Lessons from Building Deep Research Agents in Production - Sarang Kulkarni

This talk explores the complexities of deep research, focusing on the challenges of data scattering, insight extraction, and the limitations of typical LLM demos. The speaker outlines the design and deployment of a production-grade multi-agent research assistant capable of deep exploration and long-form reasoning in enterprise settings, ensuring privacy and data governance. Attendees will gain insights into context engineering techniques, such as selective retrieval and context routing, as well as reasoning strategies that streamline complex research workflows. The session highlights advancements that can reduce research discovery cycles significantly.