ANNA Money now catches twice as much fraud with half the alerts.

Balancing fraud prevention with automation and customer experience

ANNA's philosophy involves using AI and automation to minimize the friction for legitimate businesses while maintaining strong fraud and financial crime controls. 

However, this created a tension between maintaining a seamless user experience and effectively preventing fraud.

At the time of implementation, ANNA wanted to reduce the time it spent on false positives which slowed down processes for users and negatively impacted customer experience. The high volume of alerts also took time away from investigators to focus on more meaningful and complex cases.

“This is just one of the steps ANNA is taking to give our customers the strongest possible security.”

The ANNA Money Team

APP fraud reimbursement requirements have made turnaround times paramount to institution success in 2026. Anna wanted to ensure it handled this efficiently, emphasizing the need for faster and more accurate decision making.

ANNA identified that simply hiring more analysts to work a higher-volume of alerts had diminishing returns. Instead, the goal was to improve efficiency, increase automation, and deliver higher-quality alerts so they could help customers and resolve issues as quickly as possible. 

As a company actively investing in generative AI and automation across its product offering, ANNA sought specialised, high-performance models that could integrate into and strengthen its existing infrastructure.

“The Resistant AI team has been using AI to detect criminals for 16 years, and we decided to work with them because we know they’re the best in their field.”

The ANNA Money Team

Advanced machine learning models to enhance detection and efficiency

ANNA partnered with Resistant AI and Experian to introduce a joint pilot product that added an additional layer of fraud detection powered by advanced machine learning.

For over 20 years, Resistant AI’s founders have been building AI models to detect fraud and financial crime, designed to identify sophisticated fraud patterns and uncover complex criminal behaviour (including Authorized Push Payment fraud).

Experian brought access to rich UK data assets, including fraud and bureau data. When combined with Resistant AI’s models, this was a game-changer, helping ANNA improve in detection accuracy and efficiency.

Rather than deploying AI for its own sake, ANNA took a highly pragmatic approach. It only adopted models that demonstrably improved efficiency and outcomes, continuously evaluating against existing rules and systems to ensure they delivered measurable value.

The solution complemented ANNA’s broader fraud prevention ecosystem, working alongside its internal tools and rules as an additional layer without needing to replace the existing tech stack. This approach enabled more effective detection while supporting ANNA’s automation-first philosophy.

“The key differentiator for me is that it’s a mix of vendor model and data science as a service. [Resistant AI has] a much more collaborative style, which is particularly useful for small and medium-sized companies in this space.”

James Coveney Head of Data and Fraud Detection at ANNA Money

More precise detection, fewer alerts, and improved efficiency

The initial joint pilot between ANNA, Resistant AI, and Experian exceeded expectations.

Detection performance improved significantly, doubling true detection rates. At the same time, false positives were reduced by 80%, dramatically lowering the burden on analysts.

Overall alert volumes were cut in half, enabling ANNA to move closer to its goal of minimizing false positives and focusing on high value work for analysts.

These improvements helped streamline investigations, allowing analysts to focus on genuinely suspicious activity rather than reviewing large volumes of unnecessary alerts to their customers' chagrin.