There comes a point during my discussion with Netanel Kabala, Chief Data and Analytics Officer at Nuvei, when he neatly summarises the importance of the payments industry in people’s everyday lives.
“What inspires me the most is the fact that payment companies are not usually very well-known, yet they touch the lives of so many people,” he tells us. “We actually affect their lives more than they could possibly understand.”
Kabala himself is well-versed in the richness that the payments sector brings to customers’ lives, having taken the lesser-trodden path to senior management. Indeed, he joined Nuvei not through recruitment – but by acquisition. A graduate from the MBA programme at Tel Aviv University, Kabala spent almost five years at PayPal before co-founding Simplex, which designed a fiat payments infrastructure for the cryptocurrency industry, in 2014.
Simplex was duly bought out by Nuvei in May 2021, with Kabala becoming Chief Data and Analytics Officer – a post he was able to take up when Nuvei’s acquisition completed a few months later. In his new role, he finds himself at the crossroads of many fascinating technologies which are helping to define not just the payments industry, but Nuvei as a company. Not least among these is artificial intelligence, both traditional and generative, which dominates our discussion on this particular occasion.
Difference in AI scope for B2B versus B2C
Kabala is clear that, as much as we like to discuss artificial intelligence (AI) as if it were a single technology, it is instead complex and varied. It delivers significant value for those who embrace it – but what the value entails is very different, depending on whether it is deployed in a business-to-business (B2B) or business-to-consumer (B2C) setting.
“Generative AI for B2B companies mostly revolves around productivity and efficiency, while for B2C companies it could actually offer more personalised and tailored services directly to the customer,” he says.
“Nuvei prides itself in our relationship management capabilities and the fact that merchants could just pick up the phone or send an email and get a human response at any given time. Automating these underlying processes could enhance the productivity of the people answering those calls, but it's different from a scalable B2C business that actually serves millions of users and could offer a seamless, automated experience using generative AI.”
This level of personalisation can prove vital in a consumer-facing context. Despite the progress that Generative AI has made in particular, it still scans like ‘robot speech’ far too often. This has been reaffirmed in recent months by the continuing popularity of mainstream Gen AI offerings like Google Bard and Open AI’s ChatGPT.
Even though they’re surrounded by automation, consumers generally still want to feel the personal touch. Recent research from McKinsey shows that – despite their proclivity for all things digital – over 70% of consumers still expect brands to deliver personalised interactions. Indeed, it’s a revenue driver as well as a vote-winner; three-quarters of those surveyed by McKinsey claim to get frustrated when they encounter interactions that aren’t particularly personal, and companies that excel at personalisation can realise up to 40% more revenue than the industry average.
Fraud prevention continues to dominate AI use-cases
The most pre-eminent use-case for AI continues to be in fraud prevention, thanks to the technology’s ability to analyse large datasets quickly and identify patterns of behaviour. In the payment space in particular, fraud prevention represents the “perfect storm” for AI, Kabala explains.
“Fraud prevention requires looking at many different variables and data types, and giving a real-time response. AI excels at that,” he says. “Some of them are given by the user, some of them are actually behind the scenes and are passively collected. All of this needs to be calculated very quickly with a decision made before money is lost. That's especially true in digital payments and in digital goods. AI is very useful at that, and I think it’s still the dominant way to tackle fraud scenarios right now.”
Kabala believes that compliance has some catching up to do as an AI use-case, particularly when compared to fraud prevention. “For compliance, it's still a work in progress,” he tells us. “Because regulators typically require companies audited to provide a ‘show-of-work’, outlining every step of the way in coming to a particular conclusion, it makes the use-case a bit different to fraud prevention. The timeframes are longer, and the required detail is greater. But AI is getting there.”
It’s important to distinguish between these traditional applications of AI, which learn from patterns of data and act upon them, from Generative AI, which has stolen a lot of the recent attention and dominated public discourse. They are two separate channels, although that doesn’t mean that Gen AI has no role in financial services or payments; in fact, the future for Gen AI looks very promising indeed.
Gen AI is set to become a US$1.3tn industry by 2032 – an increase of more than 3,000% compared with 2022 – according to research from Bloomberg Intelligence. Indeed, Bloomberg believes that “rising demand for generative AI products” could add US$280bn in software revenue alone, driven by technologies like specialised assistants, new infrastructure products and co-pilots that accelerate coding.
“Gen AI is better for anything that involves conversations, large bodies of text, or content creation,” Kabala adds. “This content could take the form of marketing materials, sales pitches and emails, even code. That's what we're already trying to do at Nuvei. Everything related to productivity within the company could be greatly enhanced using Generative AI and all these new capabilities.
