Dots

Dots are OpenAI's always-on agents, announced on 29 September 2026 at the company's largest DevDay and rolling out the same day to ChatGPT Pro and Business Premium users in eligible markets. A dot is not a chat session; it is a persistent worker with its own cloud computer and browser, powered by GPT-6 Astra, that you can message, call, or hand a project to. Each dot connects to more than 4,000 apps through OpenAI's plugin ecosystem, learns from the feedback you give it, and keeps working while you are asleep. The first dot is included in your plan at no extra cost, and conversations with it do not draw down your ChatGPT usage limits, though tasks you hand to Codex or ChatGPT Work still count normally.

Key Features

• Its own cloud computer and browser: You can open the machine and watch work in progress, and the dot can also be granted access to your laptop when you want help in place.
• 24/7 background work: Dots take on long-running projects and run several at once without you juggling threads.
• 4,000+ app connections: Through OpenAI's plugin ecosystem, a dot can reach inboxes, issue trackers, document stores, and the rest of your stack.
• Talk to it anywhere: Message or call a dot in ChatGPT on desktop, web, and mobile, or in Slack and Teams, with texting announced as coming soon. Context follows the dot across channels.
• Proactive research: When idle, a dot looks for ways to help using read-only access to your connected apps, so it cannot send messages or change content on its own.
• Custom Rules and Activity View: Decide which actions a dot may take, which need approval, and which are blocked, then follow its progress in one place.
• Specialist dots: For organizations, dots get their own identity and credentials for defined responsibilities, with pilots starting in procurement, invoicing, email marketing, support, and contracting.

Use Cases

Who Should Use This Tool?

• Knowledge workers on Pro or Business Premium: The common case is delegating a recurring multi-app chore rather than asking another question in a chat window.
• Small teams without a platform group: A dot inside Slack gives a shared teammate without anyone provisioning infrastructure.
• Anyone who starts work in the evening: Hand off the task before you close the laptop and review the result in the morning.

Problems It Solves

1. Chat does not finish anything: A dot keeps a task moving after the conversation ends, including across apps you would otherwise coordinate by hand.
2. Context scattered across tools: One agent holds context across ChatGPT, Slack, and Teams, so you stop re-explaining the project.
3. Credential anxiety: Saved-password sign-in is handled without exposing secrets to the model, and connected apps can stay read-only by default.

Pricing

The first dot is included with no additional charge, and OpenAI says launch pricing includes extended limits for the first month. Dots themselves do not consume your ChatGPT usage limits; tasks routed into Codex or ChatGPT Work do. Check the official page for current tier detail before you plan a budget around it.

Advantages & Unique Selling Points

Compared to Competitors:

1. Bundled instead of extra: Competing persistent-agent products generally require a separate subscription; here the first dot rides on a plan many people already pay for.
2. Distribution advantage: Arriving inside ChatGPT means the dot is already where your conversations, files, and connectors live.
3. Voice-first access: You can call a dot and redirect it by voice, which matters for long-running work you want to steer without sitting at a keyboard.

What Makes It Stand Out:

• Written conversational allowances that do not draw down your normal limits.
• Read-only proactive research that limits what an idle agent can do.
• Approval rules that can be tuned per action class rather than globally.

User Reviews

Early reactions split along a familiar line. People who tried it immediately appreciated the inspectable cloud desktop, with one user describing asking a dot to review a pull request, watching it install Rust, clone the branch, and start the review. Others raised the real concern: the entry price is a Pro subscription, and the security story for an agent that can sign into your accounts is still being written. OpenAI's own docs acknowledge that dots make mistakes and should have consequential work reviewed. One widely shared joke about "open your browser and type dot.com" is worth ignoring: the product lives at OpenAI's blog and inside ChatGPT, not at that address.

Getting Started

Quick Start Guide

1. Check eligibility: Confirm your plan and market, then install or update the ChatGPT desktop app.
2. Create your first dot: Open ChatGPT on desktop and give the dot a name.
3. Connect apps deliberately: Add only the tools the dot needs, and start with read-only where possible.
4. Set Custom Rules: Decide up front which actions are allowed, which require approval, and which are blocked.
5. Hand off a real task: Start with something small and verifiable, such as preparing an invoice draft for your approval.

Integration

• ChatGPT on desktop, web, and mobile, plus Slack and Teams.
• Connections through OpenAI's app ecosystem for the tools already in your stack.
• Handoffs into Codex and ChatGPT Work when a task needs heavier engineering.
• Voice calls for steering work in progress.

Frequently Asked Questions

How is a dot different from ChatGPT or Codex?

ChatGPT answers questions, Codex writes and reviews code, and a dot is a persistent worker with its own computer that runs projects across apps 24/7.

Does it consume my ChatGPT usage limits?

Conversations with your dot do not count toward your ChatGPT limits. Tasks you ask it to start or manage in Codex or ChatGPT Work do count as usual.

Which plans include dots?

Dots rolled out to Pro and Business Premium in eligible markets at launch, with Enterprise, Edu, and Healthcare able to try the beta once a workspace administrator enables it.

Can I add more than one dot?

At launch you get your primary dot. OpenAI says additional dots and per-dot scaling will come later.

What are the risks?

Dots can sign into supported websites and act on your behalf, so scope their app access, keep approval requirements on anything consequential, and review the output before you ship it.

Alternatives

If Dots is not the right fit, consider these alternatives:

• Claude Cowork: Anthropic's file-centric coworker for people who want the agent close to their documents.
• Grok Bot: SpaceXAI's always-on teammates with their own cloud computers, bundled with SuperGrok Heavy and Cursor Ultra.
• Manus: A general autonomous agent platform with a longer track record.
• Buzz: An open-source, self-hosted workspace where agents sit alongside your team.

Tips & Best Practices

1. Start read-only: Prove the dot's judgement on research before you let it write to any account.
2. Keep approval on money and messages: Outbound email and payments are the two places an eager agent costs the most.
3. Name the dot after the job: A dot called "invoices" is easier to audit than one called "assistant".
4. Review the cloud desktop: Opening its machine is the fastest way to see whether a stalled task is stuck on a login or a decision.

Conclusion

Dots push ChatGPT from answering questions to owning work, and the fact that the first one is included in an existing subscription is a bigger deal than any single feature. If you already pay for Pro or Business Premium, the rational move is to create a dot, keep its permissions tight, and give it one recurring chore you can verify. Treat the rest of the capability as something to earn trust in gradually.