What are ChatGPT Dots? How OpenAI?s always-on agents work
OpenAI?s new Dots are always-on GPT-6 Astra agents with their own cloud computer, cross-channel context and access to 4,000+ apps. Here?s what changes in practice.
OpenAI launched Dots on September 29 as always-on agents designed to keep working toward a user?s goals instead of waiting for a new prompt each time. A Dot is powered by GPT-6 Astra, runs on its own cloud computer, learns from feedback over time and can connect to more than 4,000 apps through OpenAI?s plugin ecosystem.
What a Dot actually is
The biggest change is persistence. You can talk to a Dot in ChatGPT, Slack or Teams, and OpenAI says the agent carries context across those channels. It can also send progress updates, ask questions and surface decisions that need you. That makes Dots closer to persistent AI coworkers than a normal one-off chat session.
The launch in numbers
How Dots differ from normal ChatGPT
Chat session vs always-on agent
| Normal ChatGPT interaction | A Dot |
|---|---|
| Usually starts when you open a conversation and ask | Can continue working in the background toward an ongoing goal |
| Context is primarily tied to the ChatGPT experience | Carries working context across ChatGPT, Slack and Teams |
| You typically initiate the next step | Can proactively research and send progress, questions or decisions |
| Tools help complete the current task | Its own cloud computer and connected apps support longer-running work |
The important shift is not simply that Dots can use more tools. It is that OpenAI is packaging memory, a cloud computer, app access, proactive background work and communication channels into one persistent agent. For business workflows, that could move AI from a tool employees repeatedly invoke to an agent that owns a continuing objective. The useful test is whether the objective benefits from ongoing context and judgment; deterministic, repeatable processes can still be better handled by conventional automation.
Always-on does not mean unrestricted
OpenAI says each Dot runs on a separate cloud computer, leaving the user?s own computer and files separate unless they are connected. During proactive research, connected-app tools are restricted to read-only actions: the Dot cannot send messages, change app content or control the user?s browser or computer in that background mode.
Dots have built-in rules for when they may act independently and when they need approval. Custom Rules can allow, require approval for or block specific actions, while OpenAI?s built-in safety requirements continue to apply. Activity View lets users follow background work and redirect the agent.
Where a Dot could be genuinely useful
- A sales or account agent that follows ongoing customer context across connected tools and surfaces what needs attention.
- A research agent that keeps an eye on a market or project and brings back useful findings instead of waiting for a fresh research prompt.
- A project agent that carries context between ChatGPT and team conversations in Slack or Teams.
- A personal work agent that remembers preferences and prepares recurring context, drafts or decisions for review.
Who can use Dots now
Dots are rolling out gradually to eligible ChatGPT Pro and Business Premium users, while Enterprise users can try the beta when a workspace admin enables it. OpenAI says the first Dot is included at no extra cost for eligible Pro and Business Premium plans. Setup currently starts on desktop web or the ChatGPT desktop app; after setup, mobile access can be used as it becomes available.
OpenAI?s Help Center says rollout can take several days to reach an eligible account. It also says that, for the first month after launch, Dots usage will not count toward eligible Pro, Business and Enterprise plan allowances; later usage terms are still to be announced.
What to test before giving a Dot real responsibility
- Start with a continuing objective where background research or cross-app context is actually valuable.
- Connect only the apps the Dot needs and review their permissions.
- Define Custom Rules for actions that should always require approval.
- Use Activity View to inspect how it reaches conclusions and when it asks for help.
- Keep irreversible or customer-facing actions human-approved until the workflow is proven.
Dots turn ChatGPT?s agent capabilities into a persistent, always-available worker with its own cloud computer, connected apps and cross-channel context. The opportunity is ongoing delegated work, not merely faster prompting. The constraint is equally important: permissions, approval rules and human review still determine how much responsibility a Dot should actually receive.
Sources & useful resources
- OpenAI: Introducing dots— Primary launch announcement, Sep. 29, 2026
- OpenAI Help: Getting started with your dot— Official setup, rollout and usage guidance