Toward Agentic RAG for Ukrainian

TL;DR

This paper investigates agentic retrieval-augmented generation for Ukrainian, highlighting retrieval quality as a key bottleneck and proposing a system combining two-stage retrieval with an agentic layer.

Contribution

It introduces a novel system combining two-stage retrieval with an agentic layer for Ukrainian, analyzing its limitations and potential improvements.

Findings

Retrieval quality is the main bottleneck for system performance.

Agentic retry mechanisms improve answer accuracy.

Overall scores are limited by document and page identification.

Abstract

We present an initial investigation into Agentic Retrieval-Augmented Generation (RAG) for Ukrainian, conducted within the UNLP 2026 Shared Task on Multi-Domain Document Understanding. Our system combines two-stage retrieval (BGE-M3 with BGE reranking) with a lightweight agentic layer performing query rephrasing and answer-retry loops on top of Qwen2.5-3B-Instruct. Our analysis reveals that retrieval quality is the primary bottleneck: agentic retry mechanisms improve answer accuracy but the overall score remains constrained by document and page identification. We discuss practical limitations of offline agentic pipelines and outline directions for combining stronger retrieval with more advanced agentic reasoning for Ukrainian.