OpenAI Unveils Dots to Humanize AI Agents

Powered by the GPT-6 Astra model, these agents operate within their own dedicated cloud environments to manage workflows 24/7. Users can interact with their chosen character via ChatGPT, Slack, or Teams, with the system designed to learn personal preferences and work styles over time. OpenAI frames these mascots as extensions of the user, capable of handling tasks ranging from proactive research to ongoing project management.

Despite the whimsical branding, the company maintains strict operational boundaries. Users retain control over app permissions and can monitor the agent's progress through an inspection interface. While the agents can work independently, sensitive actions like password changes remain under human oversight. By pairing sophisticated autonomous capabilities with a approachable interface, OpenAI is leaning into a psychological design strategy: replacing the cold utility of software with a familiar face to encourage deeper user reliance.