Edinburgh, June 04, 2024 –Lovat, a leading innovator in e-commerce solutions for tax compliance, today announced significant advances in document recognition and data extraction. This cutting-edge technology uses machine learning models to streamline the processing of tax documents and save users time and resources.

“We understand the challenges that e-commerce businesses face when dealing with tax documents,” says Olga M., CEO of Lovat. “Manually entering data from poorly formatted documents is not only time-consuming, but also prone to errors. Our new machine learning models automate data extraction, eliminating these frustrations and ensuring accuracy.”

The challenge of inconsistent tax documents

Tax documents, especially those from international sources, are often inconsistently formatted have different currencies and written in a foreign language. Research shows that over 70% of businesses experience difficulties processing tax documents due to inconsistencies in layout, font styles, and data presentation. This leads to:

• Time wasted: Manual data entry from poorly formatted documents can mean a significant loss of time for businesses.
• Increased error rate: Typos and inconsistencies in manual data entry can lead to costly tax return errors and fines.
• Reduced efficiency: Inefficient data processing hinders an organization’s ability to meet tax filing deadlines and effectively manage compliance obligations.
• Lovat’s solution: Automated pre-classification based on image recognition.

Lovat’s document discernment feauture uses advanced machine learning, to automatically extract essential information from tax documents such as invoices, checks, exemption certificates and others.

Lovat’s document recognition feauture was tested on a huge documents from different countries and achieved an impressive 90 accuracy rate. In addition, a self-learning system continuously refines the model’s performance by incorporating incorrectly recognized data into the training process.

Higher accuracy through state-of-the-art technologies

Lovat’s document processing pipeline goes beyond initial extraction. The system uses a large language model (LLM) called Llama to identify and correct errors in the extracted data. This multi-stage approach ensures the highest level of accuracy for important control information.

Lovat’s commitment to innovation enables e-commerce businesses to manage the complexity of global tax regulations with greater ease and efficiency.

About Lovat

Lovat enables e-commerce businesses to manage the complexity of global tax compliance. Our user-friendly software simplifies the calculation, reporting, and filing of VAT returns in over 111 countries. Lovat also provides expert advice on regulatory changes and additional services such as registration and audit support.