Details
- Google released two new autonomous research agents built on Gemini 3.1 Pro via the Interactions API: Deep Research (optimized for speed and lower latency) and Deep Research Max (designed for comprehensive, asynchronous workflows with extended reasoning).
- Deep Research now integrates Model Context Protocol (MCP) support, enabling secure connection to custom data streams and professional repositories such as financial or market data providers, alongside web search and file uploads.
- The agents natively generate charts, infographics, and multimodal outputs (HTML, Nano Banana) and accept PDFs, CSVs, images, audio, and video as research inputs, transforming Deep Research from a web summarizer into a specialized autonomous agent.
- New workflow controls include collaborative planning (review and refine research plans before execution), real-time streaming of intermediate reasoning, and the ability to combine Google Search, code execution, and file search simultaneously or disable web access to search only proprietary data.
- Google is partnering with FactSet, S&P, and PitchBook to integrate financial data via MCP servers; Deep Research Max reportedly consults significantly more sources and identifies nuances that earlier versions missed, delivering expert-grade analysis for regulated fields like finance and life sciences.
Impact
This release positions Deep Research as a foundational layer for enterprise agentic workflows, particularly in finance and life sciences where regulatory rigor and data integration are critical. MCP support and native visualizations enable developers to build professional-grade research pipelines that blend public web data with proprietary sources in a single API call, reducing time and cost for due diligence, market analysis, and compliance workflows. Over the next 12–24 months, expect accelerated adoption among regulated enterprises and specialized data providers, with potential to reshape how financial analysts, researchers, and compliance teams conduct investigations.
