OpenAI launches always-on Dots AI agents at DevDay

OpenAI introduced Dots, an always-on artificial intelligence agent, at its DevDay conference in San Francisco on September 29, saying the assistants can work in the background, learn from user feedback and connect with applications. The company said the rollout began that day for eligible Pro and Business Premium users, with regional restrictions. OpenAI described each Dot as running on its own private cloud computer, where it can use a browser and apps connected by the user. The company said users can inspect the agent’s work and, if they choose, allow it to connect to a personal computer. Dots are intended to carry out ongoing tasks, rather than respond only to individual prompts. The company also described safeguards for the agents. OpenAI said a monitoring system can identify safety concerns and pause a Dot’s work, and that its review system checks with users before sensitive actions, such as changing a password. These capabilities and protections were presented as OpenAI’s product claims; the available reports did not provide independent testing of agent performance or safety. Access details were not fully consistent across the company’s reported rollout descriptions.