RESEARCH · GLOBAL
arXiv: Bridging language models and financial analysis—survey of multimodal finance AI
Comprehensive survey of LLM applications in finance. Identifies critical gap: financial data is inherently multimodal (text earnings calls, tables, charts), but LLM adoption in BFSI lags. Paper maps techniques for processing heterogeneous data (text + tables + visuals) using transformers and multimodal models.
WHY IT MATTERS
Highlights adoption gap between frontier LLM capabilities and actual BFSI deployment. Many BFSI systems still treat financial documents as text-only. Roadmap for multimodal AI in underwriting, trading, and risk assessment—key competitive advantage for early movers.
Source: arXiv q-fin · 2026-05-21