Augmenting with AI: Exploring new possibilities in BFSI

Evolving customer preferences and the emergence of digital disruptors are rapidly changing the banking landscape. As banks evolve from a banking as a product model to an ecosystem experience, changes are necessitated to business processes, IT Architecture, as well as operating model.

With increasing focus on efficiency and experience, digital technologies have begun to play an enabling role in the transition. Automation is today playing a pivotal role enabling firms to run lean and cost effective operations, and respond to evolving customer needs at scale.

As automation itself evolves, inter-twining it with AI technology has opened the opportunity for intelligent automation. Favorable government policies, along with AI technology maturity, is further bolstering adoption prospects in the BFS industry.

Why the hype around automation and intelligent automation?

Automation technologies, when deployed after providing for system risks, have the potential to improve  human decision making in terms of speed and accuracy. The potential for value creation is perhaps the largest across industries and use cases. The technology can help lower costs through efficiencies generated by automation at scale, lower errors rates and improved resource utilization. Additionally, it can uncover new and unrealized opportunities based on an enhanced ability to process and generate insights from vast amounts of data. This intelligence in document processing is crucial for banks given the growing volume of unstructured data, which traditional automation tools can not process efficiently.  

Despite its transformational capabilities and massive investments to embed automation in the organization’s systems and processes, few banks have succeeded in leveraging automation technologies across the organization. Banks have struggled to move from the drawing board to scaling monetizable use cases. Reasons include lack of a clear enterprise strategy, an inflexible technology core, fragmented data assets, and outdated operating models that obstruct collaboration between business and technology teams.

 IT Automation – one size does not fit all  

IT Automation contextualized to use cases and business outcomes is likely to lead to positive outcomes and measurable benefits for the organization. Leading financial institutions such as JP Morgan, Morgan Stanley, and others are integrating artificial intelligence, data analytics and Intelligent Document Processing (IDP)- into their systems, aligned to the respective Lines of Business to achieve broader automation  outcomes.

Payments:

 Fraud detection is a key use case. Automation helps reduce manual validation costs and provides improvement in response time for identifying fraudulent transactions. Robust fraud management practices also help avoid regulatory fines and heavy compliance costs 

Mortgages

Customer Onboarding: Application screening and data input can be automated to expedite customer onboarding. Routine KYC task automation leads to reduced per customer KYC costs, improvement in turnaround time and better information reconciliation.  Automated data management is key here.

Credit Underwriting: Automation in customer and regulatory reporting through income, expense and networth assessment  also assists in providing improved decisioning with time bound loan application processing to customers. Data anomalies are avoided, and sanctity of customer data is maintained.  

Capital Markets

Input feed automation (middle to back office): Automation helps consolidate multiple input fields into a single integrated downstream flow for timely execution of the trade transaction. This improves trade execution pass through rate, driving efficiency in transaction execution and reduction in cost per trade.

Similarly, reporting is an integral part of capital market trading activity. Continuous reporting for investors and regulators with capture of process ready images and documents enables reduction in FTE costs and shift in manpower focus to more customer centric activities.  

Processes such as customer onboarding and KYC, mortgages, loan application processing tend to have a large volume of documents, replete with complexity and variety. This makes them ideal contenders for IDP adoption. The current global slowdown, with an already existing remote workforce model propelled by the pandemic, are further necessitating a stronger IDP push.

Augmenting automation with AI

As enterprises progress in their automation journeys,  technologies like RPA are now enhancing  themselves with the potential of Artificial Intelligence, giving rise to what is known as Intelligent Automation. By combining automation solutions with AI technologies, financial services companies can move from automating specific tasks to automating end-to-end processes with embedded intelligence. Intelligent automation (IA) combines artificial intelligence (AI), machine learning (ML), natural language processing (NLP), and process automation to optimize business outcomes. Automating business outcomes with IA rather than automating mundane tasks improves the customer experience, increases operational efficiency, and provides a path to utilizing AI in many automation intensive areas.  

For instance, intelligent automation can assist customer care staff perform their duties better by automating logins or ordering tasks in a manner that ensures customers receive improved and faster service. Other examples where intelligent automation can be applied include account closure, triggering notifications, blocking accounts, and managing account transfers to help improve operational efficiency and overall customer experience.

A growing component of IA is Intelligent Document Processing (IDP). RPA solutions were hitherto not able to automate processes which involved reading, understanding, and extracting data from semi-structured and unstructured documents. Coupled with IDP, RPA can facilitate straight through processing of document and data-intensive processes bringing increased speed and accuracy to banking operations. With data extraction automated, the maker function in the maker-checker construct is seamlessly executed with accurate output achieved in lower processing time.  

