Firedesktop has developed a solution to automatically collect data from bundles of documents relating to deteriorated bank loans or NPL.
The solution integrates Flowbot® platform with Ethymo® search engine, in order to extract data from documents and send them to the customer data warehouse.

This robot reads information from files containing scans of documents from single unstructured or semi-structured records (contracts, judicial acts, …) and transforms them into structured data, usable both in Machine Learning processes, dedicated to decision-making support, and in the internal Bank workflow.
In this way, not only the Bank drastically reduces manual control of documents, but employees can carry out greater added value activities, bringing benefits and efficiency to the company.

Problem Statement

The goal is to quickly collect and extract valuable information hidden in documents relating to non-performing bank loans through software robots and send them to the Customer’s data warehouse for the related decision-making processes.

Value Delivered

The manual control of documents made by people has been practically eliminated, moving them from activities with very low added value and routine to activities with high added value, allowing with the same number of resources in the year before carrying out more activities and bringing benefits and efficiency to the company

The system performs in an outstanding way: the checks on NPLs that were previously carried out by about fifty people in three months, are now completed in just three weeks by two full-time people and a part-time supervisor.