OpenAI is moving from chat toward proactive, agentic AI tools

OpenAI executive Greg Brockman said the company is seeing a shift from pure chat use cases toward agentic ones, while traditional chat remains widely used. He said OpenAI is working to unify these modes into systems that can act proactively and persistently according to users’ goals, with safety as a core commitment. The timing and exact form of that unification were not specified.

Brockman said ChatGPT has more than one billion weekly users, while also citing approximately 300 million weekly health queries and estimating that another one billion to 1.5 billion people have tried ChatGPT before. These figures were presented as claims without supporting detail.

The discussion identified product discovery as a major challenge: users may not realize how much capabilities have changed. OpenAI’s proposed direction includes proactive onboarding and suggestions, while specific demonstrations—such as turning an AI-generated candle-holder image into a physical object—were presented as a way to show concrete use cases and help users discover additional workflows.