AgentIR: Reasoning-Aware Retrieval for Deep Research Agents
This work addresses the problem of improving retrieval accuracy for Deep Research agents by exploiting their reasoning traces, which are ignored by existing retrieval systems. This is an incremental improvement for the field of AI agents.
This paper introduces Reasoning-Aware Retrieval, a new paradigm that leverages an agent's explicit natural language reasoning trace alongside its query to improve retrieval performance for Deep Research agents. The proposed AgentIR-4B model, trained with a novel data synthesis method DR-Synth, achieves 68% accuracy on the BrowseComp-Plus benchmark with the Tongyi-DeepResearch agent, outperforming conventional embedding models (50%) and BM25 (37%).
Deep Research agents are rapidly emerging as primary consumers of modern retrieval systems. Unlike human users who issue and refine queries without documenting their intermediate thought processes, Deep Research agents generate explicit natural language reasoning before each search call, revealing rich intent and contextual information that existing retrievers entirely ignore. To exploit this overlooked signal, we introduce: (1) Reasoning-Aware Retrieval, a retrieval paradigm that jointly embeds the agent's reasoning trace alongside its query; and (2) DR-Synth, a data synthesis method that generates Deep Research retriever training data from standard QA datasets. We demonstrate that both components are independently effective, and their combination yields a trained embedding model, AgentIR-4B, with substantial gains. On the challenging BrowseComp-Plus benchmark, AgentIR-4B achieves 68\% accuracy with the open-weight agent Tongyi-DeepResearch, compared to 50\% with conventional embedding models twice its size, and 37\% with BM25. Code and data are available at: https://texttron.github.io/AgentIR/.