History
ChatGPT agent connects research with actions
OpenAI brought the approaches of Operator and Deep Research together in one agent.

On 17 July 2025, OpenAI introduced ChatGPT agent. The launch brought together capabilities from Operator and Deep Research. The system could combine research and actions using tools in its own computer environment. The announcement was therefore about an assistant that could also carry out a task, alongside answering in a conversation. Source.

At the time, this was an important change in the shape of the product. Gathering information and then doing something with it became parts of the same assignment. The relevant question for a user therefore became broader: what end result do I want, and which actions belong within that task?

For a small or medium-sized business, a useful first trial is an internal overview using public sources. Specify in advance what the final file must contain and where each conclusion must come from. Ask the agent to deliver a draft. Then review both the sources and the steps taken. Measure the time until the document is useful, including your own checks. This assesses the whole task. Treat this introduction as a historical moment, rather than a description of current availability.