Product Signal

AI Use Moves From Chat to Agents

OpenAI’s workplace-task research is broadening, while Mollick’s guide is being observed as shifting from chat models toward agent systems.

By All We Need Editorial System

The main AI product narrative is shifting from the chat box to agent systems. OpenAI published research on workplace use, Simon Willison observed that Ethan Mollick’s AI-use guide has moved from model choice toward agent systems, and one interview title points directly to a plan for ChatGPT as an everything app.

Work is being described as task expansion

OpenAI Blog published research saying AI is expanding what people do at work, with ChatGPT users taking on tasks across roles and reshaping work boundaries. The framing does not focus only on automating a single job; it places ChatGPT usage inside expanding task scope and changing role boundaries. For enterprise researchers, that shifts the object of study from individual productivity to organizational task flows.

Independent observation shows usage guides are being rewritten

Simon Willison commented on the evolution of Ethan Mollick’s AI-use guide, saying that a year ago it centered mainly on chat models such as ChatGPT, Claude, and Gemini, plus modes such as Deep Research. Today it is more focused on agent systems, where AI performs tasks. Latent Space published an interview titled “OpenAI’s Plan to Make ChatGPT the Everything App,” adding an adjacent product-strategy signal. The three sources differ in strength, but their direction is aligned.

Entry points and task boundaries are changing

The facts are that OpenAI is framing ChatGPT around workplace task expansion, and external observers are describing AI use as moving from chat to agent systems. The editorial inference is that if users begin treating AI as a task-execution layer rather than a Q&A tool, product entry points, permission management, and data connections all become more important. The speed of adoption still needs more usage data.

What to watch next

Watch whether ChatGPT adds more cross-application execution, enterprise permissions, and workflow functions, and whether later OpenAI research discloses task types, role changes, and retention metrics.

Sources