AI use cases for Finance & Accounting
7 documented implementations
EXL Transforms Insurance Underwriting with Generative AI Assistant Built on Amazon Bedrock
EXL, a global data analytics and digital solutions company, built LDS Underwriting Assist, a retrieval-augmented generation chatbot integrated into its Life Digital Suite (LDS) platform to help insurers streamline the underwriting assessment and review stages that previously required underwriters to manually review hundreds of pages of documents. The service uses Anthropic Claude 3 Sonnet on Amazon Bedrock with Amazon Kendra as the chatbot interface, Amazon Textract to extract data from scanned documents, and Amazon Comprehend to analyze text and redact PII and protected health information for compliance with regulations such as India's PII guidelines. EXL tested multiple foundation models (Mistral AI, Claude 2, Sonnet 3, Amazon Titan) to minimize hallucinations before launching the service in August 2024, just 60 days after starting development. EXL reports the solution reduces underwriting processing time from several days to a few hours and has the potential to cut underwriting costs by up to 80%.
EXLAmazon Bedrock · Anthropic Claude 3 Sonnet · Amazon Kendra +2
Empowering Employees to Work Strategically Using Amazon Bedrock with BDM
Big Data Mining (BDM), a Brazilian company, built LOUIS, a bespoke generative AI model on Amazon Bedrock combining computer vision and natural language processing to process complex, unstructured corporate documents (contracts, powers of attorney) across more than 100 models and 15 industries. A large Brazilian financial institution deployed LOUIS to replace a team of 150 professionals manually evaluating over 40,000 unstructured processes per month for opening legal-entity accounts and contracting credit, standardizing interpretation criteria and reducing operational effort by more than 40%, an estimated $4.2 million in cost savings over five years. In insurance, LOUIS reduced life insurance claim processing from up to six months to just a few minutes by capturing required data in about 60 seconds. BDM reports documents are processed 85% faster than manual interpretation with 98% accuracy, and saved roughly 50% on development costs by building on Amazon Bedrock rather than writing the application from scratch.
Transforming Data Into Member Value: Building a Modern Data Ecosystem
Nationwide Building Society, the UK's largest building society managing £367.9 billion in assets, implemented the Databricks Data + AI Platform in 12 months to consolidate 17 of its 19 data warehouses into a single governed lakehouse on Azure. The platform uses Unity Catalog for governance across more than 5,000 internal users, MLflow for model management, and AI/BI Genie for natural-language queries, supporting personalization, fraud prevention and regulatory reporting for over 24.5 million customers.
Nationwide Building Society· United KingdomUnity Catalog · MLflow · Delta Lake +2Dongwon Improves Price Forecasting and Inventory Efficiency
Dongwon Group, a South Korean seafood, food, packaging and logistics conglomerate, adopted the Databricks Data + AI Platform to unify previously siloed data across business units and enable enterprise-wide AI-driven decision-making. Using Delta Lake and Unity Catalog, Dongwon built a raw material price forecasting model that reached approximately 96% accuracy and an inventory optimization system that saves billions of won per month, while AI/BI Genie and Databricks Assistant enabled natural-language data exploration for non-technical staff.
Keeping Every Patient Healthier as Fast as Care Demands
Duly Health and Care, the largest multi-specialty physician-owned practice in the U.S., migrated from a 17-year-old SQL Server warehouse to the Databricks Data + AI Platform, unifying nearly 60 data sources including Epic and Workday via Delta Lake and Unity Catalog. Duly's data science team used MLflow and AutoML to build a hospital readmission prediction model, comparing Random Forest, LightGBM and XGBoost algorithms, trained on 100,000 historical patients and achieving a ROC AUC of 89.2%. The company also built a conversational "Physician Finder" tool integrated into its Epic workspace, and five Genie Agents (Patient, Provider, Finance, HR, Appointments) letting non-technical staff query live data in plain English. Duly reports a 29% drop in hospital readmission rates (14% to 10%), saving an estimated $4.5 million annually, a 30% reduction in data platform costs versus SQL Server, and roughly double the number of data projects completed in 2025 versus 2024 with the same headcount.
Duly Health and CareDatabricks SQL · Delta Lake · Unity Catalog +3
FRoSTA: Pioneering touchless invoice processing and increasing the automation rate in the procure-to-pay process
FRoSTA AG, a frozen food manufacturer with about 1,850 employees and production facilities in Germany and Poland, automated its supplier invoice processing using SAP Build Process Automation and the premium, LLM-based edition of SAP Document AI on SAP Business Technology Platform. Previously, each of FRoSTA's roughly 30,000 annual invoices for goods received was manually touched at least three times, with approvals delayed by up to two weeks when managers were traveling. The new automation monitors incoming documents, uses an LLM in SAP Document AI to extract and classify invoice data regardless of format or language, validates it against SAP and third-party systems, and creates the invoice in SAP S/4HANA Cloud Private Edition, routing exceptions to an approval workflow accessible via the Joule Work mobile app. It now takes a minute to process an invoice, from arrival to posting, 70% of automated invoices are booked without any human touch, and about 60% of the 30,000 annual goods-received invoices are processed automatically with no human intervention. The project went from ideation to go-live in three months, built primarily by an IT trainee with no programming background, with implementation support from SAP partners sovanta AG and Amista.
FRoSTA AG· GermanySAP Build Process Automation · SAP Document AI · SAP S/4HANA Cloud Private Edition +2
De Agostini automates invoice processing with SAP Document AI
De Agostini Publishing uses SAP Document AI to automate purchase order and invoice processing. The company now saves around 500 hours per month, and more than 91 percent of purchase order-referred invoices are automatically processed without manual handling.
De AgostiniSAP Document AI