On a daily basis, AP Pension receives enormous amounts of documentation in connection with claims handling for customers, often totalling more than 100 pages per case. The documentation is often unsorted and contains information that is not relevant to the case in question. In fact, rarely more than 10% of the information is relevant.

In the past, claims handlers have therefore spent a lot of time manually categorising these documents and identifying the relevant documentation for the individual citizen's claim. For AP Pension, this also meant that the work process was slow and time-consuming. Therefore, the company started looking for a more innovative solution and ended up investing in the digital tool MaIA.

The solution: A digital sort with superpowers delivered by AI

MaIA can best be described as a digital sorting with superpowers provided by artificial intelligence that both provides caseworkers with an easier and clearer user interface and, once implemented, can present the documentation using a combination of historical data and machine learning.

The implementation of MaIA and the digital sorting was done in two phases.  

Phase 1 was to manually train the machine learning component of MaIA. During the 6-month training period, the caseworker uploaded case-relevant PDFs to MaIA and then manually categorised the pages into three categories. This made MaIA smarter, and after phase 1, the solution was able to automatically presort the documents uploaded by the caseworker.

In phase 2, MaIA's algorithm was further refined using a validation module. Now, the case manager is presented with MaIA's bid for sorting a specific document, with only a few pages for manual sorting.

To address the potential risks that come with storing sensitive personal data in a cloud solution, MaIA also includes an anonymisation engine trained in the Danish language. Its primary purpose is to encrypt sensitive personal data before it is uploaded to the cloud.

The result: AP Pension has streamlined over 2,000 claims

MaIA was implemented in 5-6 months, and since the rollout in 2020, AP Pension has managed to streamline more than 2,000 claims cases - this means that MaIA has sorted more than 9,000 documents.

For AP Pension, MaIA has helped them save more than 400,000 pages of printed documentation, i.e. 800 piles of paper - or 40 trees that are still allowed to grow. The solution has also diluted the monotonous, time-consuming and slow work process for claims handlers.

Where the claims handler used to be glued to the printer for more than an hour, and then spend 3 hours mapping and identifying the relevant pages for the case, it now only takes 10 minutes. This gives AP Pension's claims handlers more time in their calendar to do what actually makes a difference - i.e. process the case faster for the customer who has been injured.