# RouteDRAG: Learning to Route and Control Agentic Execution for Multimodal Document RAG

DevFeed: [RouteDRAG: Learning to Route and Control Agentic Execution for Multimodal Document RAG](<https://devfeed.tech/articles/routedrag-learning-to-route-and-control-agentic-execution-for-multimodal-document-rag-103134.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.36444>)

Author: Zekai Chen, Hao Zhu, Yu Liu, Xunkai Li, Zejian Kang, Rui Qian, Quanjiang Li, Rong-Hua Li

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [execution](<https://devfeed.tech/tags/execution.md>), [learning](<https://devfeed.tech/tags/learning.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [rag](<https://devfeed.tech/tags/rag.md>), [route](<https://devfeed.tech/tags/route.md>)

## AI overview

RouteDRAG learns when to invoke agentic execution for multimodal document question answering and how to control its subsequent tool use and termination. Its graph-backed Unified Document Index derives query representations from initial retrieval; Adaptive Path Routing estimates whether an agent is likely to help over a shorter path while preserving exact answering; Dynamic Agent Execution maintains a workspace and refines planner proposals; and Sequential Policy Learning trains operation control before path routing. The abstract reports a 3.47 percentage-point average accuracy gain over shorter-path-only execution across two benchmarks, with agents assigned to 79.3% of queries. RouteDRAG also reportedly outperforms uniform agent execution on both benchmarks while requiring fewer answering calls. These results are limited to the two benchmarks described in the abstract.

## Source excerpt

arXiv:2610.36444v1 Announce Type: new Abstract: Web-based applications increasingly rely on understanding long multimodal documents, where answering questions often requires combining information across pages, tables, and figures. Agentic retrieval-augmented generation addresses these tasks through iterative retrieval and reasoning. However, Agentic execution is not always necessary, introducing additional computation and potentially incorrect answers. Our empirical analysis reveals that Agentic execution provides query-dependent benefits, while additional operations may improve, preserve, or degrade answer accuracy. These findings motivate controlling both Agent invocation and subsequent execution. To address these challenges, we propose RouteDRAG, a learning-based framework comprising four components. Unified Document Index organizes multimodal content into a graph-backed index and derives query representations from initial retrieval. Adaptive Path Routing preserves exact answering and predicts the relative benefit of Agent over shorter execution paths. Dynamic Agent Execution maintains an evolving workspace and refines Planner proposals to control tool use and termination. Sequential Policy Learning first trains operation control from observed outcomes and then trains path routing using the resulting Agent's performance. Across two benchmarks, RouteDRAG achieves an average accuracy improvement of 3.47 percentage points over shorter-path-only execution while assigning only 79.3% of queries to Agent. Notably, it also surpasses uniform Agent execution on both benchmarks while requiring fewer answering calls. Our code is now available at https://github.com/Ereshkigal1/RouteDRAG.