Intelligent document processing is showing significant adoption in the banking industry with efficiencies beyond rule-based RPA. The underlying technologies such as AI-ML and NLP allow financial services firms to evaluate processes that require a degree of judgement to execute them successfully. Financial institutions have adopted a number of use cases, ranging from simple integration of cognitive services to AI-powered decision making to deliver efficiency in business outcomes.  

Intelligent automation in action - Industry examples 

Bank of New York Mellon has leveraged almost 220 RPA bots integrated with Artificial Intelligence for process efficiency and cost saving. This has resulted in 100 percent accuracy in account closure across multiple systems, significant improvement in processing time, a 66% improvement in trade entry processes and high reduction in reconciliation of failed trade.

Heritage Bank is one of Australia’s oldest financial institutions. The Bank was facing increasing competition from fintechs and digitally savvy banking counterparts. Heritage implemented an IA solution to automate front end, back-office, and mid-office processes related to operations, fraud risk and contact center services. As a result, the company automated approximately 80 processes, with a level of automation as high as 90% thereby freeing its FTE resources for more customer centric activities.

To automate loan processes, Upstart, a leading AI-based lending solution, focuses on directly offering loans using its machine learning algorithm. The focus is on the segment of population with low credit history. The firm evaluates the years of credit, FICO credit scores, education background, field of study and job history to understand their creditworthiness and grant loans accordingly.

US bank PNC Financial uses the system to automate approvals for certain loan types. The bank combines prescriptive business rules with predictive data modelling to ascertain  customer eligibility for credit.

Persisting challenges

Despite the numerous benefits that Intelligent Automation offers, it comes with its own set of challenges too. Many of these relate to AI security threats, such as tampering with machine learning models or their ingested data to influence outcomes. Additionally, the possibility of malicious or erroneous code being introduced and getting amplified multiple times is a very real threat in an automated process.  

Concerns regarding data privacy and provisioning of data are likely to impact the use of AI and automation in banking. High implementation cost of advanced solutions, coupled with lack of skilled industry professionals, can also prove to be a further deterrent for adoption in the BFS landscape.

At a larger level, governance remains a formidable challenge. As and when intelligent automation begins to include AI-powered decision making, it might lead to new governance challenges, such as the risk of AI bias in lending decisions. Given the risks that arise from fully automated decisions, and regulatory eagerness to ensure transparency in lending decisions and AI algorithms, financial institutions are likely to be cautious in their adoption of Intelligent Automation.

The Road Ahead

Automation has brought significant efficiencies to the banking industry. While the initial mandate was for automation of repetitive low-end tasks, the maturity in the technology has seen banks exploring advanced use cases for reaping greater benefits from automation journeys.   

Cognitive automation, leveraging AI, is taking the industry towards a state wherein end to end decision-making is handled by automation tools, allowing complex tasks also to be automated. This has the potential to augment the human element from business processes, truncating errors and significantly improving productivity.

On the road towards automation, banks will have to standardize and digitize processes to lay the bedrock for successful automation. The product landscape will subsequently have to be geared towards systems which have inbuilt automation solutions.  Care would also have to be taken to keep the ‘human in the loop’, so that automated decisioning and execution is in line with business objectives. Given the close eye of the regulator, banks would also do well to document their automation processes and maintain reasonable controls over AI algorithms, so as to maintain regulatory discipline and compliance.  

There are many implementation examples of how intelligent automation is helping banks and how it can help banks stay competitive both today and in the future. In the end, it boils down to how well intelligent automation is executed and integrated into the end-to-end customer journey.

References :

 https://www.mckinsey.com/industries/financial-services/our-insights/ai-bank-of-the-future-can-banks-meet-the-ai-challenge

https://www.processmaker.com/blog/intelligent-automation-in-banking/

 https://www.prnewswire.com/news-releases/ai-and-automation-in-banking-industry-to-top-us-182-bn-amid-growing-adoption-of-advanced-financial-techniques-301543961.html

 https://techmonitor.ai/technology/ai-and-automation/intelligent-automation-financial-services-leading-the-way

 https://www.aspiresys.com/banking-solutions/cognitive-automation-convergence-of-AI-and-RPA-in-Banking/

 https://kristasoft.com/solutions/industry/intelligent-automation-for-banking-and-financial-services/

 https://research.aimultiple.com/intelligent-automation-in-banking/

 https://www.qentelli.com/thought-leadership/insights/leveraging-ai-banking-guide