Best Technical Documentation Solution Provider in Sweden
Founded in 1986, Excosoft is a Swedish company renowned for its expertise in technical documentation solutions. With nearly 40 years of experience, they provide comprehensive solutions that streamline the entire technical documentation lifecycle, from planning product manuals to translation, review, and publication. Their product, Skribenta, is an optimal document management system designed to simplify technical documentation processes.
Excosoft faced a significant bottleneck in their data entry process. The reliance on manual data entry for a large volume of PDF technical documents demanded significant staff time and effort. This time-consuming and prone-to-error process may affect the overall efficiency and quality of the documentation process.
They needed a solution that could automatically convert these PDF documents into a special Flex file format, thereby reducing manual effort and enhancing data quality.
Neurond AI quickly assembled a dedicated team of three experts: a Leader and two skilled AI Engineers. To start with, the team analyzed Excosoft’s sample extracted documents to understand their desired output format and specific requirements. We found out that Excosoft’s documents varied significantly in layout and formatting due to the diverse range of vendors they worked with. This complexity, including variations in text fonts, tables with and without borders, and images embedded within tables, required a robust and adaptable computer vision model.
Based on this analysis, we consulted and proposed an AI model solution to help them extract information from technical PDF files and convert it into the Flex file format.
The AI team carefully selected and trained a comprehensive computer vision model capable of handling all these cases, accurately extracting information from the PDF files into the Flex format.
Technologies Used
Computer Vision
Layout Detection Models
Table detection
Our AI model successfully addressed Excosoft’s challenges and delivered significant improvement. It streamlined their data entry process, reducing manual effort by an impressive 80%. Tasks that previously took several days could now be completed in less than an hour. Remarkably, the model achieved an 85-90% accuracy level in extracting information from PDF documents and converting it into the Flex file format.
