A massive win for data processing accuracy and speed

One of the largest cities in the US was struggling to process more than 200,000 multipage, handwritten paper housing registration forms each month. Their existing data capture partner scanned the forms and extracted handwritten data using single-pass verification, but typically only achieved 65% accuracy. A team of 4-6 full-time city employees worked to fix the errors and key in correct information from captured images across six different types of housing forms. The city fell weeks behind in their form reconciliations, leading to fines being levied on landlords for overdue documents. Disgruntled landlords who had submitted their forms on time called to challenge these fines, creating additional customer service burden for the city.

FormedAI went head-to-head against the city’s existing provider and achieved 99.8% accuracy. Using a sophisticated OCR/ICR engine to lift printed and handwritten data — even unstructured data — and pairing it with a double-blind human verification process, FormedAI has consistently achieved greater than 99% accuracy since the city transitioned the business away from its previous provider. Documents are automatically routed to one of six different departments based on classification rules. The city’s throughput skyrocketed and they were able to reconcile reports in under 48 hours through automated data posting, reducing landlord complaints and incorrect fine rollbacks. City employees can spend time on higher-value tasks instead of data error resolution.