Insurance company HanseMerkur presented PLANET AI with the task of revolutionizing their entire input management workflow. Together with IBM as an integrator, PLANET AI faced the challenge of combining the client’s 30-year-old, complex rulebook with state-of-the-art document analysis capabilities. Processing around 25 million single and multi-page documents and sorting them for various departments were the two most important tasks to master at first.
PLANET AI’s Solution:
- PLANET AI provided HanseMerkur with their IDA Textlayer, allowing employees to search, mark and copy text from scanned data, even from handwritten text and distorted scans. PLANET AI’s robust Automatic Text Recognition directly led to an improvement in the quality of subsequent processes.
- Additionally, the implemented IDA Classification feature only needs a small number of example documents to classify entire documents or single pages – beating the former rule-based approach by lengths.
Together with IBM, PLANET AI implemented their capabilities while minimally affecting and instantly improving HanseMerkur’s current workflow. IDA Textlayer
yielded an improvement in the quality of subsequent analyses of more than ten percent compared to the client’s former solution.
Additionally, IDA Classification outdid the insurance’s long-standing rulebook by achieving average ratios of 95 percent in both document and page classification. The classification does not rely on manual input anymore, whereby the potential for mistakes could be reduced dramatically. All in all, HanseMerkur could save thousands of working hours per year while speeding up their input management significantly.