“Mundane and often rigorous tasks could be automated using these new tools that didn't exist about a year ago – so things like either customer support, text summarisation, knowledge sharing within the company, writing everything from compliance papers to marketing materials. All of these processes could be greatly enhanced using Gen AI.”
Building an AI economy fit for everyone
In a turbulent economy, where many small businesses are struggling to pay their commercial rates, talking about AI might seem like a flight of fancy. Can independent businesses and everyday firms really afford to invest in this technology, or will it continue to be the preserve of large multinationals and enterprises who have the resources to invest?
Well, no matter how new Generative AI might seem, we’ve been through this process before. There have been predecessor technologies, like coding and web development for instance, that required companies to recruit skilled talent to succeed. However, what we saw with coding was an emerging wave of technology innovators who found ways to democratise access by rolling out drag-and-drop low-code editors. Not only did this lower the barrier to entry for firms with fewer resources, it made the entire end-to-end process much simpler and quicker for large companies already invested in the coding economy.
We may, in time, see something similar from the AI industry. Currently, those with the deepest pockets are most actively engaged with the technology. But, as AI becomes more mainstream, those early adopters will start creating ways to bring smaller firms into the fold – perhaps with easy-to-use applications or interfaces that leverage their own AI expertise.
The advantages of this would be two-fold: first, it will make it easier for everybody to benefit from the power of AI; but secondly, it would ensure that anyone building on artificial intelligence is doing so from a sound base, with all the requisite safeguards in place.
“Everything related to data privacy, compliance, explainability and so on is super important in the financial sector. Here at Nuvei, and in the industry as a whole, we are investing in building those safeguards and making sure that everything is done in a reliable manner,” Kabala says.
He continues: “AI has existed for some time in different forms, so it's not a brand-new concept. What's different in Generative AI is the fact that only a few companies dominate in the space – typically those that provide large language models. Building infrastructure involving these language models will require businesses to choose whether they're doing it in-house; using an open-source model; or using a vendor, and if so what are the capabilities of this vendor and their safeguards?
“In terms of data privacy, we need to get prepared for new regulations due to the rise of Gen AI. The first one, which is obvious, is explainability. We need to be able to explain why the model made such a decision, why the model offered one payment method over another, or in other cases declined payment. This will be a major shift, I think, for any company using AI. Aside from that, I believe that we'll see more regulations around Generative AI, how to tell the difference between automatically generated content and human content – not necessarily for payment companies, but in general. And I think that all the conversation around fake news will be heightened in the coming years.”
Do people need to understand AI to embrace it?
If operators can successfully navigate this delicate regulatory landscape, there is potentially an AI utopia that awaits them: a perfectly blended outcome where risks are minimised, opportunities are fully reaped, and we get a prosperous AI economy as a result. What would this look like? Kabala has a few thoughts.
“First of all, payment companies should embrace AI and understand the value it can provide to merchants, and eventually their customers,” he says. “Even if the value is not immediately apparent and not immediately revenue-generating, improving the internal processes and improving the efficiency will lead to greater customer satisfaction and quicker, more accurate decision-making.
“That's where I anticipate the payments sector, and more generally our entire work culture, will get to. I hope that regulators will help by providing a framework that balances both the innovation that comes from AI with protecting customers and their data, all while ensuring that the AI reaches reasonable decisions on their payments.”
There continues to be a debate about whether or not customers need to understand AI in order to embrace it. It would certainly help with adoption, but it probably isn’t vital to the continued growth of the sector. Nonetheless, the level of general understanding about AI might surprise us; according to a survey conducted by Ipsos, almost two-thirds of consumers (64%) report having a “good understanding” of AI. Reassuringly, most people now feel comfortable using products or services that rely on AI, with only 39% of respondents saying they feel “nervous” around AI.
But you can compare it with your car. Even as we press on with electrification, cars are still as popular today as they have ever been. Most drivers don’t understand the inner workings of a combustion engine – or even the electric motor that drives new electric cars. Yet they understand how to operate it, and that it gets them to where they need to be. They know where to refill it, or where to recharge it, and what to do when something goes wrong.
“Customers care about the value that we provide them, they care about whether they’ve got a solution to their problem,” Kabala says, raising a different analogy of his own. “If you're standing in line and the cashier doesn't work because there was an issue with the software, and the merchant couldn't get their payment company on the line, they don't really care whether it happened with or without AI. They just want the problem fixed.”
Having a detailed understanding of the underlying technology is not so critical when it comes to mainstream adoption, and that is a good omen for the future of AI in payments.
“Here at Nuvei, we are looking at different opportunities to enhance everything that we do within the company, using all available tools, from machine learning to Generative AI. That could involve either preventing fraud, improving approval rates, enhancing internal processes and customer support – anything of that nature, while eventually bringing value to the customer.”