Amrita Chatterjee, COO of Finarb, speaks to us about incomplete digital journeys in financial services being accelerated by Covid-19, future trends and the importance of mapping out an AI Strategy. She highlights the importance of building an approach based on problem solving, incorporating an understanding of the business structure and a realistic pathway to implementation. Looking ahead, she recommends Explainable AI should be hard-wired in from the start.
The financial services industry, among others, has been characterised as being on a digital journey. How far along on that journey would you say the industry has travelled?
In terms of the impact of digital technologies, the financial services industry has seen paradigm shifts in recent years. But it is important to understand that the industry – which comprises the whole gamut of services across Banking, Credit Finance, Insurance, Trading & Investment – is not where the “one size fits all” approach can work.
Each sector within financial services has made significant, but varied, degrees of progress on their digital journey. But their respective maturity and road to adoption is influenced by factors often unique to them. It would, therefore, need different strategic approaches to address existing concerns within each sector, to find the right opportunities for further progress along the digital journey.
What are some of the key forces that you see as propelling change in the industry?
Digital disruption – though not a phrase that is known to be used sparingly these days – has and will continue to drive some of the key forces shaping the industry. Moreover, the Covid-19 crisis has taught us that an online infrastructure is no longer a ‘good to have’. It is a key component for businesses to keep running in some shape and form, to make their services available to customers.
As the digital economy matures and becomes mainstream, while the competition landscape and regulatory environment shift towards greater accessibility and transparency, financial organisations will increasingly look to incorporate Artificial Intelligence as part of their strategic roadmap. The focus will not only be on using AI for improved customer service optimisation and internal operational efficiency, but also to address security concerns and make decision-making accountable.
As AI & ML models become more adept in handling unstructured information and natural language detection, they will commonly appear as the first level touchpoint in customer experience journeys, and knowledge intensive functions previously considered to be human-only processes. Think of digital financial advisors, automated wealth managers, chatbots & conversational analytics – all of these will make bigger inroads into financial services, in turn enabling humans in routine operations and freeing up their time for more nuanced and complex decision making.
The advent of Open Banking also sets the stage for a ‘platform economy’ in financial services, where traditional organisations, emerging FinTechs and large tech players collaborate to offer a bouquet of services both for the end consumer as well as B2B enterprise operations. This will be an ideal win-win for all players involved, to become the end user’s ‘platform of choice’, as well as leverage each other’s resources to gain competitive advantage. Both institutional Banks and FinTechs will try to capitalise on Data-as-a-Service and Business Intelligence – either to monetise the huge data at their disposal, or to mine the data for insights that will bring differentiating value to both B2B & B2C services.
It is hard to ignore the COVID-19 pandemic that is touching every part of the world at present. Do you think this shift to ‘online where possible’ will accelerate transformation across all industries?
Absolutely! The Covid-19 pandemic, difficult as it has been, has also made the need for digital and remotely available services even more apparent. As we have seen, almost all industries have had to re-think and re-structure their existing operations and use their online infrastructure and network, to make emergency services available, and continue running businesses wherever possible.
While this unprecedented crisis, which has sent almost the entire global economy into lockdown, is something we need to collectively come out of, it is imperative we do so by taking both immediate and longer-term measures in identifying the areas that would most benefit from digital adoption, and put a plan and infrastructure in place to make those services available.
In what ways will artificial intelligence and machine learning change the ways that businesses function? What will be some of the effects on consumers?
Device and platform agnostic anytime-anywhere services would be the new normal. With customers adopting digital practices in various areas of their lives, they will increasingly expect their financial service providers to offer the same level of experience.
As numerous surveys and customer lifetimes with a company have shown us, customers are willing to provide their financial service providers with more personal information in exchange of context-aware products and personalised services; and conversely, they can just as easily switch due to a negative experience.
Of course, a lot of this will be driven by early adoption among millennials and digital nomads. This is where digital-first service providers offering peer to peer payments, hyper personalization, mobile wallets, on-demand banking become more prevalent.
But privacy & security considerations will be critical. With Open Banking, financial service providers will largely have access to held and away data to come up with innovative solutions; however regulatory mandates like GDPR and PSD2 mean they have to operate under stricter guidelines and greater levels of transparency. Indeed, addressing security concerns will be the first frontier to overcome, to get customers to share their data for enhanced services.
