IndexAct improves agent search with smarter candidate refinement

From Delivery to Stateful Exploration: Rethinking the Index for Agentic Search

Information Retrieval

Summary

When searching through lots of documents using language models, it's tricky to decide which pieces to look at next. The authors propose IndexAct, which helps search agents narrow down their choices without always needing to read all matches. Instead, agents work with sets of possible candidates and see helpful stats about them before digging deeper. This approach led to better performance on several tasks where the agent has to find and combine evidence from many documents.

What this means in practice

  • For enterprise search teams: Enable search agents to refine document sets more efficiently, reducing unnecessary passage readings in large corporate knowledge bases.
  • For software developers: Integrate IndexAct-like interfaces to improve layered search tasks, especially those requiring multi-step reasoning across many texts.

Authors

Deogyong Kim, Sunghwan Kim, Sangam Lee, Wonjae Lee, Dongha Lee

Abstract

Recent advances in agentic search have given large language model (LLM) agents finer control over corpus exploration. However, search interfaces often return matching passages even when feedback about the candidate set would suffice for the next decision, coupling candidate refinement with source-text exposure. We propose IndexAct, an interface for Index-Native Corpus Interaction that separates candidate-set refinement from text inspection. Agents construct and manipulate persistent candidate sets through lexical conditions and set operations over an inverted index, receiving reusable state references and statistics such as candidate counts rather than matching passages. This feedback guides further refinement, while separately requested passages provide new clues or evidence that can inform subsequent operations on retained candidate sets. Experiments on five benchmarks spanning agentic search and multi-hop question answering show that IndexAct outperforms the evaluated baselines on each benchmark. On BrowseComp-Plus, it also achieves higher evidence coverage with a smaller average live context than terminal-based corpus interfaces, and maintains answer accuracy as the corpus expands. Further analyses suggest that informative refinement feedback and state reuse support continued evidence discovery, while shorter contexts or fewer search steps alone do not ensure better performance.