OpenAI Launches "Dots," Active Agents for Paid Subscribers

OpenAI introduces "dots," autonomous agents capable of operating continuously and orchestrating tasks across a wide range of applications. The deployment begins with paid offerings, accessible via ChatGPT on desktop, web, and mobile, along with announced control mechanisms to regulate their actions.

Immediate Access for Paid Subscribers and Deployment Kickoff

Users with eligible plans can now access dots. The deployment pertains to Pro, Business Premium, and Enterprise offerings in eligible markets, with an extension planned soon. To create and name a first dot, simply open the ChatGPT application on desktop or use a browser; this first agent is included in Pro or Business Premium plans. Dots are accessible via ChatGPT on desktop, web, and mobile, and users can communicate with them from the mobile app.

Professional Integrations and Connection to Over 4,000 Applications

Dots can connect to over 4,000 applications and integrate into work environments like ChatGPT, Slack, or Teams to manage professional tasks. These agents are designed to handle various tasks and can be organized into teams collaborating for the user. They gradually learn from the user's preferences and thought processes. It is also possible to link a dot to applications and plugins to expand its capabilities.

Read-Only Permissions and Behavior Monitoring

OpenAI announces that dots come with built-in protections, including read-only permissions to limit certain actions. A monitoring system is planned to detect potential security concerns or harmful behaviors. Checks are also scheduled regarding the applications accessible by the agents. According to the description, dots correspond to autonomous agents that are constantly operational, relying on the GPT-6 Astra model.