Agentic Information Retrieval

TL;DR

Agentic IR reframes information retrieval as achieving dynamic information states via LLM-driven AI agents, moving beyond static corpus item retrieval. The framework formalizes states, actions, environment dynamics, and a trajectory-based reward to guide users toward target information states. The architecture integrates profile, memory, planning, and action modules, and supports single- and multi-agent designs with multiple optimization techniques. An evaluation protocol based on a verifier reward assesses utility, efficiency, and ethics across interactive trajectories, illustrated through life and business assistant case studies. The work outlines key challenges and future directions toward adaptive, proactive, and task-executing IR systems with broad practical potential.

Abstract

Since the 1970s, information retrieval (IR) has long been defined as the process of acquiring relevant information items from a pre-defined corpus to satisfy user information needs. Traditional IR systems, while effective in domains like web search, are constrained by their reliance on static, pre-defined information items. To this end, this paper introduces agentic information retrieval (Agentic IR), a transformative next-generation paradigm for IR driven by large language models (LLMs) and AI agents. The central shift in agentic IR is the evolving definition of ``information'' from static, pre-defined information items to dynamic, context-dependent information states. Information state refers to a particular information context that the user is right in within a dynamic environment, encompassing not only the acquired information items but also real-time user preferences, contextual factors, and decision-making processes. In such a way, traditional information retrieval, focused on acquiring relevant information items based on user queries, can be naturally extended to achieving the target information state given the user instruction, which thereby defines the agentic information retrieval. We systematically discuss agentic IR from various aspects, i.e., task formulation, architecture, evaluation, case studies, as well as challenges and future prospects. We believe that the concept of agentic IR introduced in this paper not only broadens the scope of information retrieval research but also lays the foundation for a more adaptive, interactive, and intelligent next-generation IR paradigm.

Figures (5)

• Figure 1: The next paradigm shifts from (a) traditional information retrieval to (b) agentic information retrieval. We illustrate the core processes of both traditional IR and agentic IR with the example of online travel agency.
• Figure 2: Four core components of the AI agent.
• Figure 3: The design of single-agent and multi-agent systems for agentic information retrieval.
• Figure 4: The case study of agentic IR in life assistant scenarios, which conveys four key features.
• Figure 5: The case study of agentic IR in business assistant scenarios, which consists of four key stages.