MM-BizRAG: Rethinking Multimodal Retrieval-Augmented Generation for General Purpose Enterprise Q&A

Jul 1, 2026·
Hanoz Bhathena
,
Parin Rajesh Jhaveri
,
Rohan Mittal
,
Prateek Singh
,
Aymen Kallala
,
Rachneet Kaur
Yiqiao Jin
Yiqiao Jin
,
Zhen Zeng
,
Adwait Ratnaparkhi
,
Denis Kochedykov
· 1 min read
Abstract
Enterprise question answering frequently spans heterogeneous modalities — text, tables, charts, scanned documents, and structured databases. We introduce MM-BizRAG, a multimodal retrieval-augmented generation framework designed for general purpose enterprise Q&A. MM-BizRAG combines modality-aware retrieval, structured-context fusion, and grounded generation, and is evaluated on enterprise-realistic workloads.
Type
Publication
Annual Meeting of the Association for Computational Linguistics (ACL) 2026, Industry Track

Abstract

Enterprise question answering frequently spans heterogeneous modalities — text, tables, charts, scanned documents, and structured databases. We introduce MM-BizRAG, a multimodal retrieval-augmented generation framework designed for general purpose enterprise Q&A.

Retrieval-Augmented Generation Multimodal Large Language Models Enterprise AI
Yiqiao Jin
Authors
Yiqiao Jin
Ph.D. Candidate in Computer Science
My research focuses on adaptive and efficient AI systems, with emphasis on LLM agents, agent memory, self-distillation, multimodal LLMs, and structured multi-agent intelligence.