> ## Content Index > Fetch the complete content index at: https://www.tao.media/llms.txt > Use this file to discover other available public pages before exploring further. # OpenAI Launches Dots as Always-On Agents Powered by GPT-6 Astra - URL: https://www.tao.media/openai-launches-dots-as-always-on-agents-powered-by-gpt-6-astra/ - Published: 2026-09-29T17:50:40.000Z - Updated: 2026-09-29T17:50:40.000Z - Description: OpenAI’s dots are named agents with their own cloud computer, cross-app memory, and approval controls for work across ChatGPT, Slack, Teams, and thousands of connected apps. - Author: Bart Hillerich - Tags: OpenAI, AI, News OpenAI has launched [dots](https://openai.com/index/introducing-dots/?ref=tao.media), a new layer of always-on personal agents powered by GPT-6 Astra and designed to work across ChatGPT, Slack, Microsoft Teams, and connected third-party apps. OpenAI describes dots as named agents that can learn from user feedback, run multi-step projects on their own cloud computer, and keep working toward goals even when the user is not actively chatting with them. Instead of asking a model for one answer at a time, users can create a dot, give it a role, connect apps, set permissions, and let it carry work across multiple channels. The rollout is beginning for Pro, Business Premium, and Enterprise users in eligible markets, with broader availability planned. 0:00 /2:51 1× OpenAI dots launch video ## How OpenAI dots Work Each dot has its own cloud computer and browser, so it operates in a separate environment from the user's device. dots can run several projects at once, maintain context across ChatGPT, Slack, and Teams, and connect to more than 4,000 apps through plugins. That design gives dots a more persistent operating surface than a standard chatbot. A user can assign a project in ChatGPT, continue the exchange in Slack or Teams, and inspect the dot's cloud computer to see the work in progress. Voice calls work at launch, while texting is planned for a later release. ![dots are an extension of you. With a dot that knows your goals and standards, you can take on more without directing every step.](https://storage.ghost.io/c/78/0b/780ba906-b1a7-4bf0-873c-bdd5c32e5331/content/images/2026/09/dots-quote.png) “dots are an extension of you. With a dot that knows your goals and standards, you can take on more without directing every step.” dots can also adapt over time to how a person works. They can learn from feedback, bring back work in the user's preferred style, and use prior context to make future work more useful. That makes dots useful for recurring workflows such as research, project coordination, draft preparation, task follow-up, and cross-app information gathering. The bigger shift is that the agent is not limited to one chat window; it has an identity, a workspace, and permissioned access to external tools. ## Proactive Research Adds an Always-On Layer One of the main differences between dots and earlier ChatGPT workflows is proactive mode, which OpenAI calls proactive research. When a dot is not actively working with the user, it can look for ways to help by using read-only tools on apps the user has already connected. OpenAI says that mode does not let the dot send messages, change app content, or control a browser or computer. Instead, it surfaces useful context, monitors relevant information, or prepares suggestions before the user asks. Proactive agents create a different risk profile from normal assistants. A chatbot usually waits for a prompt, while an always-on agent can notice information, infer next steps, and prepare work without being directly asked in the moment. OpenAI is trying to separate that background intelligence from actions that could affect accounts, publish content, move money, or expose sensitive information. dots are [built with controls](https://openai.com/index/how-we-build-safety-security-and-privacy-into-dots/?ref=tao.media) around account access, user approvals, monitoring, and action review. ## Permissions, Isolation, and Auto-Review OpenAI's control model centers on Custom Rules, Activity View, and auto-review. Users can allow certain actions, require approval for others, or block categories of work altogether. Activity View shows what a dot is doing, and OpenAI says its monitoring systems can pause or stop work when safety concerns arise. The [auto-review documentation](https://learn.chatgpt.com/docs/sandboxing/auto-review?ref=tao.media) covers actions that could affect accounts, share information, or create other meaningful consequences. Sensitive tasks, such as changing a password, stay with the user rather than being delegated to the dot. The cloud computer setup is also a boundary. Each dot's computer is separate from the user's machine unless the user chooses to connect it. Saved passwords can be used to sign into supported sites without exposing those passwords to the model. ## Enterprise dots Point Toward Managed AI Workers OpenAI is previewing specialist dots for enterprises, with early internal testing in procurement, invoice processing, email marketing, customer support, and commercial contracting. These versions are expected to have separate identities for access management, IT-provisioned hardware, and deeper integrations with systems of record. A specialist dot could be assigned to a defined business process, connected to approved systems, and managed under company policies rather than operating only as an individual user's assistant. OpenAI says Enterprise, Edu, and Healthcare customers can access the beta when a workspace admin enables it. The company also says work is underway on Microsoft Agent 365 integration for enterprise governance and security. ## Pricing and Availability dots are beginning to roll out across Pro, Business Premium, and Enterprise plans in eligible markets, with wider access planned over time. A user's first dot is included in Pro or Business Premium at no extra cost. OpenAI says those plans include an allowance for deeper work, with extended limits during the first month after launch. dot conversations do not count toward ChatGPT usage limits, though Codex and ChatGPT Work tasks still do. Future options will include adding more dots and scaling either speed or monthly work capacity. Users can create their first dot from the ChatGPT desktop app or desktop browser, connect apps during setup, and use mobile messaging afterward. The company's setup page is available at [chatgpt.com/dots](https://chatgpt.com/dots?ref=tao.media).