How MUFG Bank aims to cut flowchart work by up to 90% with ontology-grounded AI on Amazon Neptune and Amazon Bedrock

MUFG Bank built a custom AI Flowchart application that grounds generative AI with an ontology and knowledge graph layer on Amazon Neptune and Amazon Bedrock, addressing hallucination problems that made a pure-LLM proof of concept unusable. The system models business concepts as RDF/OWL triples, stores branch-level procedure instances as knowledge graphs in Neptune, and uses the HermiT reasoner alongside a Strands Agents-based agent to constrain and ground flowchart generation. Results cut the system-development effort behind each branch-procedure standardization from 11 person-days to 1 (90% reduction), with flowchart creation itself down 95% and comparisons/revisions dropping from person-days to minutes. MUFG is now extending the pattern to UAT test-case generation, requirements documents, knowledge transfer, M&A due diligence, and new-branch openings, partly using AWS's open-source Context Ontology Accelerator.

Nguồn: https://aws.amazon.com/blogs/database/how-mufg-bank-aims-to-cut-flowchart-work-by-up-to-90-with-ontology-grounded-ai-on-amazon-neptune-and-amazon-bedrock. 8 Sync News chỉ tóm tắt và dẫn link; bản quyền nội dung thuộc tác giả và nguồn gốc.