Deep Research Agents: A Systematic Roadmap for LLM-Based Autonomous Research Systems

Deep Research Agents (DR agents), developed through a collaboration between the University of Liverpool, Huawei Noahs Ark Lab, Oxford, and University College London, represent a significant advancement in autonomous research systems powered by Large Language Models (LLMs). Unlike traditional retrieval-augmented generation (RAG) models or static tool-use frameworks, DR agents are engineered to perform complex, long-horizon tasks that demand dynamic reasoning, adaptive planning, and iterative tool integration, including structured API use and browser-based retrieval. This enables DR agents to navigate evolving user intents and ambiguous information landscapes more effectively, addressing key limitations of prior