OpenAI's Dots Bring Always-On AI Agents to ChatGPT

Dots were the headline announcement at OpenAI's DevDay 2026 keynote, held September 29 at Fort Mason in San Francisco in front of roughly 2,500 developers. Described by OpenAI as "remarkably capable, always-on agents built to handle everything," dots are a new way of working with AI that stays active on a person's behalf around the clock, rather than waiting for a new prompt each time. The announcement arrives just weeks after OpenAI released GPT-6 Astra, the underlying model that now powers dots, suggesting the two launches were designed as companion pieces from the start rather than separate, unrelated releases.

What Dots Actually Do

Per OpenAI's official announcement, dots are powered by GPT-6 Astra and run on their own cloud computer, giving them a persistent workspace separate from a user's own device. Dots can connect to more than 4,000 apps through OpenAI's plugin ecosystem, and users can reach their dot through ChatGPT on desktop, web, or mobile, as well as through Slack and Microsoft Teams, with texting support coming soon. Users give their dot a name to make it their own, and OpenAI has said it envisions entire teams of dots eventually working together on a person's behalf. OpenAI CEO Sam Altman described dots on stage as agents "built to handle really anything you can think of," positioning them as OpenAI's answer to similar always-on assistants like Meta's Muse.

How Do Dots Learn a User's Preferences?

Dots are designed to improve the more someone works with them. OpenAI says a dot learns a user's goals, standards, and working style over time, whether through direct conversation, voice calls, or simply observing the outcomes of tasks it completes. OpenAI's own examples include a dot that turns customer feedback into tested bug fixes for a developer, one that revises launch materials as a product's scope changes, and one that reruns a scientist's analysis whenever new data arrives.

Dots Can Work Independently on Ongoing Projects

Unlike a typical chatbot exchange, dots are built to keep working on a task even after a conversation ends. OpenAI says a dot can take on a project and continue running it while simultaneously handling several others, letting users add new tasks without personally directing every step along the way. Inside OpenAI itself, the company says dots have already started investigating bugs reported in Slack and turning new designs into working apps while human teams stay focused on customer feedback.

What Is Proactive Research in Dots?

Proactive research is what OpenAI calls the background activity a dot performs when a user isn't actively interacting with it. During this time, a dot looks for ways to help using only the apps a user has already connected, restricted to read-only tools that cannot send messages, change app content, or control a device. That restriction is one of several safeguards OpenAI built specifically to limit what a dot can do without direct oversight.

How OpenAI Keeps Dots Under User Control

OpenAI has emphasized safety and control throughout the dots rollout. Each dot runs on its own cloud computer, kept separate from a user's personal device unless they explicitly choose to connect it. Dots use built-in rules to determine when to act independently versus when to pause for approval, and users can set Custom Rules to allow, require approval for, or fully block specific actions. An auto-review system checks potentially sensitive actions against a user's instructions and safety requirements before they proceed, and certain tasks, like changing a password, always require the user to handle them directly.

Can Dots Make Mistakes?

Yes. OpenAI is explicit that dots can still make mistakes and encourages users to review consequential work rather than treat a dot's output as final. An Activity View lets users follow a dot's progress, including background work, and redirect it as needed if something looks off.

Who Can Use Dots Right Now

Dots began rolling out on September 29, 2026 to ChatGPT Pro and Business Premium subscribers in eligible markets, with OpenAI planning to expand access to more users over time. Enterprise, Edu, and Healthcare workspace users can also access a beta version once their workspace administrator enables it. A user's first dot is included in their existing Pro or Business Premium plan at no additional cost, and conversations with a dot don't count against standard ChatGPT usage limits, though tasks a dot manages through Codex or ChatGPT Work do count as usual.

Are Specialist Dots Different From Regular Dots?

Yes. Alongside personal dots, OpenAI previewed specialist dots, a separate offering built for organizations rather than individuals. A specialist dot is set up by a company with its own identity, credentials, and access to specific internal systems, allowing it to take on well-defined responsibilities like procurement, invoice processing, or customer support. OpenAI is starting specialist dots with focused enterprise pilots and is also working with Microsoft to integrate them into Microsoft Agent 365 for enterprise governance and security management.

Dots Arrived Alongside a Packed DevDay 2026 Lineup

Dots shared the stage with more than 20 other announcements at DevDay 2026, according to CNBC's live coverage of the event. Among them was GPT-6.1 Sol, a new model OpenAI positions between the faster GPT-6 Luna and the more powerful GPT-6 Astra, offering close to Astra-level intelligence at a fraction of the cost. OpenAI also introduced ChatGPT Space, a collaborative workspace for teams, alongside an Ultrafast speed tier and a new $500-per-month Pro plan. With dots as its centerpiece, DevDay 2026 marked what CNBC and other outlets described as OpenAI's clearest step yet toward positioning AI as something that actively manages ongoing work, rather than simply answering questions when asked.