Visma's AI team processes over 18 million documents monthly using specialized models. Their journey began to streamline accounting workflows and has evolved significantly. Initially using traditional deep learning methods, the team switched to transformer technology around 2020, recognizing its potential for language processing. Their flagship product features a document extraction engine capable of identifying key fields in various documents. Additionally, Auto Suggest labels transactions based on learned behaviors, enhancing user experience and organizational efficiency.
Visma's flagship product is a robust document extraction engine that processes documents in the countries where Visma companies are active, supporting a variety of languages.
"We scrapped all of our development plans and have been transformer-only since then," says Claus Dahl, Director ML Assets at Visma, highlighting a strategic pivot to transformers for document processing.
Collection
[
|
...
]