The Transaction Enrichment Engine is a multi-layered system developed by Plaid to process and enrich over 500 million new transactions daily for leading fintechs, enterprises, and financial institutions. The engine uses a combination of rule-based approaches, machine learning models, and named entity recognition techniques to extract valuable information from transaction descriptions, including merchant names, locations, and categories. The system also leverages an internal Knowledge Base to deliver comprehensive insights beyond what is available in the transaction string itself. By applying a layered approach with multiple methods, Plaid strikes a balance among performance, accuracy, and ease of use, achieving over 90% coverage for merchants involved in transactions while maintaining 99% precision. The engine also provides advanced insights such as Personal Finance Category (PFC) predictions, enabling customers to understand where users are spending their money. With ongoing improvements and new product features on the roadmap, Plaid's Transaction Enrichment Engine is poised to continue delivering sustainable high-quality enriched transactions that best serve customers' needs.