OpenAI introduced dots as an always-on personal assistant at its developer day in late September. The feature appears inside the ChatGPT desktop and mobile apps for subscribers on the Pro tier or higher, with plans starting at $100 a month, and is designed to keep working in the background once connected to a user’s calendar, email, computer and other tools.
From coding tools to a persistent agent
According to Alexander Embiricos, who leads OpenAI’s enterprise product team, dots follows a path that began with the company’s coding tools. Embiricos said much of the assistant’s functionality could be reproduced with OpenAI’s Codex app, released earlier this year, but dots packages those abilities into a single, persistent conversation aimed at feeling more intuitive for users.
Embiricos described the product as an “entity” that maintains one continuous conversation rather than many separate chats. He recounted early tests in which the assistant flagged calendar conflicts and deadlines, noting one instance when it warned about a conflict only three minutes before the event.
Privacy, controls and proactive research
OpenAI built a feature it calls “proactive research,” which Embiricos said lets dots periodically assess what it knows about a user, what conversations it has had and which connected resources it can access, then propose ways to be more helpful. The company says users can turn off model training to stop OpenAI from using their interactions to improve its models, and can control which resources the assistant can access.
Embiringicos acknowledged limits in the current controls and said they plan to build “super, super, super granular controls.” The app currently refuses read-only access to cloud drives, and the defaults were initially conservative, a choice that received user feedback according to Embiricos.
Questions about scale and permissions
Embiricos pointed to examples—such as an engineer whose assistant detected an outage before the human noticed—to illustrate potential benefits. He also noted friction: businesses and people do not always want to interact with agents, and some corporate partners prefer human contact.
OpenAI is reportedly in discussions with payment processors about letting agents act on behalf of users, a step that would raise further permission and safety questions. Dots is currently available to Pro subscribers in ChatGPT apps; pricing and controls are as described by OpenAI.
Original source: TechCrunch AI