What’s the reality of AI implementation in banking? Does the right IT strategy exist? We interviewed 15 banking leaders to hash out what actually works.
An IT strategy is now a vital cornerstone of the overall business with advancements in technology and artificial intelligence taking flight. But strategising one can be complicated for the Banking, Financial Services and Insurance (BFSI) industries, considering the complex nature of their operations.
How are banks approaching tech and AI implementation, why and what works?
We asked 15 banking leaders
We sat leaders from multinational banks down on several roundtable discussions on the topic: The Business of Working Smarter: Cognitive Abilities to Drive the Next Wave of Growth in the Financial Sector.
Representatives included SVPs, heads and directors. They span the bank’s IT Operations, Big Data, AI and Innovation departments, as well as lines of businesses like Regulatory Operations, Custody, Trade Finance, OTC Derivatives, Audit and more*. Together, we discussed the approaches leaders are taking today in developing an IT strategy to implement smart technologies.
Developing and implementing an IT strategy is important for any organisation pioneering digital transformation initiatives to maximise workplace potential. Especially with market competition, regulation and lofty spending forecasts on the line, a clear strategy is essential to ensure that the entire bank is aligned on what goals to shoot for, and how to do it.
We found 4 IT strategy approaches to implement AI in banking
Implementing AI is a highly calibrated process for banks. Only 50% of these firms have a clear idea as to what entails the process.
Through multiple interviews, we looked beyond theory to explore the reality of AI implementation in banks today. From there, we deduced four key approaches that shadow their IT strategy, summarised in the figure below.
While they range across the spectrums of business value and technical feasibility, each approach has its merits, and must similarly be used with caution. How these characteristics surface primarily hinge on the specific frame each bank is looking at. In other words, the right IT strategy is the one that works for them.
In the sections below, hear from banking leaders themselves about the nuances surrounding each approach.
4 IT strategy approaches to implement AI in banks
1. Most banks are laying the groundwork for a holistic ‘all-in’ approach
Large global banks are on their way towards massive IT overhauls through an ‘all-in’ approach. They’re looking at the big picture and laying out a long runway to build the right technology foundation from the ground up.
If resources allow, leave no one behind
A holistic approach behind an IT strategy prioritises making a substantial impact across all operations and all stakeholders, through the whole of organisation. Various divisions of the company should be involved to roll out each part of the IT strategy to take into account the synergistic natures of customer journeys and workplace operations.
The motto is to leave no one behind. This has huge potential for significant digital enablement, although heavy on resources.
The rationale behind the approach is sound. Banks understand the transformative advantages of AI and other smart technologies ,and the necessity to outpace competitors from all fronts. Reaping the most benefit from digitalisation, automation and AI requires structured yet metamorphic moves to be made.
We found banks to go all-in in with in-house capabilities or external POCs
The first is in-house. Internally, we’re seeing firms ramping up in-house resources and talents to build custom tech solutions on demand. Banks that have sufficient resources are investing in-house research teams to build technological capabilities and evaluate solutions. Solutions are tailored to fit the company’s IT infrastructure and targeted operational requirements, as well as navigate specific regulatory controls.
When brought up to speed, in-house teams can deliver solutions faster, which will prove invaluable to the productivity and success of the company. But of course, these come at the cost of heavy investments and maintenance effort.
Externally, firms may need to seek the expertise of FinTechs by conducting proofs of concept to experiment with outsourced solutions. These specialists have more experience and tools to remotely manage various different parts of the ship. The approach tends to be less expensive than internal resources, aside from factors like time to evaluate vendors and collaborate on projects.
2. Financial regulation forces many to prioritise a risk-centred approach
However, regulation is a recurring obstacle quoted by banking leaders. Finance is notorious for being one of the most regulated industries in the world. Legacy to current regulatory restrictions tightly govern the lending landscape. To many banks, these restrictions are regulatory burdens hindering AI adoption.
Technologies come with risks on multiple fronts
Entering the world of digital through automation, AI, cloud computing and data processing means more cybersecurity threats and data privacy risks. Various banking and finance use cases are high-risk as well, such as creditworthiness. Regulatory controls must be put in place at every step of the data journey within each AI implementation, from collection and use to protection, retention and disposal.
Fairness and accountability are areas of concern as well, pushing banks to make room for practical interventions to protect trust and ethics. Many are taking time to educate stakeholders and thoroughly test technology for fairness before they go live. A risk-centred approach also requires more documentation every step of the way, be it internal or external processes.
Pressures to stay compliant are forcing many financial institutions to adopt a risk-centered approach in their IT strategy. Public guidelines that govern the use of technologies such as the Monetary Authority of Singapore’s Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of Artificial Intelligence and Data Analytics in Singapore’s Financial Sector are a valuable starting point.
