Rosbank connected artificial intelligence to work with customer documents

2020: Introduction of AI systems from Smart Engines for document processing

Rosbank, with the support of experts from Philosophia.IT, has introduced Smart Engines artificial intelligence systems for processing documents necessary to form a client dossier. The solution automates the entry of data from a set of client documents, including identification documents, into a form approved by the bank that meets all the requirements of the law. The technology allows you to reduce the time to open customer accounts from 20 to 5 minutes, Rosbank reported on July 16, 2020.

Integrated recognition technologies allow you to automatically process and enter customer data when opening accounts and performing banking transactions, where identity confirmation is required. To process identity documents, Smart IDReader technology is used, which automatically enters recognized passport data into the corresponding fields of registration forms. Smart DocumentReader technology is used to recognize and automatically enter KYC data and consent to the processing of personal data. Artificial intelligence in two seconds recognizes data from scans and photographs of documents (more than 70 details for each client) and performs about 15 automatic data checks.

The high speed of document recognition and classification is ensured by the use of new architectures of ultralight neural networks. The computational visual memory algorithms used in Smart Engines solutions allow you to correct projective distortion and achieve high quality data recognition even in mid-light photos from webcams and mobile phones with low quality images.

Smart DocumentReader and Smart IDReader technologies are implemented on the basis of GreenOCR's own AI platform, which is focused, among other things, on reducing power consumption by neural networks during training and execution of operations on end devices. GreenOCR allows you to reduce the "dimensions" of the used neural networks and increase their performance.

In the process of document recognition, systems ensure the security of working with personal data, since they do not create copies and do not transfer information for processing to external servers or cloud services. When embedded in mobile applications, deep hardware and software optimization allows you to recognize complex A4 format documents in real time in a few seconds. On the basis of Smart IDReader, automatic recognition of the Russian passport and certification documents of more than 200 countries is implemented.

"Every day, every employee of the bank's front office accepts many documents from current and potential customers. The Bank is obliged to work with documents not only in compliance with all legislative norms, but also in compliance with internal requirements of efficiency, speed and accuracy. The goal of the recognition technology implementation is to comply with the quality of customer data entry, reduce the number of operational errors and, as a result, optimize the entire process. Thanks to Smart IDReader technology, we managed to significantly reduce the time spent on identifying new customers and opening accounts for them, on changing personal data, "said Natalya Voevodina, member of the board, director of operations at Rosbank.

"The presented solution implements a full set of capabilities that artificial intelligence tools allow. The system recognizes in photographs not only typical forms of identity documents, but also flexible forms with "floating fields." Despite the complexity of the task, our specialists have fulfilled all the technical requirements of the customer, and the entire range of services has become available to the bank's customers, "comments the CEO of Smart Engines, named after Vladimir Arlazarov.

"A team of experts has implemented a solution that reduces maintenance time by automating the completion of a customer card based on a passport photo or any other document. Data entry is an important step in the service process, its informatization is a real indicator of the bank's digital transformation. This leads to an increase in the quality of client data by minimizing the human factor. We are glad to be part of such a high-tech project that will bring visual benefit to the customer and his clients, "said Dmitry Siluyanov, project director from Philosophia.IT LLC.