OpenAI Dots Review: Features, Pricing, Privacy, and Verdict

This OpenAI Dots review examines an always-on ChatGPT agent that can keep working between conversations. A dot has its own cloud computer and browser, can use connected apps, remembers ongoing context, and returns to the user for approvals or decisions. It is powered by GPT-6 Astra and currently targets higher-priced personal and business plans.

The promise and risk are broader than one chat: a persistent agent can read connected information, create files, use tools, and take approved actions. This article evaluates current documentation and independent reporting rather than claiming a hands-on test. Details were checked on October 12, 2026.

OpenAI Dots Review: Quick Verdict

OpenAI Dots is a serious attempt to turn ChatGPT into an ongoing collaborator. Its strongest ideas are persistent context, a separate cloud computer, connected apps, recurring work, proactive read-only research, Custom Rules, Activity View, and a second safety check called Auto-review.

It is not a replacement for human ownership. Dots can misunderstand instructions, retain context from disconnected apps, and make changes that are hard to reverse. The rollout also focuses on ChatGPT Pro and Business Premium, with an admin-enabled beta for Enterprise workspaces. Dots is worth a controlled pilot for ongoing work with clear review points, but not for unrestricted sensitive data or poorly defined approvals.

What Are OpenAI Dots?

OpenAI Dots are persistent agents inside ChatGPT. The current product starts with one nameable primary dot. Specialist dots for organizations remain focused enterprise pilots.

A dot can work in its own cloud environment, browse the web, create files, run tools, and use apps the user has connected. Optional local-computer access lets it create ChatGPT Work or Codex tasks, use local skills, reach files, and use a local browser when the cloud browser is blocked. Local access starts off and must be explicitly enabled.

The distinction is continuity. A dot can retain project responsibility, remember relevant context, schedule future work, and make progress while the user is away.

Dots use GPT-6 Astra for reasoning and delegation. SD's GPT-6 Astra review covers the model's strengths, cost, and safety evidence in more detail.

OpenAI Dots Features That Matter

Cloud Computer, Browser, and Connected Apps

Every dot receives an isolated cloud computer and browser. Users can inspect it, take over when needed, and return control afterward. Supported secure sign-in keeps saved passwords outside the model context, although secrets in readable files or messages may still be visible.

Plugin permissions are shared across Dots, ChatGPT, ChatGPT Work, and Codex. Review every connected account and its read and write scopes before starting long-running work.

Persistent Context, Memory, and Channels

A dot receives ChatGPT memories and recent context and can form memories from connected apps. Users can create and reach it through the ChatGPT mobile app, desktop app, or desktop web, and can message it in Slack and Teams. Users can call their dot; calls initiated by the dot are planned for after launch.

Turning off ChatGPT Memory stops future sharing but does not delete context already received. Disconnecting an app stops new access but does not remove information already incorporated into the dot's context.

Scheduled Tasks and Proactive Research

Dots can run reminders, recurring checks, and proactive research without a new message. OpenAI says proactive research tools are enforced as read-only: they cannot directly send messages, modify connected-app content, or control a browser or computer.

Later actions are governed separately by app permissions, Custom Rules, approvals, and Auto-review.

Custom Rules, Activity View, and Auto-Review

Custom Rules let users allow, pre-approve, confirm, or hand back supported actions. Activity View shows ongoing and delegated work so the user can redirect or stop it.

Auto-review separately checks certain planned actions, including emails and file changes, against instructions, Custom Rules, and safety requirements. Core restrictions cannot be disabled or overridden. OpenAI still warns that Dots can make mistakes.

For a broader view of projects, permissions, and agent workflows inside the same ecosystem, see SD's ChatGPT Work review.

OpenAI Dots Pricing and Availability

The first dot is included at no additional charge with eligible ChatGPT Pro or Business Premium access. OpenAI's current personal Pro plans start at $100 per month. A Business Premium seat costs $100 per user per month with annual billing or $125 with monthly billing, and a self-serve Business workspace requires at least two paid seats. Enterprise pricing is custom.

