Scanning service provider optimizes document digitization through automatic document separation Emma Lebat Case Study January 28, 2025 Leading Swiss BPO significantly increases efficiency in document capture thanks to AI-based document separation. Overview – Management Summary By implementing automatic document separation with Parashift, a major specialist IT service provider in the Swiss market can significantly optimize document digitization. The service provider is very familiar with the challenges involved in this process step and the associated optimization potential. With the newly designed process based on AI procedures, the goals set were achieved. Automated document separation with Parashift saves costs and processing time. Challenges Every day, the specialist IT service provider processes thousands of documents for its customers. In this case, the documents enter the capture process in paper form. However, incoming digital documents also benefit from this functionality. The documents practically always enter the process in larger quantities (batches/files) and have to be separated. In this case, incoming mail was separated manually by people on the screen after scanning in document capture software; in many other cases, documents are separated during scanning using barcodes affixed during document preparation (AVOR) or inserted barcode or patchcode separator sheets. Both variants require a significant amount of manual work. In addition, incorrect handling can result in corrupt documents, and currently available tools for automatic separation have hardly become established due to their lack of reliability. The solution Automated document separation was implemented on the basis of GNNs (Graph Neural Networks), a powerful artificial intelligence tool for analyzing and predicting relationships between objects. Document awarenessSituational separation based on pages that are much earlier or later in the document. Multi-layered AIUses layout (textual) and images (visual) Contingency decisionProbability “First page” and “Last page” are used for separation suggestion of each page. (*Figure 1) High automation, little validationAutomation optimizes validation effort Continuous, autonomous learningLearns continuously based on human feedback (validations and corrections) Simple configurationSet an acceptable error rate (e.g. 2%) and leave the rest to the machine The results Saved costs, Optimized process,More satisfied customers, Continuous further improvement The implementation of automatic document separation with Parashift has led to significant cost savings for the service provider due to the significantly reduced manual effort. As a result, throughput times and therefore the overall performance of document capture have been improved. The faster processing of incoming mail in turn led to higher customer satisfaction. Specifically, according to previous surveys, more than half (54%) of the documents have already been separated fully automatically. In this case, this corresponds to 82% of all pages recorded. However, the remaining documents can also be separated more efficiently and quickly. The separations suggested by the AI can be easily and quickly confirmed or denied by the human validator. A nice side effect of this is that the system learns during validation and further improves the degree of automation through its use. Interested? Contact us Peter ZurfluehEnterprise Sales Development Representative of ParashiftPhone: +41 79 252 88 50peter.zurflueh@parashift.io linkedin.com/peterzurflueh/ Share on Facebook Share on Twitter