The customer, a leading low-cost airline in East Asia, was grappling with the challenge of processing approximately 2,000 invoices each month, each ranging from 100 to 200 pages in length. The invoices, coming from over a thousand suppliers, arrived in a wide array of formats. The task of manually extracting data from each invoice and processing it into the airline's BPM system was repetitive and time-consuming. The addition of new destinations only increased the volume of suppliers and invoices, further complicating the process. The airline's pre-existing OCR platform struggled with the variability of the invoice formats and required constant intervention. The frequent changes to these documents made a template-based automation approach difficult. The company was in need of an AI-based solution that didn't require data science expertise.

The client, a top asset manager based in the UK, was grappling with the challenge of manually processing a substantial volume of paperwork, which was both costly and prone to errors. The company decided to start with invoice processing, a task that involved dealing with over 750 suppliers generating more than 19,000 invoices per annum. These invoices were mostly in English, but also in Italian, Portuguese, Dutch, and Polish. The client had previously used in-house technology based on OCR, but it was not delivering the desired results. The company was seeking a third-party technology that could automate its invoice processing end-to-end, deliver higher levels of accuracy, and minimize manual work so that it could focus purely on quality control.