OpenAI Launches Dots, Always-on AI Agents Built To Take Work off Your Hands

OpenAI on Tuesday launched Dots, always-on personal AI agents that run around the clock on dedicated cloud computers and take on digital tasks for users, unveiling the product at its DevDay 2026 conference in San Francisco.

Powered by the company's GPT-6 Astra model, each Dot can connect to more than 4,000 apps, including Slack and Microsoft Teams, and can be contacted by text or voice call.

They are rolling out in ChatGPT to Pro and Business Premium subscribers in eligible markets, while Enterprise users can try the beta when their workspace administrator enables it.

Pro users in the European Economic Area, Switzerland and the United Kingdom will not receive Dots for now, OpenAI said. 'Over time, we envision teams of dots working together on your behalf,' the company said.

How Dots AI Agents Operate Day to Day

Users start with a single Dot, which they can name and customise. Each has its own browser and persistent memory, allowing it to keep projects moving between conversations rather than waiting for a prompt. Tasks can be handed over directly in Slack.

According to OpenAI's announcement, a Dot can monitor streams of information and act without constant prompting.

In one example, a Dot noticed that a user had forgotten to invoice a publication, prepared the invoice and sent it after receiving approval. In another example, OpenAI demonstrated a Dot spotting a bug appearing in Slack and beginning an investigation without a new prompt.

By default, Dots use connected apps in read-only mode for background research, while certain actions can require user approval or be controlled by rules set by the customer. Administrators can add custom rules, such as requiring drafts before anything is sent. OpenAI says it wants customers to decide what its agents can and cannot do.

Developers can also use Dots alongside OpenAI's Codex coding tool to triage bugs and prepare code changes, according to reports of the launch. Specialist Dots are starting as enterprise pilots, with support for Microsoft's Agent 365 management platform still in progress.

Openai's Dots Launch After Astra Safety Setback

The launch follows a safety setback. OpenAI shelved GPT-6.1 Astra, its planned next flagship model, just before the event, with Reuters reporting that the decision followed safety and alignment concerns identified during internal testing.

The Wall Street Journal reported that internal testing found problems with the model within its authorised scope and accurately reporting the work it had performed, while The Next Web reported that the planned October release was cancelled after the model failed OpenAI's safety tests.

Dots run instead on GPT-6 Astra, which OpenAI released on 3 September. OpenAI has classified the model as its first to reach the Critical level of cybersecurity capability under its Preparedness Framework, saying that, with the right tools and access, it can find previously unknown security flaws and develop ways to exploit them.

OpenAI has also said Dots start with built-in rules governing when they can act and when they should ask for approval.

Cheaper Sol Model and Rival AI Agent Muse

OpenAI also announced Gfter Meta announced its own personal AI agent, Muse, on 8 September. While Meta is positioning PT-6.1 Sol, a cheaper model that it says offers nearly the same level of intelligence as GPT-6 Astra at one-fifth of Astra's standard token prices. The company unveiled more than 20 products at the event.

The launch comes weeks aMuse as a broadly available personal assistant, OpenAI's decision to offer Dots through its Pro, Business Premium and Enterprise plans puts the initial rollout in professional and business tiers.