Case Study: Thomson Reuters powers AI document understanding with Invisible Technologies

Thomson Reuters builds document AI training data in under 2 weeks with Invisible Technologies

Invisible Technologies partnered with Thomson Reuters, a provider of information services and legal & financial technology, to address a significant challenge. Thomson Reuters needed high-fidelity training data for a document layout segmentation model. Their goal was to create an AI capable of interpreting the visual structure and contextual meaning across diverse document types, which traditional OCR tools could not accomplish.

Invisible Technologies built a flexible data annotation framework with an advanced schema to represent document components with pixel-level accuracy. They established a seamless workflow for expert annotation and review. This solution enabled a 2.5x growth in tasks due to high-quality, on-time results and allowed Thomson Reuters to build its novel data structure and annotation process in under two weeks.

Thomson Reuters