Pieter was attracted to Ocrolus by the company’s human-in-the-loop AI workflow and how we are able to generate proprietary labeled training data at scale. According to Pieter, “This is an end-to-end strategic asset that is rare in the AI world, and it creates a flywheel effect where machine learning algorithms are always improving.” Despite advances in tools for data science and machine learning, the fundamental process of gathering data remains difficult and inefficient. Since every document that Ocrolus processes is validated, Ocrolus naturally creates tremendous volumes of labeled training data. Whereas other companies typically purchase training data or resort to using third party labeling services of varying quality, Ocrolus uses its own human and algorithmic QC techniques to create novel data in a highly controlled environment.

https://www.ocrolus.com/ocrolus-adds-senior-technical-teadership/?utm_content=128249557&hss_channel=lcp-3651177