Isla Fulford, a researcher at Openai, struggled that deep research would be a hit even before it was released.

Fulford helped build the artificial intelligence agent, which autonomously examined the web and decided which links to click, what to read and what to collect in a in-depth report. Openai made deep research available for the first time; When it goes off, Fulford says, she was inundated with inquiries from colleagues who were eager to get it back. “The number of people who made me disappear made us quite excited,” says Fulford.

Since Deep Research went to the public on February 2, Deep Research has also been a hit with many users outside the business.

“Deep research has written 6 reports so far today,” Patrick Collison, the CEO of Stripe posted on X a few days after the product was released. “It’s indeed excellent. Congratulations to the people behind it. ‘

“Deep research is the AI ​​product that really gained a significant part of the policy -making community in DC to start feeling the AGI,” wrote Dean Ball, a fellow at George Mason University, specializing in AI policy.

Deep research is available as part of the Chatgpt Pro plan, which costs $ 200 a month. It takes a query, such as “Write me a report on the Massachusetts Health Insurance Industry” or “tell me about Wired’s coverage of the Department of Government Efficiency”, and then comes with a plan, search for relevant websites, to combat their content and decide which links are to click and what information deserves further investigation. After you have sometimes explored ten minutes, it synthesizes its findings in a detailed report, which may include quotes, data and maps.

Many instruments that are currently branded AI agents are essentially chatbots linked to simple programs without much sophistication. The deep research model itself goes through an artificial kind of reasoning before devising a plan and moving forward with each step. The model contains details of this reasoning behind his research in a side window.

“Sometimes it’s like” I have to track back, it doesn’t look so promising, “said Josh Tobin, another open -minded researcher involved in building deep research. “It’s pretty kind to read some of these trajectories, just to understand how the model thinks.”

Openai seems to see deep research as a tool that can tackle more office work. “It’s a thing we can scale,” Tobin says, adding that the agent can be trained to complete specific white collar work. For example, an agent with access to the internal data of a business can quickly set up a report or presentation. Tobin says the longer purpose is to build an agent who is not only good at building reports by searching the web, but also good with many other types of tasks. “

Since deep research has been trained to analyze and summarize the human written text, Tobin says his team was surprised to see that many people use it to generate code. “It’s an interesting thread to pull,” he says. “We’re not quite sure what to make of it.”