OpenAI releases AI agent ‘Deep Research’ designed to act like analyst

OpenAI is releasing a new artificial intelligence tool that’s designed to carry out time-consuming online research for users about everything from complex science questions to car recommendations — expanding the startup’s portfolio of AI agents that act on a person’s behalf.

The service, called Deep Research, will be available to certain paying customers through OpenAI’s ChatGPT chatbot online, the company said in a blog post on Sunday. In response to a prompt, the tool will scour words, images and PDFs online, as well as files uploaded by the user, to create an in-depth report. OpenAI compared the feature to a research analyst and said it’s meant to do in “tens of minutes” what would typically take a person “many hours.”

Deep Research is the second AI agent that San Francisco-based OpenAI has released this year. Last month, OpenAI introduced Operator, which can help book flights, plan grocery orders and even complete purchases for users. Both services are initially available just to those who pay $200 per month for OpenAI’s recently introduced ChatGPT Pro option.

The rollouts are part of a broader industry push toward agents, or AI software that can complete multi-step tasks for users with minimal supervision. OpenAI-backer Microsoft Corp. and rival Anthropic have launched their own takes on agent software, as have a number of other startups. The companies hope such tools can save users time with their personal and professional tasks and thereby live up to the long-held promise that AI will make people more productive.

OpenAI Chief Executive Officer Sam Altman has previously said agents will be “the next giant breakthrough” for AI. The stakes of that bet have only increased amid renewed concerns that chatbots from Chinese companies like DeepSeek are rapidly catching up to top American AI developers, including OpenAI.

The ChatGPT maker cautioned that Deep Research is still in the early stages and can present made-up information as factual. It may also have difficulty distinguishing rumors from accurate information.

The research tool is also “very compute intensive,” according to OpenAI. To start, users will only be able to submit 100 queries per month.

OpenAI plans to eventually offer the service to other paying customers, including those who subscribe to its Plus, Team and Enterprise options, but the company did not provide a timetable for doing so.