What Are ChatGPT Dots? A Simple Guide for Normal People

The problem

Normal ChatGPT usually works like this: you ask something, ChatGPT answers, and the conversation stops until you come back. A Dot is different. You give it an ongoing job or goal, and it can continue working toward that goal between conversations, using the tools and connected apps you allow it to use. Then it can bring the results back for you to review. Think of normal ChatGPT as someone you ask for help when you need it, and a Dot more like an AI assistant with an ongoing responsibility. For example, with normal ChatGPT you might say "Summarize these five customer reviews." With a Dot, you might say "Keep helping me track customer feedback and surface the important patterns I should look at." OpenAI announced Dots on September 29, 2026, describing them as always-on agents powered by GPT-6 Astra that can take on ongoing work.

What you need

• A ChatGPT account that has access to Dots. They are NOT necessarily available to every ChatGPT user yet.
• As of September 30, 2026, Dots are rolling out gradually. Eligible Pro and Business Premium users are among the first to get them, and you must be 18 or older.
• Availability depends on supported markets. For example, OpenAI says the Pro rollout currently excludes the European Economic Area, Switzerland and the UK.
• Enterprise access is currently a beta and depends on workspace settings. It is turned off until a workspace admin enables it.
• Eligible users create a Dot in the ChatGPT desktop app or ChatGPT on desktop web, not on mobile.
• Because this is a new feature, availability can change quickly.

Safety first

START SMALL. Do not give a new Dot broad permissions just because it can use them. For your first Dot, choose one clear responsibility, connect only the apps it genuinely needs, start with reading, research and preparation work where practical, review what it produces, and expand permissions only after you understand how it behaves. Before connecting any app, ask yourself three questions: 1. Does the Dot actually need this app to do its job? 2. What information will the Dot be able to access? 3. What actions am I comfortable allowing it to perform? If you cannot answer those questions, do not connect the app yet. For your first experiment, avoid giving a Dot authority to send money, approve purchases, delete important files, make irreversible account changes, send sensitive communications without review, make legal, financial or medical decisions, or freely access information it does not need. Also avoid giving unnecessary access to financial accounts, confidential client information, sensitive personal records, important business systems, or anything where a mistaken action would be difficult to undo. The goal is useful assistance, not maximum autonomy, and human review still matters.

Steps

1. Understand the big difference: a Dot can keep working. NORMAL CHATGPT You: ask for something. AI: responds now. Good for: one-off tasks. DOT You: give it an ongoing goal. AI: can continue making progress between conversations, within the permissions and tools you provide. Good for: work that continues over time. This does NOT mean a Dot should make every decision for you, has unlimited access, or should be trusted without review.
2. Know what "its own cloud computer" means. OpenAI says each Dot has its own cloud computer: a computer environment in the cloud that it uses for supported work. It does NOT mean the Dot automatically controls your personal laptop or sees everything on it.
3. Know how connected apps work. Dots become more useful when you intentionally connect supported apps through OpenAI's plugin ecosystem, which OpenAI says reaches thousands of apps. Examples might include email, calendars, documents, work tools and communication tools, and depending on supported integrations you can talk to your Dot through ChatGPT, Slack or Teams. Connecting ChatGPT does not automatically grant access to everything. The Dot should only have access to the tools and information needed for the job you gave it, and you can configure what it may do on its own.
4. Paste the starter prompt into ChatGPT and describe your recurring job. It helps you plan a safe first version before you connect anything. For example: USER GOAL: "I want help keeping track of customer complaints." BAD FIRST VERSION: "Read everything, contact customers and fix problems automatically." BETTER FIRST VERSION: "Review approved feedback sources, group recurring complaints, and prepare a weekly summary for me to review." The better version gives the AI a useful job without immediately handing over important decisions: OBSERVE → PREPARE → REVIEW.
5. Pick one simple first job. Good first Dot ideas include: 1. CUSTOMER FEEDBACK WATCHER: surface recurring customer complaints and compliments. 2. MEETING FOLLOW-UP HELPER: keep track of action items from approved meeting information. 3. RESEARCH HELPER: continue researching an ongoing topic and bring useful findings back. 4. PROJECT UPDATE HELPER: collect approved project information and prepare a progress summary. 5. WEEKLY BUSINESS BRIEF: gather information from approved sources and prepare a weekly brief for review. These are examples of how an ongoing agent can be used. What works for you depends on the apps and plugins available to your account and the permissions you provide.
6. Review what the Dot brings back before acting on it. Only expand its access or let it do more on its own once you understand how it behaves.

What success looks like

Before: "ChatGPT Dots" sounds like another confusing AI buzzword. After: you understand what a Dot is, how it differs from a normal chat, why ongoing work is different, what connected apps mean, why permissions matter, how to start safely, and whether you currently have access. Current as of September 30, 2026. ChatGPT features, plans and availability can change quickly, so for current availability check OpenAI's official ChatGPT release notes and the "Introducing dots" announcement.

If it goes wrong

Don't see Dots yet? That does not necessarily mean anything is wrong. OpenAI is rolling Dots out gradually, and availability depends on your plan, region and workspace settings. Do not change your country, use a VPN, or try to bypass availability restrictions. Check your ChatGPT account for current availability, and read OpenAI's official ChatGPT release notes for the latest information.