OpenAI has introduced Dots, persistent AI agents that can continue working on goals between conversations. Powered by GPT-6 Astra, each Dot has its own cloud computer and browser, can use apps a user connects, and can handle recurring tasks. OpenAI says Dots can also delegate work to other agents. The initial rollout is limited to eligible paid plans and is gradual, so access may not appear for every account immediately. 

For businesses, the change is significant: an AI tool can keep track of a continuing workflow, gather new information and prepare follow-up work over time. That makes Dots potentially useful for research, reporting, content preparation and project support. It also makes clear permissions and human review more important. 

What Are OpenAI Dots? 

A Dot is an always-on agent in ChatGPT. A user gives it a goal and defines what it may do independently. It can make progress in the background, remember relevant context, and return with updates, questions or work for review. 

OpenAI says Dots can connect to more than 4,000 apps through its plugin ecosystem. Users can interact with them through ChatGPT, Slack and Microsoft Teams. Dots can use connected apps and their cloud computer, and can delegate parts of a task to subagents. App availability and permissions vary. 

Examples include investigating a reported software bug, updating launch materials as plans change, revising an analysis when new data arrives, and preparing content drafts from an interview transcript. These are product examples, not independent measurements of real-world performance. 

How Are Dots Different From a Regular ChatGPT Agent? 

A typical chat interaction responds to a prompt and ends when that task is complete. Dots are designed to retain context and continue working on an ongoing responsibility, including scheduled checks and background research. 

Persistent and multi-agent systems are not new. The product change is packaging persistence, a cloud computer, app connections and recurring work into ChatGPT. 

What Can a Dot Do? 

A Dot’s practical abilities depend on the apps connected, permissions granted and task instructions. Potential business workflows include: 

• Lead follow-up: Review new enquiries, summarize needs and draft a response for a team member to approve. 

• Campaign reporting: Gather campaign information from connected sources and prepare a recurring report. 

• Content operations: Turn approved source material into draft newsletters, social posts or content briefs. 

• Website quality checks: Review specified pages and prepare a list of issues for a developer. 

• Sales research: Collect information about a prospect and prepare a proposal or meeting brief. 

• Support triage: Organize incoming support issues and suggest responses for review. 

These are possible workflow applications, not a promise that every task will run reliably or be available through every connector. Users can schedule recurring checks, review active, paused and completed work, pause a Dot or redirect it as work develops. 

OpenAI Dots Plans and Availability 

As of 30 September 2026, OpenAI says Dots are rolling out to: 

• Pro users in supported markets, excluding the European Economic Area, Switzerland and the United Kingdom. 

• Business Premium users across supported ChatGPT regions. 

• Enterprise, Education and Healthcare users through a beta that a workspace administrator must enable. Enterprise access is off by default. 

OpenAI says the first Dot is included at no additional cost with Pro or Business Premium. The plan includes an allowance for deeper work, with extended limits for the first month after launch. OpenAI says it will share the usage terms for each plan after that initial period. Dots are rolling out gradually, so access may take several days to reach an eligible account. 

There is no separately stated Dots price in the current launch guidance. The subscription price depends on the applicable ChatGPT plan and market. OpenAI also says Dot conversations do not count toward ChatGPT usage limits, but tasks started or managed in Codex or ChatGPT Work count toward those products’ limits as usual. Check the plan and usage terms shown in your own account before subscribing or budgeting for a team. 

Are OpenAI Dots Available in India? 

India is listed by OpenAI as a supported ChatGPT country. OpenAI’s Dots guidance describes Pro availability in supported markets, with exclusions for the EEA, Switzerland and the UK. This indicates India is within the relevant market scope, but it does not guarantee that Dots have reached every Indian account. The rollout is gradual; Business Premium and workspace beta access also depend on the plan and workspace settings. Check the ChatGPT desktop app or desktop website while signed in to the eligible account. 

OpenAI says users must create a Dot on desktop web or in the desktop app. After setup, mobile ChatGPT access is available, but Dot creation is not currently supported on mobile, and Dots are not supported on mobile web. Texting is a limited beta for some Pro users in the United States, not a general India feature. 

OpenAI Dots Safety: What the Tests Show 

Persistent agents encounter new emails, documents and other information after their original task begins. Some material can contain malicious instructions designed to make an agent reveal information or take an unauthorized action. This is commonly called prompt injection. 

OpenAI’s launch safety document reports the following results from internal tests: 

OpenAI says its open-ended evaluations are preliminary and will be expanded. It also says human red-teamers probed Dots with malicious third-party content and that the company changed confirmation policies in response to findings. These are company-reported test results, not an independent certification or proof that a Dot cannot make a mistake. 

Dots include several controls. Users choose which apps to connect, can set Custom Rules, and can review activity or ask a Dot to stop. OpenAI says Auto-review checks certain planned actions, while some sensitive actions—such as changing a password or transferring money—must be completed by the user. OpenAI also warns that Dots can make mistakes and that some actions cannot be reversed. 

Privacy and Business Data 

OpenAI says it does not use Business, Enterprise or Education workspace content to train its models by default. For personal plans, the “Improve the model for everyone” setting controls whether eligible Dot conversations and work may be used to improve models. OpenAI also says background research notes are not used directly for training, though information brought from that research into an eligible conversation or task may be used depending on the user’s settings. 

Disconnecting an app stops new access through that connection, but it does not automatically erase information the Dot has already incorporated into its context. OpenAI says deleting a Dot deletes its own context, while files and conversations created elsewhere are stored separately. 

What the Early Numbers Say About the AI-Work Trend 

OpenAI’s DevDay recap says ChatGPT has 1.2 billion weekly users as a platform for new experiences. Separately, OpenAI said Codex and ChatGPT Work together had more than 35 million weekly users. A June comparison of about 5 million Codex users covers Codex alone, so those figures are not a like-for-like growth comparison. They show the scale OpenAI is building for AI-assisted work, not Dots adoption: OpenAI has not yet published a Dot user count. 

The product trend is moving from one-off prompts toward systems that can carry responsibility across time, tools and changing information. That shift creates a governance challenge alongside the productivity opportunity: the longer an agent operates and the more connected services it can access, the more important it becomes to define its scope, review its work and know how to stop it. 

OpenAI Dots vs n8n Automation 

For a defined task such as moving a form submission into a CRM and sending a confirmation, a conventional workflow may be easier to control. For work that needs to interpret varied information and prepare a response, an agent may be useful—provided a person checks the result and external actions have appropriate approval. 

How Indian Businesses Can Prepare for Persistent AI Agents 

Agencies, startups and service businesses in India can start by selecting one low-risk, measurable workflow. Examples include preparing a weekly campaign report, summarizing incoming leads, or drafting content from already approved material. 

• Define the task: What recurring job should it handle, and what counts as completion? 

• Set boundaries: Which actions are prohibited, and what should trigger escalation? 

• Limit data access: Connect only the apps, folders and accounts needed. 

• Require approvals: Review before sending messages, publishing, purchasing or changing important records. 

• Measure the pilot: Record time saved, corrections required, missed items and interventions. 

• Plan how to stop it: Decide who can pause the Dot, revoke app access and review retained context or files. 

Start with read-only access where possible. Do not use an early pilot for unsupervised payments, public publishing, customer commitments or changes to critical business records. 

Frequently Asked Questions