Highlights:

  • For now, Deep Research is available exclusively to paid Google Workspace users with access to Gemini Advanced.
  • According to Google, Deep Research is designed to save time, as conducting thorough research is often tedious and time-consuming.

Google LLC introduced Deep Research, an experimental artificial intelligence research tool. It is now available to Gemini Advanced users through the Google Workspace platform.

Initially introduced last year, Deep Research is designed to explore complex topics and conduct in-depth analysis, providing precise answers to users’ inquiries. By scanning the internet and reviewing hundreds or even thousands of sources, it compiles findings into a detailed yet easy-to-understand report, according to the company.

The key advantage of Deep Research is its ability to accomplish in minutes what would typically take a human researcher hours, making it valuable in various business and educational settings.

For example, Google highlights its usefulness in industry research, enabling businesses to stay informed about emerging trends, such as advancements in sustainable battery technologies, which can guide clean technology strategies.

Another promising application is competitive research, where organizations can use Deep Research to conduct in-depth analyses of local and global competitors or identify the best location for launching a small business, according to Google. It can also assist with customer research—for instance, a salesperson could generate a detailed report on their company’s products and services, customized for a specific client.

Additionally, educators can leverage Deep Research for various tasks, including lesson planning, class project development, presentations, and grant writing, allowing them to create reports on virtually any subject, Google noted.

For now, Deep Research is available exclusively to paid Google Workspace users with access to Gemini Advanced. Google Workspace is Google’s suite of productivity tools, including Docs and Sheets, while Gemini Advanced is the premium version of the Gemini AI application, offering significantly more token capacity than the standard version, along with benefits like expanded storage.

To begin using Deep Research in Google Workspace, Gemini Advanced users must select the “1.5 Pro with Deep Research” model within the app and enter a simple prompt outlining the topic they want to explore in detail.

Deep Research will then generate a multistep research plan, which users can modify as needed. Once finalized, it will immediately start analyzing the most relevant information from its web search. As it processes data, it continuously refines its analysis based on new insights. For example, if it’s investigating emerging trends in clean battery technology and identifies a promising innovation, it will automatically gather as much information as possible on that breakthrough.

Once the research is complete, Deep Research generates a comprehensive report summarizing its key findings. Google states that the report is presented in a clear, easy-to-read format, complete with relevant links to original sources, allowing users to explore specific aspects in more detail if needed. Additionally, users can ask follow-up questions to refine their research further.

According to Google, Deep Research is designed to save time, as conducting thorough research is often tedious and time-consuming. It also requires a certain level of expertise in knowing how to search effectively. By automating the process, Deep Research can condense hours of work into just minutes, covering virtually any topic.

Currently, Deep Research is available to Google Workspace users on the web, with mobile support planned for a later release. At this time, it only supports English, but Google intends to expand it to more languages soon.

Alongside Deep Research, Google also introduced two new experimental models within Gemini Advanced: 2.0 Flash Thinking Experimental and 2.0 Pro Experimental. These models are currently ranked as the world’s second and third most powerful generative AI models on Imarena’s Chatbot Arena LLM Leaderboard.

The 2.0 Flash Thinking Experimental model is Google’s most advanced reasoning model to date. Unlike traditional generative AI models, it deconstructs prompts into a series of logical steps, enhancing its reasoning capabilities and producing more accurate responses. What sets it apart is its ability to transparently outline its entire thought process, allowing users to track its reasoning and understand the assumptions behind its conclusions.