OpenAI CEO Sam Altman unveiled Dots, the company's new always-on AI agents, during its annual DevDay developer conference in San Francisco on Tuesday.

The agents, described as remarkably capable tools that handle tasks autonomously, run on OpenAI's GPT-6 Astra model and operate on dedicated cloud computers with browsers. They connect to more than 4,000 apps via plugins and can perform multistep work such as research, data analysis, coding, scheduling, and monitoring projects in the background.

Altman pitched Dots as personalized digital assistants depicted as cute animated blobs that users can name and assign responsibilities. The system allows teams of Dots to collaborate, with users communicating via ChatGPT, Slack, Microsoft Teams, and eventually phone or text.

Availability begins Tuesday for ChatGPT Pro and Business Premium subscribers in eligible markets. The rollout emphasizes enterprise features, including integrations with Microsoft security controls and custom rules for permissions on sensitive actions.

The launch positions OpenAI against recent competitors, including Meta's Muse agent released earlier this month. OpenAI highlighted safeguards such as read-only background access and required approvals for actions like data deletion.

Dots represent a shift toward persistent AI that continues working independently after initial instructions, learning user preferences over time through feedback.