OpenAI’s Sept. 29 DevDay Could Shed Light on Persistent AI Agents
The unannounced consumer AI product could continue tasks after users leave ChatGPT, with possible support for research, monitoring and email workflows.
OpenAI could introduce a consumer AI agent called O at its Sept. 29 DevDay event in San Francisco, potentially expanding ChatGPT into tasks that continue after users end a session.
OpenAI has not formally announced O. The name appeared in ChatGPT settings and upgrade-page elements, including a brief appearance as a benefit on the $100 Pro upgrade page. The settings also included an email suffix ending in “-o.”
The event’s opening keynote is scheduled for 1 p.m. ET (17:00 UTC). O could be introduced at the event, but its appearance on ChatGPT pages does not confirm a launch date or final product design.
O is described as an always-on or persistent agent that could keep working after a user leaves ChatGPT. That would distinguish it from a system that works only while the user is actively interacting with ChatGPT.
Potential uses include recurring research, monitoring events or conditions, managing email and follow-ups, and completing workflows over several hours or days. The concept would move ChatGPT beyond single-session exchanges toward software that users can assign ongoing work.
O may be connected to an internal OpenAI project called “Aeon.” That name has been associated with always-on agents and with the infrastructure for custom agents in ChatGPT Workspace. The references suggest O could be a consumer version built on the same or related technology.
OpenAI has not disclosed what permissions O would have, whether users could schedule tasks, how long-term memory would work or how users would control the agent. The possible email identity also suggests O may be able to receive and process messages directly, rather than only prepare draft replies.
The next concrete opportunity for OpenAI to discuss the product is the Sept. 29 DevDay keynote in San Francisco.
