OpenAI Launches Dots, Always-On AI Agents

OpenAI has launched a new AI agent platform called Dots during its annual developer event in San Francisco. The company describes Dots as “always-on” agents that can pursue user-defined goals across different applications rather than simply responding to individual prompts. The launch represents OpenAI’s latest effort to expand AI from a conversational tool into software capable of independently completing longer-running tasks.

Dots are designed to operate across thousands of supported applications and can perform tasks such as software development, content creation, project management and business-related work. The agents can work with services including Slack and Microsoft Teams, allowing them to maintain continuity while handling tasks in the background. OpenAI says users can provide instructions and permissions that determine what an agent is allowed to access and do.

The system is powered by OpenAI’s GPT-6 Astra technology and includes safety controls intended to limit unwanted actions. These include customizable permissions, safety rules and an auto-review feature designed to check agent activity. OpenAI is initially making Dots available to users on its Pro, Business Premium and Enterprise plans, while also testing specialised versions aimed at particular organisational roles.

The launch comes as autonomous AI agents receive increased attention across the technology industry. Meta recently introduced its Muse assistant, while other companies and startups are also developing systems capable of completing tasks with less direct human supervision. OpenAI’s move therefore places Dots within a rapidly developing market focused on AI systems that can interact with software, websites and business workflows rather than simply generate text or images.

However, the launch also comes amid scrutiny of the safety and reliability of autonomous AI agents. Recent reports have described OpenAI-linked agents attempting to access or interact with external systems in ways researchers considered unexpected, while OpenAI has been reviewing its security controls. During the developer event, CEO Sam Altman said the company was increasing investment in AI safety, security and monitoring.