OpenAI has introduced Dots, a new line of AI agents designed to function as persistent assistants. Powered by the GPT-6 Astra model, these agents operate in the background using a dedicated cloud computer to interact with over 4,000 applications and web browsers. Users can communicate with their agents via text or voice through ChatGPT on web, desktop, and mobile platforms.
The agents integrate with tools like Slack and Microsoft Teams to maintain context across different devices. Currently, users are limited to creating a single Dot, but the company plans to enable the deployment of multiple agents simultaneously. OpenAI also intends to offer options to scale agent performance by adjusting processing speed and monthly task capacity in the future.
Dots feature built-in safety protocols, allowing users to define custom rules that restrict specific actions or mandate permission requests. An auto-review function helps the agents verify compliance with these safety requirements and identify tasks that require human intervention, such as password changes. These interactions do not count toward standard ChatGPT usage limits.
The service is available starting today for ChatGPT Pro and Business Premium subscribers. Additionally, OpenAI is testing a feature for organizations to develop specialist agents tailored for specific professional roles. This launch follows the recent release of Meta’s Muse assistant, positioning Dots as a direct competitor in the agentic AI market.
