The service check is designed to provide quick, secure and unbureaucratic payment and insurance for marginal employment. However, processing the large volume of submissions presents a massive challenge with high time requirements. That’s why the Insurance Institution for Railways and Mining (then VAEB, now BVAEB) was looking for a solution that would automatically process documents.
The solution – fully automatic processing
In cooperation with Kapsch and the Insurance Institution for Railways and Mining, we have further developed the processing of service checks. The solution, which is being used in the healthcare sector for the first time, is based on Artificial Intelligence with LinkThat PRISM. This allows handwritten submissions to be processed automatically.
Learn more about the project and the benefits in this video:
350,330 service vouchers…
…were sold in 2018. Manual processing of a service check used to include the following:
- the opening and scanning of the mail
- the delivery to the agents
- further verification of the information for accuracy
- entering the data into the back-end systems
- the payment to the employee
Four of these five steps are now fully automatic processing in most cases.
“Because the service check has been well received by the public. The number of forms that need to be processed has increased significantly. So the goal of the project at VAEB was therefore primarily to reduce the workload of employees and to reduce the processing time between submission and payment,” recalls Manfred Pigl.
The project implemented by LinkThat included the provision and implementation in the workflow of an appliance that, based on AI technologies, enables the automated reading of information from the DLS with high quality. Of particular note is the appliance’s ability to read handwritten input.
The interplay between people and technology
Noman Khairul, head of software development at LinkThat, describes it as follows:
“Incorrect service checks continue to be reprocessed manually. For example, if the date is filled in incorrectly, then crossed out and the correct date is written over it. In this case our AI recognizes that as an error and forwards it to staff responsible for it.” The majority of submissions, however, can be processed fully automatically, he says: “As a rule, the information can be read very well and without errors. Our AI system is also continuously trained on known errors, so it can recognize and correct them or act accordingly.”
Greatly reduced processing times
The implementation of this project has made it possible to process the steadily increasing number of submitted DLSs with high quality without expanding the existing team. Therefore time from submission to payment has been significantly reduced. From a few weeks to a few working days.