Feedzai launches Railgun after $100 million invested in R&D against financial crime
“Railgun is a significant weapon in the fight against financial crime, allowing banks and other financial institutions to accurately and cost-effectively turn the tide on the rising levels of fraud. With Railgun, banks no longer have to make assumptions or generalizations when it comes to risk profiling – helping protect millions of people worldwide.”
Feedzai has announced the launch of Railgun, an AI engine designed to target and intercept financial fraud before it can occur.
The launch of Railgun, which was designed to address the compromises inherent in today’s fraud detection systems, comes on the heels of a $100+ million investment by Feedzai to build its RiskOps platform, including in basic research, having created nearly 100 patents and pending patents in the US and Europe over the last three years.
Feedzai’s R&D department innovates in a wide variety of areas including applied AI and machine learning, fraud detection, streaming data processing and analytics, money laundering detection, rules management, and AI explainability and fairness. Feedzai’s RiskOps platform is intended to be the world’s most comprehensive suite of solutions to combat financial crime.
“Turn the tide on the rising levels of fraud”
Pedro Barata, Chief Product Officer at Feedzai said, “As technology continues to evolve and tools such as AI become more readily accessible, fraud detection systems need to be able to keep pace and reliably combat criminal activity. Railgun is a significant weapon in the fight against financial crime, allowing banks and other financial institutions to accurately and cost-effectively turn the tide on the rising levels of fraud. With Railgun, banks no longer have to make assumptions or generalizations when it comes to risk profiling – helping protect millions of people worldwide.”
According to Feedzai, Railgun’s patented technology features:
• Enhanced accuracy – Railgun enables real-time calculations based on data across much longer time windows, providing financial institutions with better observability and understanding of customer behaviors, resulting in more precise detection of suspicious activities.
• Greater agility – Early production results show that Railgun improves the speed of risk strategy updates by 4x or more, enabling swifter responses to emerging fraud threats. New rules become effective almost immediately and risk recalculation no longer imposes a heavy burden on data science teams.
• Scalability and lower latency – With Railgun financial institutions can confidently handle increasing transaction volumes without compromising accuracy or decision latency
Feedzai’s Railgun addresses generative AI-powered financial crime
Feedzai argues that today’s risk engines force financial institutions to limit the data they use to make risk decisions – typically by looking only at a limited history of data, and by using only a subset of relevant data inputs.
Unlike other risk engines, Feedzai’s Railgun removes these constraints, so financial institutions can use a complete history of all relevant data – enabling them to assess the likelihood of fraudulent activity with unprecedented accuracy.
The RegTech firm also stated that financial institutions must use new AI-based fraud-fighting technologies to address the ever sophisticated financial crime that embraces generative AI and other techniques.
Railgun’s technology is the result of years of intensive research and development, and the product of multiple Feedzai patents in the area of streaming analytics.
Additionally, Feedzai said fraud detection engines typically rely on risk profiles that capture customer information, such as transaction history over specific periods, and assess the likelihood of a fraudulent payment based on a real-time comparison between current transactional activity and past typical behaviors.
“In the past, fraud detection suffered from a paucity of data to build these risk profiles, but now our digital age ensures almost limitless information. However, risk engines have struggled to keep up with this change, because storing profiles for numerous customers, cards, and terminals requires massive amounts of low-latency memory space at scale. As a result, current systems simplify data inputs and limit historical context, diminishing fraud detection accuracy in order to achieve scalability. And because updating risk profiles is an offline process that is time consuming and expensive, these systems are also less agile, taking more time to adapt to rapidly evolving fraud patterns. Altogether, the result is a failure to prevent significant financial crime.”
