Isla Fulford, a researcher at OpenAi, had a premonition that deep research would be a hit before it was released.
Fulford had helped to build the artificial intelligence agent, who autonomously explores, decides to click which links, what to read and what to collect in an in-depth report. OpenAi made in -depth research available for the first time; Every time it went, Fulford says, she was flooded with questions from colleagues who wanted to give it back. “The number of people who was me made us pretty excited,” says Fulford.
Since he goes to the public on 2 February, deep research has also turned out to be a hit with many users outside the company.
“Deep Research has written 6 reports today,” Patrick Collison, the CEO of Stripe who posted X a few days after the product was released. “It is indeed excellent. Congratulations with the people behind it.”
“Deep research is the AI product that has really received a meaningful part of the policy community in DC to feel the Agi,” wrote Dean Ball, a fellow at George Mason University that specializes in AI policy.
Deep research is available as part of the Chatgpt Pro plan, which costs $ 200 per month. A question is needed, such as “write me a report about the Massachusetts Security Insurance sector”, or “Tell me about the coverage of Wired about the Ministry of Government Efficiency”, and then comes with a plan, looking for relevant websites, by combing their content and deciding which links deserves further research. After exploring sometimes tens of minutes, it synthesizes his findings in a detailed report, including quotes, data and graphs.
Many tools that are currently branded as AI agents are essentially chatbots that are connected to simple programs without much refinement. The deep research model itself goes through an artificial reasoning before it comes up with a plan and goes ahead with every step. The model offers details about this reasoning behind his research in a side window.
“Sometimes it is as” I have to go back, this does not seem promising, “says Josh Tobin, another OpenAi researcher who is involved in building deep research.” It is quite cool to read some of those processes, just to understand how the model thinks. “
OpenAi clearly sees deep research as a tool that could take on more office work. “This is something that we can scale,” says Tinbin, 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 company can quickly prepare a report or presentation. Tinbin says that the longer goal is to “build a cop who is not only good at building reports through the web, but is also good in many other types of tasks.”
Because deep research has been trained to analyze and summarize people written by people, Tobin says that his team was surprised to see many people use it to generate code. “It's an interesting thread to pull,” he says. “We are not entirely sure what to make of it.”
