CoSearchAgent: A Lightweight Collaborative Search Agent with Large Language Models
Authors: Peiyuan Gong, Jiamian Li, Jiaxin Mao Published: 2024-02-09 Source: https://arxiv.org/abs/2402.06360 Type: Academic demo paper (Conversational Search)
TL;DR
CoSearchAgent is an LLM-powered collaborative search agent implemented as a Slack plugin, enabling multiple users in a group conversation to search the web together, receive grounded answers, and get clarifying questions when their queries are ambiguous.
Background and Motivation
Collaborative search supports multiple users working together to accomplish a specific search task. Research has found that lightweight collaborative search plugins embedded in instant messaging platforms align better with users' collaborative habits. However, due to the complexity of multi-user interaction scenarios, building a fully functioning lightweight collaborative search system is difficult — so previous studies had to rely on the Wizard of Oz paradigm (human-simulated system behavior).
With LLMs now demonstrating natural interaction and the ability to complete complex information-seeking tasks via agent frameworks, the authors propose CoSearchAgent to make real, deployable collaborative search practical and to support further research in this area.
What CoSearchAgent Does
- Slack plugin: Operates within multi-party conversations on Slack, fitting users' existing collaborative workflows.
- Multi-user context understanding: Comprehends queries and conversational context involving multiple users.
- Web search via APIs: Retrieves relevant information from the web to ground its answers.
- Clarification: Asks clarifying questions when user information needs are unclear.
- Flexibility: Designed as a flexible platform to support further research on collaborative search.
- Moves lightweight collaborative search from simulated (Wizard of Oz) setups to a working, LLM-driven system.
- Demonstrates how LLM-based agents can handle the added complexity of multi-user information-seeking, beyond single-user conversational search.
- Provides an open, extensible research substrate — code and a demo video are publicly accessible.
- A Survey of Conversational Search (ACM, Sep 2025)
- Engineering Conversational Search Systems: A Review of Applications (arXiv:2407.00997)
- ConvGQR: Generative Query Reformulation for Conversational Search (arXiv:2305.15645)
- ChatRetriever: Adapting Large Language Models for Generalized and Robust Retrieval (arXiv:2404.13556)
Significance
Original Abstract
> Collaborative search supports multiple users working together to accomplish a specific search task. Research has found that designing lightweight collaborative search plugins within instant messaging platforms aligns better with users' collaborative habits. However, due to the complexity of multi-user interaction scenarios, it is challenging to implement a fully functioning lightweight collaborative search system. Therefore, previous studies on lightweight collaborative search had to rely on the Wizard of Oz paradigm. In recent years, large language models (LLMs) have been demonstrated to interact naturally with users and achieve complex information-seeking tasks through LLM-based agents. Hence, to better support the research in collaborative search, in this demo, we propose CoSearchAgent, a lightweight collaborative search agent powered by LLMs. CoSearchAgent is designed as a Slack plugin that can support collaborative search during multi-party conversations on this platform. Equipped with the capacity to understand the queries and context in multi-user conversations and the ability to search the Web for relevant information via APIs, CoSearchAgent can respond to user queries with answers grounded on the relevant search results. It can also ask clarifying questions when the information needs are unclear. The proposed CoSearchAgent is highly flexible and would be useful for supporting further research on collaborative search. The code and demo video are accessible.