The National Development Bank (formerly the Czech-Moravian Guarantee and Development Bank) provides financial products mainly to small and medium-sized enterprises. The bank offers them discounted loans from the state budget, structural funds and regional funds. StringData was approached by the National Development Bank with a request for the automation and digitisation of administrative operations relating to financial monitoring clients. This requirement involved identifying and extracting data from defined types of financial statements and exporting extracted data to the bank’s internal systems. Thanks to automation, the NDB was able to address insufficient capacity and to scale these administrative operations without the need to increase staffing requirements. At the same time, thanks to automation, we eliminated the risk of errors caused by the manual transcription of large amounts of data.
After an in-depth analysis, specialists at StringData proposed two alternatives to achieve the National Development Bank’s target status, namely the partial or comprehensive E2E automation of all related administrative operations. This included the extraction of data from documents and the engagement of software robots to transfer data to the bank’s internal systems.
StringData’s task was to automate processing financial documents required by the bank from its clients for financial monitoring purposes.
The National Development Bank uses a total of 4 applications for financial monitoring, in which the user must manually perform several individual steps. We resolved the limited or lack of integration between systems very efficiently.
The NDB had forms available on its website into which the bank’s clients could manually transcribe all their details according to their legal form. Unfortunately, it was clear from practice that the willingness of clients to transcribe data into a predefined form was very low and clients preferred to send the bank financial statements generated from their accounting systems, which were then manually processed by the bank’s back-office staff. Transcribing the required data from these documents was time-consuming and posed the risk of increased error for the bank. In view of the fact that this often concerned scanned documents, it would be difficult to machine process them in a conventional manner.
The aim of the project was to automate administrative operations relating to processing financial statements and to propose how to engage robots to transfer the acquired data to the bank’s four internal systems without the need for complex intervention in existing systems.
After a detailed analysis of all administrative operations, we delivered a solution for the automated processing of various types of documents to the client, in combination with robotisation through software robots to ensure the integration of data into the bank’s internal systems.
Thanks to our comprehensive approach, we ensured E2E automation of all administrative operations. Software robots connect the DocumentAnts tool to the bank’s four internal systems at input and output. A robot uploads each document to the DocumentAnts tool, which automatically detects the type of input file. Format is detected by reading the file header, where metadata about the file is stored. Accepted formats are PDF (PDF-data, PDF-image, PDF-scan), Excel (xls, xlsx), Xml-Fú, P7S and ZFO. Image formats (JPEG, PNG) are also accepted, as are compressed formats (ZIP), if they contain the aforementioned types of files.
Based on the detection of format, documents are further filtered into machine-readable and other input file formats. The tool then sends these to the ABBYY OCR layer for conversion into machine-readable input. As an ABBYY solution partner, StringData provided not only the software licence, but also installation and configuration in the NRB environment.
Business checks are an essential part of the automated processing of incoming documents. For example, checking the completeness of documents sent by the client, or checking that the sum of extracted assets is equal to the sum of extracted liabilities. Thanks to business checks, the accuracy of extracted data is verified and validated, so it is no longer necessary to check documents manually. Once all process steps and checks have been completed, the whole case is queued for the robot, which integrates all data into the bank’s systems.
The chosen solution enabled the National Development Bank to scale the entire process without the need to hire additional staff and to process an increasing number of documents with the same team.
The comprehensive solution has several modules to automate and cover required functionalities. In the given context, this largely concerns the extraction of machine-readable files (PDF, Excel and XML) and universal connection to the OCR solution, which ensures the processing of other input file formats. The StringData robotisation team resolved the limited or lacking ability to integrate with third-party systems without the need to invest in the creation of web services or integration bridges. This function was replaced by robotic process automation in the framework of financial monitoring.
The project was implemented by the DocumentAnts team and the StringData robotisation team, which worked closely with representatives of the National Development Bank’s IT and business departments.
We began implementing the solution virtually the moment the order was placed and we received the cooperation needed to begin the project. The overall implementation of the project took place in several stages and took a total of 6 months.
"Automating the processing of incoming documents for financial monitoring enabled the Czech-Moravian Guarantee and Development Bank to scale these administrative operations and process an increasing number of documents with the same team.
The proposed solution processes approximately 100 cases per day, and thanks to the comprehensive approach, the entire process has undergone E2E automation, including the integration of automatically extracted data into the bank’s internal systems using software robots. The benefit for our clients is eliminating the need to manually fill in and process financial statements.”