Towards Retrieving Interaction Spaces for Agentic Search
This work addresses the scalability problem of direct corpus interaction for search agents, proposing a new retrieval paradigm that balances exploration and efficiency.
The paper argues that retrieval for agentic search should construct an interaction space (a bounded subset of the corpus) rather than just selecting documents. Their proposed method, RISE, uses BM25 to build this space and processes documents for shell-style navigation, achieving 78% accuracy on BrowseComp-Plus with gpt-5.4-mini at one quarter the cost of a pure-shell baseline, and 81% accuracy at 1M documents while the baseline degrades to 60% with failures.
Retrieval for search agents is still inherited from non-agentic information retrieval: a retriever ranks the corpus and the agent reads a small set of returned documents. Recent direct corpus interaction (DCI) work shows that agents can instead interact with the raw corpus through shell tools such as grep and file reads. But unbounded interaction does not scale: every broad shell command is a scan over the whole corpus, and latency degrades sharply as the corpus grows. We argue that the role of retrieval for agentic search is not just to select documents that fit in the LLM context window, but to construct an interaction space: a bounded subset of the corpus the agent can explore with associated tools. Two design consequences follow. The space needs a boundary supplied by retrieval, and the objects within it should be processed for interaction. As a proof of concept, we propose RISE (Retrieving Interaction SpacE): we use BM25 to construct the interaction space; meanwhile, its documents are processed during indexing for shell-style navigation. On BrowseComp-Plus, RISE matches the pure-shell DCI baseline at 78% accuracy with gpt-5.4-mini at roughly one quarter of the per-query cost. At 1M documents, RISE-BM25 reaches 81% on gpt-5.4-mini, whereas DCI on gpt-5.4-nano degrades to 60% with 33 of 100 wall-clock failures.