WebSwarm: Recursive Multi-Agent Orchestration for Deep-and-Wide Web Search

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Xiaoshuai Song, Liancheng Zhang, Kangzhi Zhao, Yutao Zhu, Zhongyuan Wang, Guanting Dong, Jinghan Yang, Han Li, Kun Gai, Ji-Rong Wen, Zhicheng Dou
• Venue: arXiv: cs.CL
• Published: 2026-07-09
• Collected: 2026-07-10
• URL: http://arxiv.org/abs/2607.08662v1
• PDF: https://arxiv.org/pdf/2607.08662v1
• DOI: UNKNOWN

Plain-Language Brief

The paper proposes a new framework called WebSwarm for multi-agent search systems. This system aims to improve deep-and-wide web searches by dynamically creating search nodes that can delegate tasks recursively and collaborate with each other. WebSwarm outperforms single-agent and existing multi-agent baselines in various experiments.

What the Document Says

• The text discusses a framework named WebSwarm for large language model (LLM)-based web search agents.
• WebSwarm is designed to handle complex, deep-and-wide search tasks that require simultaneous depth and coverage.
• A single agent has limitations in terms of one long trajectory and limited context, which WebSwarm aims to address.
• Existing multi-agent systems have limitations in recursive depth, collaboration adaptability, and evidence-grounded expansion.
• WebSwarm constructs task decomposition, recursive expansion, and agent collaboration during inference.
• Each node in the system can solve its objective or delegate child nodes.
• Nodes return evidence and results upward to enable parent nodes to expand, revise, or aggregate the search process.
• The system first probes how task-relevant information is organized on the web to guide subsequent node expansion.
• WebSwarm reuses process-level experience across homogeneous sibling nodes.
• Experiments show that WebSwarm outperforms single-agent and multi-agent baselines on deep, wide, and interleaved deep-and-wide tasks.
• Further analyses explain WebSwarm's effectiveness and provide insights for multi-agent search systems.

Key Concepts Mentioned

• WebSwarm
• large language model (LLM)-based web search agents
• deep-and-wide search
• multi-agent systems
• task decomposition
• recursive expansion
• agent collaboration
• search nodes
• evidence-grounded expansion
• homogeneous sibling nodes

Claims Made by the Document

• WebSwarm outperforms single-agent and multi-agent baselines.

Evidence Used

• Experiments on BrowseComp-Plus, WideSearch, DeepWideSearch, and GISA show that WebSwarm consistently outperforms single-agent and multi-agent baselines.

Methods / Approach

• dynamically instantiating agentic search nodes
• coupling a local objective with a search mode
• delegation of child nodes
• returning evidence and results upward
• probing how task-relevant information is organized on the web
• reusing process-level experience across homogeneous sibling nodes

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

The text does not provide information about potential limitations of the WebSwarm framework as outlined by the agent.

Open Questions

• What are the specific details of the experiments conducted?
• How does WebSwarm handle task-relevant information organization on the web?
• What is the process-level experience reuse across homogeneous sibling nodes?

Notes for Future Q&A

May include interpretation. Labeled where it does.

The text discusses experiments but does not provide specific details. It also mentions reusing process-level experience and probing how task-relevant information is organized on the web, but no further explanation is given.

Raw Abstract / Excerpt

Large language model (LLM)-based web search agents are transforming information seeking from simple factoid question answering into complex, deep-and-wide search and research-oriented tasks. A single ReAct-style agent is constrained by one long trajectory and limited context, making it difficult to handle depth and coverage simultaneously. Existing multi-agent systems improve search coverage through parallel execution and aggregation, but still exhibit clear limitations in recursive depth, collaboration adaptability, and evidence-grounded expansion. We propose WebSwarm, a progressive recursive delegation framework that jointly constructs task decomposition, recursive expansion, and agent collaboration during inference. WebSwarm dynamically instantiates agentic search nodes, each coupling a local objective with a search mode that specifies how the node should organize search and collaboration. Each node can either solve its objective itself or further delegate child nodes; after solving, it returns evidence and results upward, enabling parent nodes to further expand, revise, or aggregate the search process. To guide this process, WebSwarm first probes how task-relevant information is organized on the web to ground subsequent node expansion, and reuses process-level experience across homogeneous sibling nodes. Experiments on BrowseComp-Plus, WideSearch, DeepWideSearch, and GISA show that WebSwarm consistently outperforms single-agent and multi-agent baselines on deep, wide, and interleaved deep-and-wide tasks. Further analyses of ablation, task difficulty, web tool efficiency, and model generalization explain WebSwarm's effectiveness and provide insights for multi-agent search systems.