Background
A global food and beverage giant with a complex retail supply chain was experiencing operational inefficiencies due to fines and delivery-related disputes. Manual review of emails and attachments linked to purchase orders (POs) resulted in high effort, delayed resolutions, and increased compliance risk.
Solution
Affine implemented an AI-powered email parsing and validation engine to automate fines dispute workflows: Leveraged Azure OpenAI (GPT models with custom prompt engineering) to extract PO details, delivery acknowledgments, and exception indicators from emails and attachments (.xls, .eml) Built a recursive content pipeline to process both structured and unstructured formats across diverse customer communications Validated extracted fields and stored them in Snowflake for downstream analysis Integrated auto-verification logic to detect hallucinations and flag data inconsistencies Delivered outputs via dashboards and reports to supply chain and finance teams for faster decision-making
Impact
80% reduction in manual effort for email and fine validation Faster and more accurate dispute resolution Improved audit readiness and full PO lifecycle transparency Lower compliance risk through consistent, automated workflows