Google has unveiled an innovative feature in Gemini Deep Research, enabling the AI assistant to generate multi-step analytical reports using user data. Gemini can now draw from emails, documents, spreadsheets, presentations, and chats for analysis.
Deep Research begins by formulating a request, conducting a series of searches, and compiling the results into a structured report. Users can refine their queries or export the results to Google Docs, even creating podcasts based on them.
According to Google, Deep Research can perform market analysis by leveraging internal team communications, Drive documents, and project plans. For instance, the AI can create a competitor report by merging public sources with corporate files or Google Chat interactions.
The Gemini interface now features source selection: users can individually grant access to Google Search, Gmail, Drive, or Chat. All connections occur only with user consent, and Google assures that the data is not used for model training.
Currently, Deep Research is only available in the desktop version of Gemini, with plans to launch mobile support soon. Initially, the feature will be available to Workspace and Gemini Advanced users, followed by broader access.
Recently, Gemini also introduced a feature for creating presentations based on user prompts.
