How SymphonyAI empowers financial crime investigators to become more efficient
As budgets tighten, financial services firms require more efficient ways to handle their financial crime prevention. Jason Shane, Head of Strategy & Innovation at SymphonyAI’s financial services division, explains why SymphonyAI is the perfect solution to this problem.
SymphonyAI was founded in 2017 by Dr. Romesh Wadhwani as part of the SAI Group. The mission was to empower enterprise businesses to realise their digital transformation efforts. Shane said, “There were companies like Google and Amazon who had been successfully using AI for a number of years. But the solutions they had were not well established for enterprise businesses.” It has rapidly grown since 2017 and now boasts a team of 3,000 talented leaders, data scientists, and other professionals. There were two major factors that have helped SymphonyAI become the industry leader it is today.
The first of these was the development of the Eureka Gen AI platform internally. Sensa is the tailored version of this platform that has been specifically tuned and optimised for financial services, which leverages advanced machine learning to help institutions improve their AML procedures by reducing false positives, increasing effectiveness by finding hidden risk, and streamlining investigator workloads. The second was the purchase of NetReveal from BAE Systems in 2022.
NetReveal, which has over 25 years’ experience in fighting financial crime, offers an enterprise-grade suite of tools for tackling financial crime, including AML transaction monitoring, KYC and customer due diligence, transaction screening, sanctions screening, negative news screening, payment fraud, and investigation management. Combining SymphonyAI’s Sensa and NetReveal technologies has created an end-to-end financial crime prevention detection solution, all powered by AI.
SymphonyAI’s ubiquitous technology can be used in any sector, but financial crime prevention is a prime location for its AI to transform operations. Shane noted that many firms are struggling to keep up with the rapidly changing regulatory landscape and heavy workflows. This results in ballooned compliance teams that become very costly on already tight budgets.
“We’re trying to help companies be more efficient and effective,” says Shane. “Finding the risks that perhaps they wouldn’t otherwise find and supporting some of the more mundane investigation type of activities, so that they can focus on value add.”
Today, SymphonyAI offers financial services organisations a full suite of financial crime prevention products, a unique combination of cutting-edge AI and enterprise class solutions coupled with unparalleled knowledge and expertise in financial crime prevention. The company is continuously driving innovation and bringing valuable capabilities to market, which cater to the evolving demands of customers and their workflows and use cases. This includes a significant integration of generative AI technology with, for instance, the launch of the Sensa Investigation Hub which features a generative AI copilot. This is an intelligent AI assistant that supports the role of investigators with a case management tool that collates alerts and data from available sources and uses generative AI to produce summaries and draft reports.
What makes SymphonyAI special is its unique blend of expertise. Shane said, “We’ve got state of the art AI engineering capability from a Silicon Valley based company and experts who have come from big tech. The products have been proven across most of the countries in the world and in most of the regulatory jurisdictions as well, which is incredibly important.”
This allows SymphonyAI to keep its finger on the pulse of the market and explore potential new services where they can add value for customers. Supporting this effort is parent company SAI Group, which has committed to a $1bn capital pool, providing SymphonyAI with the funds it needs to explore new features.
One of the most recent product launches is SensaAI for Sanctions. This is a platform-agnostic AI upgrade for any sanctions solution. The API-based product works with a firm’s existing rules-based systems, leveraging AI to support the match process from rationalising unstructured data, including SWIFT transaction data. Beyond this, the tool applies matching models to produce a context-aware risk score, which can be used to help investigators sort higher risk alerts and likely false positives. Ultimately, SensaAI for Sanctions fits perfectly with SymphonyAI’s mission of helping financial crime teams bolster efficiency and spend more time working on important tasks.
Shane hinted that SymphonyAI has several other product launches planned for later this year, with the ambition being to grow the ultimate capabilities of predictive and generative AI within financial crime prevention.
Challenges adopting AI
AI is on the agenda for most financial services firms, but because of its complexity, figuring out how to implement it can be tough. Firms can either invest internally to increase their capabilities or look to partner with third-party providers. The latter is something Shane sees as a great way to start.
Rather than replacing entire workflows with new AIpowered systems immediately, Shane believes it is better to start by partnering with a third party to help augment what they already have. It is an easy, low risk way to get going with AI. He said, “As an example, you may have an existing rules-based system, which produces a lot of false positives. By implementing a third-party AI product, you can do something called alert triage and actually reduce the volume of false positives and find the true risk. That’s a good way of starting to get into AI.”
Discussions of AI are often met with concerns for the impending doom of human workers. However, Shane doesn’t think there is a future where financial crime operations are completed solely by machines. Instead, he believes a human investigator will always be needed in the loop. This aligns with accountability regulations, like the UK’s Senior Managers and Certification Regime (SM&CR) and provides a mechanism where the user can provide feedback on the AI outcomes, enabling it to get better over time. These are principles SymphonyAI embeds into its own technology. Sensa Investigation Hub’s copilot functionality serves as a base for investigators to easily manage everything and spend more time on important tasks rather than manual data gathering or simplistic monitoring tasks.
Shane also noted the importance of building design and operational controls into AI solutions. For example, it’s essential that AI recommendations are supported by an explainability model that enables the user to understand why a particular course of action is being proposed. Techniques such as ‘retrieval augmented generation’ can be used to minimise the risk of hallucination, where large language models create false information that is presented as fact. In operation, careful consideration should also be given to the selection of training data for machine learning models to minimise the risk of bias.
Although hallucination and bias are intrinsic challenges with AI, SymphonyAI has comprehensive tools in place to combat such occurrences. Alongside these stringent mitigation processes is the role of human workers, who are present to ensure that the output of generative AI is accurate. Having a human in place is still a vital aspect of accountability regulations, such as the UK’s SM&CR, and the symbiotic relationship between humans and AI helps compliance teams more than ever as they combat fraud, with results constantly improving as AI continues to evolve.
Due to the highly regulated environment financial services operates in, there is no room for uncertainty with AI. Shane said, “My counsel is always going to be that you need to make sure that you have the right understanding, the right knowledge of what you’re implementing, and be able to sit in front of a regulator and fully explain the results you’re getting and explain the data it has been trained on.”
However, this doesn’t mean a financial services firm can blindly rely on their own data. Shane pointed to a recent event he attended where a regulator expressed that banks have biased data. “Don’t necessarily think that you’ve done your job by training models on your own data, because that’s not always going to hold true to being non-biased. The fact is, because we have many customers, we have synthetic data that we’ve generated as well, the way that we train our models across the board enables us to actually have a multitude of factors coming into this. So, therefore, we’re able to kind of really produce, if you like, the best type of outcomes on that basis.”
Shane offered firms some advice looking to start seeing the benefits of AI – start small. An incremental approach to the adoption of AI will ensure the technology is implemented correctly and adds tangible value as quickly as possible. As part of this, firms should be monitoring market trends and engaging with teams from across the business to ensure new advancements can be deployed where they would have the biggest impact.
A final point of advice from Shane was to seek a technology partner that has AI at its core to ensure they are getting the most out of AI’s potential. As to why SymphonyAI is that perfect partner, Shane pointed to its technological capabilities, and most importantly its team. “I love the fact that we have such strength of understanding and depth of experience in this firm. It presents companies with that kind of safety, soundness, and comfort of knowing they’re working with a company that has this level of knowledge and understanding.”
