stack.tools

MUFG Bank

Japan's largest bank, providing capital markets solutions to major corporate clients.

MUFG's AI/ML team built a retrieval and summarization system for the Global Capital Markets Division's FX & Derivative Sales team, pulling the relevant fraction out of 100–200 page annual reports and drafting the client presentations that follow. It prototyped on the Python LangChain stack with Streamlit, using the Retriever interface to compare vector databases cheaply, then moved to the TypeScript version on Next.js for the production application.

bk.mufg.jp
Facts
1
Receipts
2
Layers
1

Sources last verified Sep 7, 2026

Orchestration & Agents

1 fact

LangChain

Powers the production RAG app extracting financial data and drafting presentations.

High confidence

Receipts · 2

Show quotes (2)
  • “Using LangChain, MUFG developed a system to extract critical financial data efficiently” — blog.langchain.dev, Feb 26, 2025
  • “The production RAG application now serves as a knowledge-sharing tool for corporate sales teams” — blog.langchain.dev, Feb 26, 2025

Source last verified Sep 7, 2026