Source: Huang et al. 2025: Deep Research Agents survey
Survey defining deep research agents and classifying them by workflow, agent composition, planning and API- versus browser-based information access.
Contents
- What: survey paper, arXiv:2506.18096
- Author: Yuxuan Huang, Yihang Chen, Haozheng Zhang, Kang Li, Huichi Zhou, Meng Fang, Linyi Yang, Xiaoguang Li, Lifeng Shang, Songcen Xu, Jianye Hao, Kun Shao, Jun Wang
- Date: 2025-06-22 (v2 2025-09-03)
- Original: arXiv:2506.18096
Summary
Systematises the deep research agent space and its open problems. Read at abstract level.
Key claims
- Defines deep research agents as LLM-based systems that combine dynamic reasoning, long-horizon planning, multi-hop retrieval, iterative tool use and structured report generation.
- Taxonomy axes: static versus dynamic workflows; single-agent versus multi-agent; planning strategy; API-based retrieval versus browser-based exploration.
- Also covers tool frameworks: code execution, multimodal input, MCP.
- Challenges: limited access to external knowledge, inefficiency of sequential execution, and evaluation metrics and benchmarks misaligned with practical goals.