Google has announced "Deep Research," a new AI feature that automatically conducts research and summarizes information into reports

Google has announced that they are introducing a new feature called "Deep Research" in their AI product, "Gemini." Deep Research automatically creates a comprehensive research plan based on the questions input by users.

Google has announced "Deep Research," a new AI feature that automatically conducts research and summarizes information into reports

Deep Research automatically collects information from across the web and creates a comprehensive report. This feature is being launched today for Gemini Advanced on desktop and mobile web, and is expected to be available in the mobile app in early 2025.

AI that Automatically Conducts Research: "Deep Research"

Google has announced a new feature called "Deep Research" that is integrated into Gemini. "Deep Research" automatically creates a step-by-step research plan based on the questions input by users.

Deep Research
Deep Research

Once users modify or approve that plan, Gemini will search for and analyze related information from across the web. Gemini will perform repeated searches over several minutes to find information, and based on the results, will conduct new searches.

By repeating this process multiple times, Gemini can produce a comprehensive report. This report can be exported to Google Docs and also includes links to the sources of information.

The combination of web search, Gemini's advanced reasoning capabilities, and the context window of 1 million tokens makes the generated reports useful and easy to read.

Deep Research is being launched today for Gemini Advanced on desktop and mobile web, and is scheduled to be available in the mobile app in early 2025. As of this writing, this feature is only accessible to English-speaking users.

Users can access it by selecting [Gemini 1.5 Pro with Deep Research] on the model selection screen.

Google also plans to integrate agent functions into its products this year, and Deep Research is the first feature from that plan.

On the same day, Google also announced "Gemini 2.0 Flash," which supports multimodal input and output as well as agent functions.


Google has integrated "Deep Research" into Gemini, allowing users to automatically generate comprehensive reports based on the questions asked. Additionally, the latest experimental model, "Gemini 2.0 Flash," has also been introduced with performance enhancements. This enables more sophisticated research and information gathering.