How Infosys uses Amazon Nova to extract critical data from technical documents and diagrams

Oil and gas businesses lose valuable time analyzing complex data that’s scattered across both written and visual documents. Logs and reports are typically dense and highly technical, while diagrams and schematics often contain equally vital information within the images. Conventional Retrieval-Augmented Generation (RAG) systems tend to fall short when processing multi-modal documents, particularly when tasked with maintaining the critical relationships between text and images.

Infosys created a powerful AI solution to help make sense of complex technical documents that include both text and images. Amazon Nova Pro made it possible to accurately interpret over 1,000 technical images — such as drilling reports and diagrams. Fine-tuning the system in stages improved the ability to handle different types of information, combining smart search techniques to find both technical terms and deeper meaning in the data. Thanks to Amazon Nova, a time-consuming, manual document review transformed into a fast, intelligent process that keeps critical information intact.