Panos Skliamis, CEO of SPIN Analytics: “It’s important to see the changes we’re living through in the context of the transformations that have come before”

We have just published an opinion piece about why we need to concentrate on potential employees’ growth mindsets rather than past experience – but as this interview with Panos Skliamis, CEO of SPIN Analytics, makes clear, decision-makers need to have some years under their belt. Especially in the world of credit risk management and regulatory compliance for financial institutions, which is where SPIN Analytics’ AI-driven platform is focused.

And Panos himself has much experience to lean on, with more than 20 years’ experience in technology and investment banking projects. Before leading the global expansion of SPIN Analytics – clients include Accenture, Fujitsu, Bundesbank Innovation and many more – he was previously Head of Strategy for an internationally awarded fintech corporation and an investment banking boutique.

Read on and you’ll discover what traditional finance and banking can learn from the fintech disruptors, how to successfully manage technological change and AI’s impact on fintech – the good and the bad.

Recommended reading: Fintech 2024 shows green shoots with backing from Alphabet and Uber

What do you think traditional finance and banking companies can learn from disruptors in the fintech space?

We live in a highly interdependent and turbulent world that is turbocharged by rapid technological change. All financial institutions as well as the regulatory organisations that help keep the financial system stable need to embrace such change and manage it well.

As we all know, many organisations of all sizes have trouble dealing with disruptive transformations of their processes and systems. For those of us with helpful solutions, the modest pace of change and avoidable missteps can be frustrating for all parties, including our target bank clients and our many partners that believe our solution will help them assist the clients they already support.

Change management experts have advised for decades the importance of senior management leadership when taking on such changes, close collaboration among all those involved (including those affected when possible) and aggressive yet realistic time frames. Perhaps most of all is to help the people affected to feel safe – not always easy when implementing radical technology that simplifies the way work is done.

When these requirements are present the chances of success increase dramatically. Finance and banking companies can learn a lot from understanding methods used by disruptors who have been successful in helping other peer group financial organisations as well as the lessons learned from other types of companies who have successfully transformed the way they do business.

Related reading: The biggest challenges for Fintech in Africa in 2024

In what ways is artificial intelligence affecting the fintech sector?

Everyone who has been keeping up with the news knows that AI has good impacts as well as some unattractive ones. Whatever the mix, it’s clear to everyone that AI is creating a bit of a feeding frenzy in which both breakthroughs and breakdowns have become daily events.

Naturally, the application of advanced technologies makes people wary, but backing away from the opportunities means that we don’t learn the valuable lessons that failures can teach. Failures can also guide us to focus our engineering and R&D on understanding the combination of factors that led to the failure.

MIT’s Nancy Leveson uses the example of 19th-century steam engines to make the point that science and engineering coupled with the obvious need for greater caution was a much better path than abandoning the transformative power of steam. I think this is good advice about AI – as it is about change in general.

Fortunately, our corporation has some experts who have done this before. They lived through the emergence of large-scale IT, the births of mini-computers and PCs, and the euphoric rise of the internet as well as its dotcom bust. We also have a board which consists of senior banking and technology executives who lived through the early dreams and disappointment of AI that caused it to be largely written off as an impossible dream.

My view is that it’s important to see the changes we’re living through in the context of the transformations that have come before: in addition to the transition from sail to steam is the staggering growth of 1970s microelectronics – and the repeated booms and busts it engendered in video games. That industry – condemned and written off more than once – is today a $350 billion behemoth that eclipses much conventional media and in which 70% comes from mobile devices that didn’t even exist before the late 1990s.

My belief is that AI will be a part of every electronic product, every business and productive process, and virtually all software. This does not mean that one can pick winners with any greater reliability than we experienced during the dotcom frenzy.

At the same time, we have to move carefully to manage the risk. While “move fast and break things” may be good advice in some domains, the financial sector is not one of them. Our company has taken some steps to dramatically lower the risks of transforming key parts of the credit risk modelling value chain in which we specialize.

First of all, our “flavour” of AI software is completely predictable: it always produces the same outputs from the same inputs and is incapable of “hallucination”. Second, we are working closely with a set of established, trusted, global partners like Microsoft, Accenture and some of the most experienced management consultants. They know our banking customers and have “boots on the ground” in all of them.