Extra controls can hinder growth, especially for particularly material use cases
Of course, adhering to strict regulatory frameworks translate to extra steps before technologies like AI hit the ground running. Additional oversight is also necessary to regularly review or validate data and models for biases, and internally authorise technology implementation. It might be the case that projects get abandoned midway for failing to pass a periodic review.
Too strict, and regulatory controls can become red tape. Red tape that dampens the benefits of AI and advanced technologies, by reducing the speed for deployment, and potentially increasing investments in talents or outsourcing costs.
It’s undeniable that measured steps to regulate technology implementation are crucial. But how much oversight is needed? One consistent theme echoed through the digital roundtables is to calibrate the level of control based on the materiality of each use case. Straightforward and simple applications can suffice with less frequent checks. But the greater the reward, the greater the risk. Riskier applications which are particularly material need more testing and validation.
3. Some rely on piecemeal solutions in place that sidestep strict regulations and lofty costs
With the above considerations in mind, what we noticed was more banks looking for piecemeal AI solutions that are easier to implement from the regulatory standpoint. Such solutions are less material and can bypass heavy resources required in the development or regulation aspects.
Piecemeal applications are low hanging fruit
Piecemeal solutions could surface as narrow applications that don’t cover the end-to-end workflow. For example, in a complex client onboarding process, banks might only introduce automation to a small step in the entire workflow. They may apply Robotic Process Automation (RPA) rules to interface with clients, but still manually process submitted documents and validate information.
We found this approach to be more accepted in specific operations that tend to be more antiquated, such as Trade Finance which is notoriously manual and paper-based. For operations like this, piecemeal solutions can be the stepping stones to gradually build better AI capabilities. Simple solutions can be a testing bed at lower risks and costs.
But isolated applications must be used with caution
A piecemeal IT strategy poses some danger as well. Solutions which are temporary and limited in their applications across time will only see short-term returns. Unless solutions are built to be modular, future additions to a growing tech stack could see complications in integrations or duplicated functions.
There’s a lot of debate on this topic with many camps against piecemeal technology implementation that doesn’t create a data culture from the ground up. However, piecemeal doesn’t mean banks are always unsystematic in their IT strategy. A small percentage of them are reaping benefits from the approach’s flexibility and speed. Using practical lenses, piecemeal solutions can be a valuable interim measure to sidestep real challenges in regulatory controls and cost.
4. An innovative platform and people-driven approach is the most ideal, but not mature
We’re now driven to examine the final piece of the puzzle; the piece where we see simple investments in AI fetching substantial returns. Is there a sweet spot between effort and outcomes in developing an IT strategy? According to who we spoke to, not yet.
From what we know in the tech space, the approach that ticks the boxes for maximal value and low effort lies in application development. More specifically, low-code or no-code hyper-automation platforms for end-users to develop and use applications.
An easier an effective way to scale is via low to no code platforms
Low to no code platforms are able to deliver back-office applications to customer-facing portals without the typical resources and IT talent associated with traditional software development. People–meaning anyone and everyone–are the driving forces behind each application.
For example, Omnitive IDP’s no-code platform allows users to build intelligent document processing models all on one web interface. The process is 99% faster than the time taken to custom-build an outsourced Machine Learning solution with a team of IT talens and data scientists.
However, only about 15% of the banking leaders we spoke to are veering into application development on low to no code platforms. Others primarily quote regulatory overheads and training as the main hurdles.
Platform providers have the bigger role to play here
Generally, what we found was that reception to end-user platforms is intertwined with their understanding of what the tools are capable of. Many have just an abstract understanding of such platforms. Additionally, each platform has a different range of functionalities, which builds on the air of mystery surrounding the new trend.
With the move towards no-code still in its infancy, it seems that the ball is in the hands of platform developers and tech vendors, to create tools in line with the challenges and requirements of banks.
TLDR: It all lies in the frame
The reality of implementing AI in banking is that there’s no hard and fast answer to what’s the right approach to developing and executing an IT strategy. How you frame the situation matters most. Think about the extended trajectory ahead, and consider what are your challenges and needs now and in the future? What’s your spending forecast over the next few years? What are the short to long term goals to hit?
There’s a time and place for each approach we discussed. If just starting out, systematically adopting piecemeal solutions can be a valuable prelude to future strategies. Once the bases are covered, leaders can begin planning how to fit AI more thoroughly through the whole organisation.
If resources allow, go all-in to reap the benefits of deploying AI at scale. If the materiality of use cases is high, take a more conservative approach to navigate the regulatory barriers and mitigate risks. It’s inevitable to see a mix of use cases when ramping up a culture of digital within the organisation. But if each room for application is thought out synergistically with others, companies will see less duplication in technology and effort, and more benefit from structure and economies of scale.
Finally, while the tech landscape continues to evolve, stay open to what innovative technologies it might bring. Keep an eye on the move towards no-code, which will empower people to access AI through platforms.
*Designations are reworded to protect anonymity
What does a no code platform for banking look like?