The Merlon Intelligence platform flags and ranks risky transactions for banks. This requires both a top level prediction model and a robust data extraction and enhancement pipeline. Merlon crawls PDFs and other unstructured text to create a graph of financial and contextual knowledge. A natural language processing system enriches this information and allows separate predictive models to make connections between events and people. That predictive model outputs a rank of suspicious activity that can then be analyzed by a human. Traditional banking analysts can provide feedback and improve the functioning of the entire system. With a pedigree consisting of both finance and machine learning experience, Cross hopes to best legacy players like Oracle and Lexus Nexus that rely on outdated technology while getting the jump on newer entrants like Palantir that are bogged down by clunky consulting operations.

https://techcrunch.com/2017/05/23/merlon-intelligence-raises-7-65-million-in-seed-financing-to-combat-money-laundering/