OpenAI introduces Dots for persistent background agent tasks

OpenAI unveils Dots at Dev Day

OpenAI has launched Dots, a personal agentic assistant powered by GPT-6 Astra. Chief executive Sam Altman introduced the product at the company’s Dev Day event at Fort Mason in San Francisco, describing a new class of agents intended to pursue user-defined goals continuously in the background.

Dots are available from Tuesday in ChatGPT to Pro and Business Premium customers in eligible markets. OpenAI says a user can begin with a primary Dot, give it a name and customise it, while the longer-term vision is for teams of Dots to work on a user’s behalf.

Persistent agents beyond a single interface

Unlike Codex or ChatGPT, Dots are presented as operating independently of a particular hardware device or user interface. The emphasis is on ongoing work with minimal oversight rather than a single prompted interaction.

OpenAI outlined several potential deployments. A software developer could assign a Dot to monitor customer feedback and implement bug fixes or requested features. A scientist could use one to rerun analyses and investigate unexpected results as new experimental data arrives.

The release extends OpenAI’s business-facing ChatGPT offering, where ChatGPT Work’s 20 million users illustrates the scale of its workplace user base and the importance of managed access for persistent agents.

Identity, tools and security controls

Users can communicate with Dots through Slack, Microsoft Teams and other organisational platforms, with text-message support planned for a later stage. OpenAI also describes specialist Dots with discrete responsibilities. Each agent can be provisioned with particular identities, credentials and tools through existing systems.

OpenAI is working with Microsoft on integration with Agent 365 security controls. The company says much of the underlying functionality was already possible with Codex and similar agentic harnesses; Dots packages those capabilities around independent agentic action.

The product also adopts a bubbly, cartoon-like visual identity: small floating dots intended to make the software feel more relatable. OpenAI’s business implication is practical: organisations considering Dots should match each agent to a defined responsibility, set credentials and tool access through existing controls, and retain oversight over the actions it performs.