Delivering AI solutions in Banking at Scale
bankingciooutlook

Delivering AI solutions in Banking at Scale

By Antonina Skrypnyk, Client Partner, Head of Financial Services Lab, SoftServe

Antonina Skrypnyk, Client Partner, Head of Financial Services Lab, SoftServe

Banks operate in a highly competitive environment. To keep up with Fintechs that constantly ‘attack’ their market share, more conservative players of Financial services market (FS) have to embed the most recent technology trends and rapid innovation pace into their core business model. On the other side, customers also push banks to provide offerings far beyond their legacy products and services. Customers continue to bring high user experience expectations from their interactions with social media networks and e-commerce platforms into all spheres of life.  And financial services applications are no exception.

Smart solutions that leverage artificial intelligence (AI) might appear to be a good option for banks to form a firm basis of their strong position in this competitive race for new business and highly demanding customers. But just on-boarding cutting-edge technology and new tech trends is not a guarantee of success. The bank has to be able to scale smart solution and new innovative initiative throughout all business streams to gain maximum impact for the whole organization.

"AI is of high interest in banking nowadays. And whomever you would talk to in a bank – each department will say that they should be considered centric for both AI-enabled solutions development and implementation"

Here are a few thoughts for bankers to keep in mind if adopting AI at scale is defined as part of their strategy and embedded into their core business value proposition.

The one and the only. What to choose? AI use cases

AI is of high interest in banking nowadays. And whomever you would talk to in a bank – each department will say that they should be considered centric for both AI-enabled solutions development and implementation. 

But for such a highly regulated industry like FS, it is easier to implement and adopt AI solutions at scale in sub segments where tolerance to mistake is higher and the cost of a mistake is relatively lower. Those to be named here might be tailoring offerings and personification for end-users in retail banking; client satisfaction management; product portfolio segmentation; load optimization or compliance improvement solutions for client support in live helpdesks and integrated chat boxes.

If we talk about the FS industry sub segments that are more sensitive from a compliance and regulatory perspective (like risk management, corporate and investment banking), here smart solutions also have a chance to be leveraged at scale but with one small specific. A reasonable approach for a bank to adopt AI into core business flow here might be if it is leveraged not as an ultimate tool in some critical function but as a recommending one. The applicable use case here might be when AI does not ‘make a final decision’, but instead supports the financial advisor, trader or risk analyst, enabling the expert to make more accurate financial decisions based on the AI-enriched data.

Play hard – Work hard

When the ‘What?’ question is answered and the use case is defined, the next question to address is ‘How?’

The answer is – start small. Proof of concept (PoC) developed according to lean product development framework and directly addressing narrow business challenges is the best way to validate the use case, gain a quick win and create credibility in the eyes of bank executives to support wide adoption and cross-organizational scaling of this solution.

Sandboxes would be an ideal environment to test hypothesis and validate assumptions for use cases chosen by the bank for PoC development. They provide defined development and deployment standards, controlled versioning for all contributors and other benefits. A Sandbox is a relatively safe environment enabling the bank to experiment with the most recent technologies and playwith the most disruptive ideas with no jeopardy to the main business. Besides, banks do not necessarily need to establish and maintain their own internal development eco-system. A good option to start might be to partner with tech and market leaders that provide opportunity to leverage open source environments.

Ready to Go Live

One of the things that put banks aside from others in the market is tolerance to risk in case of delivery failure and performance issues. So the ‘go live’ journey of smart solutions in sandboxes might consist of extra validating and testing steps before the AI solution will be launched to process a full load of live data. This has to be taken into consideration by bank executives, as quick wins that AI-enabled solutions usually bring to the businesses can appear to be not so quick specifically for the bank because of long implementation time, going live and adoption cycle.

Know Your…Data

Banks do have a lot of data. They are even in a better position than FinTechs, as they are obliged to collect and store data as part of the regulatory compliance. But just having a lot of data is not enough. Data quality is the other critical item to keep in mind when considering large scale AI solutions as part of the bank’s data strategy. AI solutions do not ‘self-contain’ the smart knowledge but develop it based on the logic of data from data sets they are trained on. Having biases in that training data directly influence the creation of biased recommendations by smart algorithms. Efficiently designed and established data governance practices help banks to transform lakes of aggregated unstructured transactional and customer data (including legacy ones) into such that are applicable for AI solutions training and running.

Conclusion. And so…‘What to do?’

Data by itself is not smart and can bring no value. The same can be said about hyping tech trends and new technologies that evolve. Smart solutions for a solid and pre-validated business case built on top of the high-quality data is what really might bring value and form good establishment for the new revenue stream in banking.

In order to make these results accelerate cross-organizationally the bank should invest in building a ‘scaling-ready’ and maintainable environment, either by establishing an in-house AI competence or by partnering with open market tech-service providers to leverage their latest expertise and smarttech experts.

Read Also

Removing Data Silos To Reach The Holy Grail Of Financial Services

Removing Data Silos To Reach The Holy Grail Of Financial Services

Alan Donnelly, Head of Financial Services, Salesforce
What To Do If You Are Not Paid

What To Do If You Are Not Paid

Donna Torres, Director of Small & Medium Business Channel & Commercial Operations UK & EMEA, XERO
Request-to-Pay is the Next Generation of Payments and Banking

Request-to-Pay is the Next Generation of Payments and Banking

James Dickerson, Client Partner, Motive Labs
Rethinking Enterprise Networks

Rethinking Enterprise Networks

Andrew Halliwell, Product Director, Virgin Media Business

Weekly Brief

Top 10 Banking Tech Solution Companies in Europe - 2019