OpenAI Launches Dots, Always-On AI Agents That Work in the Background

OpenAI introduced Dots at its DevDay event on Tuesday, presenting a new personal assistant designed to keep working on users’ goals beyond a single conversation.

Powered by GPT-6 Astra, Dots are intended to operate continuously in the background, carrying out ongoing responsibilities with minimal supervision. OpenAI’s announcement describes a future in which users have teams of agents working together on their behalf.

The launch combines capabilities already available through Codex and similar agent systems into a product centered on persistent, independent activity. It also gives those capabilities a more approachable identity through floating cartoon characters.

Dots Focus on Ongoing Responsibilities

Dots are designed to function independently of a particular device or interface. Users define a goal, and the assistant continues pursuing it rather than waiting for a fresh instruction at every stage.

That approach could make Dots useful for work that changes over time or depends on new information arriving.

OpenAI offered the example of a software developer assigning a dot to monitor customer feedback. The agent could then implement bug fixes and requested features as necessary, connecting incoming feedback with development work.

In another scenario, a scientist could ask a dot to rerun analyses as experimental data becomes available and investigate unexpected results.

These examples illustrate OpenAI’s ambitions for the product. They do not establish that every dot will perform every task successfully or that users can leave all decisions entirely unsupervised.

Access Begins With Pro and Business Premium Users

Dots became available Tuesday through ChatGPT for Pro and Business Premium subscribers in eligible markets.

Users can launch the assistants from either ChatGPT or Codex. OpenAI says people can begin with a primary dot, give it a name and personalize it.

The company has also built communication options beyond its own interfaces. Users can message Dots through Slack, Microsoft Teams and other organizational platforms, allowing the assistants to participate in existing workplace workflows.

Text message support is planned, although the announcement did not provide a release date for that feature.

The combination of background operation and multiple messaging channels is central to the product: a dot can maintain an ongoing responsibility while users interact with it through the tools they already use.

Specialist Agents Will Have Their Own Access and Tools

OpenAI also outlined plans for specialist Dots with defined responsibilities.

Individual agents can be provisioned with specific identities, credentials and tools through existing systems. That would allow organizations to configure agents for particular roles and determine which resources they can use.

OpenAI is working with Microsoft to integrate Dots into Agent 365 security controls. The collaboration connects the product’s workplace ambitions with systems intended to govern agent access and activity.

Over time, OpenAI envisions multiple Dots collaborating on a user’s behalf. The announcement presents that team-based model as a longer-term direction rather than detailing a complete rollout schedule.

A Familiar Agent Model Gets a New Identity

Much of the underlying functionality resembles capabilities already offered through Codex and other agentic tools. Dots packages those features around the idea of an assistant with continuing responsibilities.

The branding is part of that shift. Its small, floating cartoon characters give the software a playful appearance, echoing the approachable presentation of Meta’s recently released Muse agent.

For OpenAI, Dots brings persistent agents into a more personal product format. Its practical value will depend on how reliably those agents carry out ongoing work and how effectively users and organizations can manage their responsibilities and access.