OpenAI’s Deep Research Agent Is Coming for White-Collar Work

Isa Sa Fulford, a researcher OpenWas a crunch Deep research It will hit before the publication.

Helped to create Fulford Artificial intellect The agent, which autonomously explore the web, what links to click, what to read and what to collect in the depth report has been decided for himself. Opeena first made internally deep research; Fulford says whenever it goes down, the questions were submerged from colleagues interested in getting it back. “The number of the numbers that I had had made was quite excited to us.”

Deep research has also been proven as a hit to many users outside the company since February 2 went straight to the public.

Stripe CEO Patrick Collison Patrick Collison Patrick Collison X posts A few days after the publication of the product. “It’s really great. Congrats for the people behind it.”

“Deep Research is the AI ​​product that AGI was realized in DC to realize the truly meaningful part of the community in DC,” Wrote Dean Ball, an associate of George Mason University who experts in the AI ​​policy.

Deep research is available as part of the plan to pay all the money, although most users are bound to 10 queue per month. ($ 200 Chat for Plan Get 120 Query per month) It takes a question, such as “Write me a report on the Massachusetts Health Insurance Art,” or “Tell me about wired coverage Government skill department“And then brings a plan, searching the relevant websites, clicking on their contents and clicking on what links and demanding further investigation. Sometimes after exploring a few minutes, it synthesizes its search in a detailed report, which may include quotations, data and charts.

Many of the current branded tools currently branded as an AI agent are mainly connected to ordinary programs without much sophisticated. The deep research model itself goes through an artificial type of logic before creating a plan and moving forward with each step. The model provides the details of this argument behind its research in the adjoining window.

“Sometimes it needs to be ‘backtrack, it doesn’t seem commitment to it,” “Another OpenAI researcher involved in the creation of deep research, Josh Tobin says.” Just understand how the model is thinking, very cool to read some of those tragicular “”.

Open is clearly seen as a tool that can do more office work. Tobin says, “This is something we can scale,” the agent can be trained to complete the specific white-collar work. An agent with access to an organization’s internal data can quickly prepare a report or presentation, for example. Tobin says that the long goal is “Creating an agent that is not only good in making reports through web search, but also good for many other types of work” “

Since the deep research was trained to analyze and shorten human-written text, Tobin said that his team was surprised to see many people using it to create the code. “It’s an interesting thread to pull,” he says. “We’re not completely sure what it will make.”