The plan includes an allowance for deeper work, with temporarily extended launch limits. Dot conversations do not count against standard ChatGPT usage limits, but Work or Codex tasks use their normal allowances. OpenAI has not published permanent pricing for additional dots or higher capacity.

Personal Pro access excludes the European Economic Area, Switzerland, and the United Kingdom at launch. Business Premium is rolling out worldwide. Enterprise workspaces require an administrator to enable Dots, which is off by default. Rollout is gradual.

OpenAI Dots Privacy and Security

OpenAI says each dot's cloud environment is isolated and sandboxed, while supported sign-in keeps credentials outside model context. Plugin permissions, Custom Rules, confirmations, monitoring, and Auto-review add layers around actions.

A dot may retain context from conversations and plugins for as long as it exists. Users cannot inspect or delete individual dot memories. Deleting the dot removes its context, while files, tasks, conversations, and shared ChatGPT memories may require separate cleanup.

Business, Enterprise, and Edu content is not used for training by default. On personal plans, the “Improve the model for everyone” setting controls eligible Dots conversations and work. Proactive research is not used directly, but information moved into an eligible task may be used depending on that setting.

Prompt injection remains a central risk because a dot reads websites, email, documents, and connected data that may contain malicious instructions. OpenAI describes model safeguards, tool restrictions, isolated execution, action checks, and monitoring, but none makes consequential work error-free. SD's AI agent security guide provides a practical least-privilege and approval framework for this category.

OpenAI Dots Limitations and Risks

The largest limitation is evidence. OpenAI's examples are demonstrations and early scenarios, not independent success rates. Long-term reliability data is unavailable.

Other practical limits include:

• only one primary dot is broadly available today;
• specialist dots and deep enterprise integrations are still pilots;
• local access requires the connected computer to stay online with the desktop app open;
• cloud browser sessions are separate, and some sites may block them;
• mobile web remains unsupported;
• individual dot memories cannot currently be inspected or deleted one by one;
• some completed actions cannot be reversed.

Useful deployments need a bounded goal, approved sources, explicit recipients, permitted actions, escalation conditions, and a review cadence. Start with reversible work and expand access only after reliable results on representative tasks.

OpenAI Dots vs Meta Muse and Other Alternatives

The current SERP often compares OpenAI Dots with Meta Muse because both use persistent cloud computers and background work. Dots is more tightly integrated with ChatGPT, GPT-6 Astra, plugins, Work, and Codex. Muse is positioned more broadly as a personal agent and offers a free starting tier. SD's Meta Muse review covers its tradeoffs.

ChatGPT Work is a closer alternative for a defined, observable project. Conventional automation platforms remain better for deterministic triggers and actions. A human checklist may be safer for rare, high-impact work.

Choose the smallest level of autonomy that solves the problem. Persistence is valuable only when continuity, monitoring, and approved follow-through matter more than the additional security and governance burden.

Who Should Use OpenAI Dots?

Dots fits recurring, multi-step work: tracking feedback, monitoring evidence, maintaining connected assets, triaging issues, and coordinating systems. The strongest use cases produce reviewable drafts, analyses, code changes, or recommendations before external action.

It is weaker for low-cost use, organizations that prohibit hosted agents, or unrestricted sensitive data. Every workflow still needs a human owner, decision records, and a quick way to stop access.

A sensible pilot connects few apps, enables multi-factor authentication, confirms irreversible actions, and tests representative tasks. Measure completion, errors, review time, unintended disclosures, recovery effort, and total cost.

Conclusion

This OpenAI Dots review finds a promising but high-responsibility product. Persistent context, connected apps, scheduled work, proactive research, and cross-channel access could reduce the coordination burden of changing projects. Custom Rules, Activity View, isolated execution, and Auto-review address some risks of ongoing authority.

Adopt it conservatively. Start with reversible work, grant minimal access, specify approvals and recipients, and measure accepted outcomes. An always-on agent is useful when it earns trust through controlled results, not when it merely stays busy.