: Getting Started with Your First Dot

1

Confirm Your Plan Eligibility

Check whether your account is Pro or Business Premium. Enterprise, Education, and Healthcare users can try the beta only if their workspace administrator enables it.

2

Wait for Rollout and Create Your Dot

Access rolls out gradually even on eligible plans. Once the feature appears on your account, create your first dot. Note that you can currently run only one dot per account.

3

Configure App Permissions

Set rules for each connected app: allow, require approval, or block. Apply the strictest setting to anything involving payments, credentials, or sensitive data.

4

Assign Your First Project

Give the dot a clear, well-scoped task with a defined definition of done. Start small, such as monitoring an inbox or preparing a recurring document.

5

Review, Give Feedback, Repeat

Check the activity view, approve or reject pending actions, and give feedback. The dot learns your preferences from this loop and needs less oversight over time.

First-Week Strategy

Use your first week to calibrate. Approve actions deliberately, correct mistakes explicitly, and let preference learning do the heavy lifting before you hand over larger projects.

Access Plans and Rollout Status

Plan Availability Table

PlanAccess StatusNotes
ProRolling outFirst dot included with eligible plans
Business PremiumRolling outSame rollout tier as Pro
EnterpriseBeta availableRequires workspace admin opt-in
EducationBeta availableRequires workspace admin opt-in
HealthcareBeta availableRequires workspace admin opt-in

OpenAI states the first dot is included with eligible plans, alongside an allowance for deeper work. Multiple dots working together is a stated future direction, not a feature available today.

Before You Delegate Real Work:

  • Verify your plan is Pro, Business Premium, or beta-enabled
  • Create your first dot (one dot per account for now)
  • Set app permissions to allow, approve, or block
  • Review the activity view during early sessions
  • Give regular feedback so the dot learns your preferences

FAQ

Q: What is OpenAI Dots in ChatGPT?

OpenAI Dots are always-on AI agents inside ChatGPT. Each dot runs on its own cloud computer and browser, powered by GPT-6 Astra, and continues working on your projects in the background even when your device is off.

Q: Which plans get access to Dots?

Rollout begins with Pro and Business Premium users. Enterprise, Education, and Healthcare users can try a beta if their workspace administrator enables it. Access is gradual, so not every eligible account has it immediately.

Q: Can I run multiple dots at once?

No. For now, each account can have one dot, with the first dot included on eligible plans. OpenAI has said multiple dots working together is a future direction, but it is not available today.

Q: How much control do I have over what my dot does?

Substantial control. You set per-app rules (allow, require approval, or block), monitor everything through an activity view, and a review system flags sensitive or account-risking actions for your explicit approval.