How do companies set about creating a strategy for AI and ML? What are some of the challenges in putting together an effective strategy?
The first step towards creating a successful AI & ML strategy is to identify the right problems to solve and the right areas to transform. While the possibilities can seem endless, it’s important to start at the right place, which will then be the foundation to build incremental value from further AI interventions.
On the flip side, it is counter-intuitive to have an AI solution that cannot be extended or scaled in future. At the same time, it is important to be mindful of the law of diminishing returns and to therefore not lose the competitive advantage that an organisation can have from an effective AI strategy, that can be implemented sooner rather than later.
Most data-heavy and process-driven business functions are suitable use cases for AI intervention. Companies today have no dearth of data, in fact, most often the challenge is to make sense of the huge amounts of data – Big Data – that they have. Some of the data they have may not be usable, or bias-free; some may be redundant. Also, sometimes companies may not be looking at the right sources of data, or keep in mind future scale, when it comes to data scope and quality.
While most companies believe that AI would be useful, they often struggle to make it work as part of their existing systems and enterprise architecture. Using out-of-the-box AI products and solutions can often become challenging when they can’t be integrated with existing systems without an overhaul and impact on other processes.
Without the right AI strategy, the difficulty in finding a cost/benefit trade-off, and the need to show quick ROIs, often mean that AI programs don’t take flight or reach enterprise level adoption within companies. AI programs best work with a results-driven product approach, that typically have shorter turnaround times than other conventional projects. Therefore, managing that change, and driving adoption, is also an area that effective AI strategies look to address.
Security and compliance considerations should be factored in, right at the start of any AI strategy. Otherwise companies end up trying to retro-fit these measures into a solution, or reinventing the wheel to arrive at a solution that works – both of which can prove to be costly and ineffective.
Tell us about the ways in which Finarb can support companies in the process?
At Finarb, we try to bring a “consult to implement” mindset when partnering with our clients on their end-to-end AI transformation journey – right from AI strategy and consulting, solution design and development, architecture definition, through to production deployment and BAU operations.
As part of our Consulting Services, we conduct Discovery Workshops & Design Thinking Sessions, where we start collaborating with client stakeholders early on, to map out their processes, user journeys and identify the pain points and areas of next best action. The other part of the exercise is to do an AI Maturity Assessment to understand their current application landscape & digital adoption. All of this then results in a Business Case Roadmap for AI intervention, and a recommended POC as a first step.
We strongly focus on the design and development of an MVP for our clients, as this is where we show value right at the start of an engagement. This is also the framework on which the rest of the solution is built, so this phase involves taking a lot of client feedback and tweaking and fine-tuning the AI models. We also use the Proof of Concept to get buy-in from the business users and take their inputs, which forms the base for the business user enablement phase that eventually follows.
In working with clients across industries and business verticals, we usually try to take a use -case driven approach to implement the right AI solution that would not only address the current pain-points and give them maximised benefits, but also bring them sustained value over a period of time. We have ‘productized’ solutions for these use cases, where depending on the client’s needs, the suitable approach could be a lift-and-shift, or degrees of customizations and add-ons to meet their enterprise requirements. We also work with clients on the recommended system architecture and to integrate and deploy the AI models to work with existing operations.
In most of our client engagements we have also noticed, that whether as part of bigger scale transformations or a specific business-case driven solution, clients want to be able to scale & extend their AI solutions to leverage across other channels and business functions. To ensure that we bring that value for our clients, we extensively analyse their current and potential sources of data, their data needs & limitations, and also security & regulatory compliance guidelines in handling data.
Compliance and accountability are also something we are especially interested to help our clients address. At Finarb, our approach is to design AI/ML models that retain their explain-ability and can simulate the impact of changing parameters on their decision making. This is also our focus on one of our next offerings – Explainable AI. We believe that as Artificial Intelligence looks to become a trusted tool for humans to empower their decisions, companies would increasingly have to make their AI/ML models explainable and accountable to show that they are bias-free and secure.
Every year we put together a FinTECHTalents playlist. What would you like to add to this year’s list?
Heal the World (by Michael Jackson)
Meet and hear more from Amrita Chatterjee and the Finarb team at FinTECHTalents 2020 in November.