These partnerships bring the relationships, knowledge, and trust that allow financial organisations to move rapidly and with high confidence to transform the way their credit risk models are created and updated. This upgraded organisational capability increases the speed and reliability of regulatory reporting, and improves bank capital allocation as well as management’s responsiveness to changing conditions.

Developing regulatory-grade credit risk models in weeks rather than months and modifying them in hours not only creates greater flexibility and innovation, but it allows modellers to do what they love to do instead of the tedium that inherently comes with existing artisanal methods.

What are some of the biggest regulatory challenges affecting the fintech sector?

Regulators are focused on risk, particularly industry-wide contagion, so the enthusiasm for applying new methods – especially Gen AI’s well-publicised risks – is likely to concern them.

It seems likely that regulators will seek to closely monitor the role that AI plays in model development, computation and other aspects of the credit risk reporting pipeline. Certainly, I think we should expect disclosure rules for the use of all forms of AI and that some assurance evidence will be necessary. We have built our product with this expectation in mind.

At the same time, we trust that regulators will be cautious to avoid nullifying the benefits of AI by creating overly restrictive rules. We think that the best approach is to provide a safe alternative to manual methods since innovation in the regulatory process is clearly necessary – frankly, the earlier the better – since simply applying more capital to compensate for out-of-date models is hardly a good answer.

Recommended reading: Top ten breakthrough fintech companies

How does your company differ from its direct competitors in the fintech space?

The short answer is that our existing competitors consist of teams developing models using traditional coding tools. This allows us to say that technology-based competition is not present at the moment. To be clear, there are companies who have solutions that facilitate parts of the credit risk modelling pipeline. However, our approach is effectively end-to-end starting with the raw data all the way through regulatory quality documentation – an element that takes a great deal of the total development time-line.

RISKROBOT’s model development, revision and full regulatory output production happens in hours and can be repeated as many times as the modeller wants. For all intents and purposes, using RISKROBOT modellers can work without worrying about deadlines. Our solution includes incremental data cleansing through final regulatory report generation for over 80-90% of the regulatory credit risk models in use. Beyond this, we can create new models using our proprietary development language and we willingly teach our customers and partners how to do so on their own.

A second distinction – already mentioned – is that we have a wide range of partners who are deeply embedded in our target market and who see our technology as an opportunity to serve our joint customers more successfully using our tools than their labour-intensive methods.

We have two service delivery models that depend on the size of the financial firm. For Tier 1 banks we work with them to create a secure process within the bank’s cloud-based infrastructure. For small banks, we deliver models at scale that fit the needs of their portfolios and budgets with a Modelling-as-a-service approach.

Since it’s not our approach to pick “favourite” providers to support us at ground level we work with a wide range of firms. In fact, not long ago our partners saw us as competition for their traditional model-building work. No longer. Transforming the modelling process and creating new strategies and methods is better for the clients as well as the partners.

In summary, we see our mission as working with partners and clients to transform the credit risk modelling process and all that it touches. Our collective goal is improved use of capital, faster product development, greater responsiveness to change, less tedium for modellers, and fewer loads on regulators through better documentation and fewer problematic submissions.

We also think that our approach demonstrates the complementary way that different AI approaches can work together to develop solutions that leverage both sets of strengths and create fewer surprises.

What are your top three fintech predictions for the upcoming years?

There is a global understanding that if you cannot compete with something new then is better to follow it. This is currently true for the application of generative AI technology and how it can accelerate trivial manual tasks. In banking, however, it depends on the level of regulation and flexibility on capital requirements.

For these reasons, I could imagine that the priorities for change with the application of banking technologies will be focused on customer service, which allows more growth opportunities and automation of the processes that are non-core for every banking institution.

One more topic is assessing the performance of each model based on the internal policies which are driven by banking regulations and management business plans. This topic has plenty of space to grow and it has already started taking advantage of the generative AI speed.

More interviews worth a read

• Mark Allen, Head of Cybersecurity at CloudCoCo: “It’s alarming to witness the extent to which deepfakes can be weaponised”
• Taavi Tamkivi, Founder and CEO of Salv: “Collaboration between financial institutions, or rather lack of it, has traditionally been a challenge”
• Kathie Johnson, CMO at Sitecore: “Generative AI allows marketers to go back to what they love best”