Accounts receivable

• ML matching of incoming payments to open invoices, credits and adjustments, with confidence thresholds and a review queue for the rest: 60% of payments allocated automatically, $2.5bn a year.
• Email ingestion and document parsing for remittance advice in whatever format it arrives: free text, attachments, spreadsheets, scans.
• OCR for printed and handwritten amounts on scanned checks, across regional layouts: 10,000+ checks read, cursive included.
• Search over remittances, invoices and payments, full text and fuzzy references, as the retrieval layer the matching engine and the review queue share: document search 30× faster.
• Receivables analytics on a columnar store, wide pre-joined tables keyed by tenant, entity and period, so enterprise tenants with millions of invoices get ageing, DSO and collections in seconds.
• Change-data pipelines (Debezium, Kafka) from the transactional database into the search and analytics stores, so nothing is re-keyed and the application database carries no reporting load.

All four cases below come from one accounts-receivable product: 60% of payments allocated automatically, $2.5bn a year matched, 10,000+ checks read including handwriting, document search 30× faster, enterprise analytics on a columnar store. The audit starts with where your team still matches by hand and what the data looks like when it arrives.