The rise of challenger banks has been a particular hallmark of the fintech industry over the last decade. Created to disrupt the traditional banking sector, challengers are full to the brim with innovative, often digital offerings aiming to serve customers in a variety of ways. With the customer taking centre stage and newfound co-operation with incumbents, this month we explore some of the classic attributes of challenger banks and their efforts to stay one step ahead of the industry.
Both challenger and incumbent banks are increasingly using artificial intelligence (AI) to transform their customer experiences. Not only can it help detect and stop fraud, but AI can also remove friction from customer journeys and ensure client satisfaction. To learn more about how AI can be used in banking, we spoke to some of the key industry players to hear their thoughts.
Peter Sanchez, global head of banking & treasury services at Northern Trust believes the benefits of AI are seen in personalisation.
He said: “For an established global bank with sophisticated institutional and wealth clients, artificial intelligence and machine learning can enable personalised service, security and regulatory compliance at scale and across borders, with efficiency that benefits our clients and the global enterprise.
“Starting with the individual client, voice, facial, fingerprint recognition can provide ease of access as well as security. AI data gathering and analysis of client activities can also be used to provide more accurate forecasts of account balances and likely transactions, and suggest targeted product offerings that fit a client’s needs. At the business, fund or enterprise-level, fraud profiling uses machine learning to spot unusual activity. AI can increasingly help in the compliance world, quickly scanning regulatory changes and identifying impacts to terms and conditions within websites and mobile apps, as well as internal systems.”
Aki Eldar, CEO of Mirato, a third-party risk management (TPRM) platform said: “Banks are increasingly dependent on third parties to deliver critical business processes and services, potentially exposing themselves to significant risk.”
He continued: “Yet since new regulations were implemented in 2013, TPRM programs still have not changed much, even though the global risk landscape has become increasingly complex and fraught. Banks’ current TPRM programs are not up to the challenges of the evolving risk environment, where extreme events such as climate catastrophes, pandemics, and massive cybersecurity breaches are occurring more frequently. Many TPRM approaches still require manual tasks to be performed by subject matter experts, who spend much of their time on tedious, labour-intensive processes, leaving little room for the more challenging work of planning for and navigating through actual risk events.
“Advanced cognitive technologies such as AI and machine learning are helping banks strengthen their TPRM programs by automating the manual effort, empowering banks to better identify and anticipate risk and more quickly conform to rapidly evolving regulatory requirements. For example, AI and proprietary natural language processing (NLP) algorithms can analyse third-party content and validate and cross-reference it with other data sources. These tools complement workflow automation, saving banks significant time, effort, and cost associated with manual TPRM work. AI enables data mining from questionnaires, evidence documents, data feeds, etc., and transforms it into actionable risk exposure insights with specific action plans. An AI-powered TPRM intelligence platform continuously monitors and digitises data collection from numerous sources around-the-clock –allowing banks to leverage previously unused or underutilised data sources due to a lack of manual bandwidth.
“In the future, banks will continue to use AI to digitally transform their operations and take advantage of the power of advanced data analytics to help drive meaningful results, from enhanced employee and customer experience to streamlined processes to reduced third-party risk.”
Helping the underbanked
Nicole Valentine, fintech director of the Center for Financial Markets at the Milken Institute thinks AI “has yet to realise its full and true power.
She continued: “When optimised, it can be instrumental to improving the high rate of households that remain unbanked and underbanked and create adequate access to capital for the many small and medium-sized enterprises that need it most.
“AI has shifted the culture of banking to automation and machine learning from legacy banking practices, which is pivotal in strengthening lending and credit decisions that enhance the banking experience for consumers. We are focused on not just what AI can do but how many people and businesses it can impact. To truly measure success and ensure that broad access to financial services is achieved, we need a new benchmark that leverages inclusivity as a key component of any performance metric.
“Traditional financial institutions, industry disruptors, and financial regulators have an incredible opportunity to work together to harness the power of fintech to increase economic and social mobility in underserved communities that have historically been excluded.”
Data, data, data
Kavita Singh, VP of AI Product Management at Payrailz said that AI is helpful when examining data.
“Financial institutions have large volumes of data they can use to improve customer experience,” she said. “In payments, something incredibly useful that AI can do is help manage an accountholder’s finances. Machine learning can analyse their payment habits when it comes to bills, spending and saving to make proactive recommendations that help account holders better manage their day-to-day finances and reduce financial stress. For example, AI can see a pattern that a particular banking user pays their large bills on the 15th of the month and have historically needed an aggregate amount. Based on that pattern, AI can predict how much money they will need to cover those bills this month and can alert the user about having sufficient funds to cover those bills. The possibilities like this go on and on.”
Louis Brown, head of data science and advanced analytics at Chetwood Financial said: “AI and Machine Learning are used differently in various parts of banking.
“As a digital bank, we’re more interested in what’s going on with the use of AI and ML on the credit side of things.
“In credit, there are a number of exciting areas which are increasingly using AI and Machine Learning. For credit scoring it can be used for boosted decision trees, this technique is being adopted more and more across the market, replacing traditional ML techniques like Logistic Regression. In turn, tools such as SHAP (Shapley Additive exPlanations) can then be used by practitioners to explain the decisions made by ML. Internal Ratings-based approach to capital requirements for credit risk also has the potential to be transformed. In the credit market, protecting against fraud remains one of the largest adopters of ML, however, we expect ML to continue to transform the whole credit side of banking.
“In the future, Probability of Default Model practitioners may see powerful uplifts to their incumbent models when using Boosted Decision Trees, and possibly Neural Networks as well. The biggest change, however, is likely to be in IRB Model Operations; this is where the adoption of ML techniques may lead to model change being delivered more quickly.”
Finally, Joost van Houten, CEO of Sentinels thinks that AI can help with anti-money laundering strategy.
He said: “Banks are spending an increasing and unsustainable amount of resources on fraud, money laundering and terrorism financing detection and prevention because the tools they use are no longer up to the job. The European financial sector spends about €100billion trying to identify dirty money in the global financial system, but less than one per cent of that money is seized. Clearly, something is not working.
“Banks are now recognising that incorporating AI and, as a subset, machine learning into their anti-money laundering (AML) and transaction monitoring is a more effective strategy. This technology can be purpose-designed to meet the challenge head-on. It can detect suspicious behaviour and activity more acutely, then learn and adapt to changing criminal activities.
“Challenger banks are now incorporating automated systems that are trained to monitor transactions for unusual events. When the system is fed with substantial high-quality datasets (everything from relationships between customers and the exact time of day a transaction took place can be analysed), it can identify behaviours, patterns, and connections that even the most meticulous compliance officer could not.
“Sentinels is unique in that we focus on the institution’s clients and their counterparties (as the actors) to uncover criminal patterns and typologies in transactions. Our machine learning approach spares compliance departments time and effort as it is utilised to omit the transactions that do not present a threat to the businesses’ risk-based approach to the customer’s profile and behaviour. We concentrate on anomalies, the potentially risky transactions, all to replace the masses of (predominantly false positive) rules-based alerts with precise, risk-prioritised alerts.
“While AI and machine learning applications in compliance remain in their infancy, it is an area that shows great promise and must become ubiquitous if the banking industry is to keep up with the evolving fraud threat.”