CONTENTS

1. Introduction .................................................................................................................................... 3
2. Overview of RPAand AI technology applications across the banking value chain 3
I. Robotic Process Automation (RPA)............................................................................ 4
II. Artificial Intelligence (AI)............................................................................................... 6
3. Key Strategy Considerations for banks while deploying RPAand AItechnology 9
4. Outlook for AI and RPAin banking ........................................................................................... 10

2
1 lntroduction
Evolve orperish, sosaystheold adage,andthiscouldn't Almost four in five
be truer for the global banking industry currently. Recent persons - 79% - who
developments in the market such as the proliferation of smart
use mobile banking in
phones, the explosion of social media and growing
Europe have boughtan
dominance of online retailers and shared economy platforms
such as Uber have brought about aparadigm shift in the
item using their mobiledevice
concept of customer experience and the quality and speed of in the last 12 months
service delivery. Customers are increasingly getting accustomed
- EuropeanBanking
to tailored customer service and the convenience and speed of Federation report
making online purchases and transacting through mobile
devices; which have inadvertently raised their expectations
from banking and other financial services providers to offer the
same user experience.
For Banks, a big hurdle when competing with fintech start-ups is
Unfortunately, for banks, this gap in customer expectation is
its inability to achieve operational efficiencies in its back-end
being serviced by the flourishing breed of fintech start- ups who
processes which in turn inhibit the level of front-end system
have the agility and flexibility to use emerging technologies to
enhancements that can be made to improve customer
introduce innovative business solutions, offer better customer
experience. Two technologies – Robotic Process Automation
touch points, and most importantly have the ability to pick and
(RPA) and Artificial Intelligence (AI) - play a critical role in
choose the parts of banking that they want to focus on. This is
empowering banks to rapidly automate its business operations
driving disruption across the banking industry, and incumbent
banks, especially in mature markets such as Europe and the US andeffectivelytransformtheentire business.

have already embarked on the digitalization journey to Through this whitepaper, lBS lntelligence and Hexaware
transform their services by investing in the latest technologies endeavor to explore how RPA and AI powered cognitive
and revisitingtheir businessstrategy - both internalandexternal technologies are transforming the banking business industry,
and make it moreconsumer-focusedthanproduct-focused. specificallyin maturemarkets such asEurope andtheUS.

2 Overview of RPA and Al technology applications

across the banking valuechain
The last 5 years have seen RPA and Al technologies evolve to a solution and compliments workflow solutions perfectly in end-
point where they are now beyond the hype cycle and are to-end automation of standardized process in banking
making a visible impact in the banking and financial services operations. On the other hand, the more complex artificial
industry. These two technologies have applications in every intelligence (Al) technology, which uses machine learning, image
area of the banking value chain right from customer origination processing and natural language processing software to deploy
and on-boarding to the back-office processing of loans, cognitive capabilities, hasseen relatively slower uptake. The year
deposits, managing investments and the closure of a customer 2016 saw many of the banks announce investments in AI
account. and are already working on integrating the technology into its
However, there is a marked variance in the rate of adoption of banking processes. However, the application of this technology
these two technologies. The inherently cautious and structurally is still seen being done in pockets and is yet to witness an
complex banking industry has been more forthcoming in organizationwide deployment.
implementing RPA, which is a less risky and cost-effective

3
The early adopters of both these next gen technologies have mainly been large global banks mainly in Europe and the US who
have the spending power to experiment with new technologies. Direct banks and challenger banks with digital only banking
models are fast emerging as foremost adopters of these emerging technologies. Challenger banks which have proliferated in
Europe and are now entering the US market and are most likely to be a bigger user of RPA and Al technologies than incumbent
banks.

Robotic Process Automation (RPA)

Robotic Process Automation, which is essentially the use of An RPA function operates by mapping a workflow in the RPA
software "robots" to automate repetitive, clerical and typically tool for the software robot to follow computer path-ways
high-volume tasks, has become one of the most sought after between screens and various data repositories. An RPA tool
technologies in banking over the last 2-3 years. According to can be triggered manually or automatically, move or populate
HFS Research, the worldwide RPA software and services data between prescribed locations, document audit trails,
market is expected to grow at a CAGR of 35% reaching $1.2 conduct calculations, perform actions, and trigger
billion by 2021. This is not surprising considering the demand downstream activities.
in the banking industry itself, where more than one third of
banking operations are manually executed with dedicated
back-office staff.

Typically, there are two types of RPA models that can be

deployed:
• Attended RPA: Deployed where processes can be
fully automated, with no need of manual
intervention.
Usually these are back-office sub processes.
• Unattended RPA: Usually deployed in partially automatable processes. Robots are deployed on an employee's
workstation, typically in front-office or mid-office banking functions.

The popularity of RPA can be attributed to its benefits and characteristics:

• Non-intrusive: Banks typically operate several large legacy platforms in each different business vertical making it very
complicated, time-consuming and expensive to connect them all together into one cohesive IT system . The non-
intrusive RPA software helps banks achieve automation without the need for complex system integration.
• 24/7 support: Unlike humans, an RPA bot can operate continuously without a drop in efficiency and accuracy levels.
By leveraging RPA, banks can achieve productivity gains of 35-50%-compounded across thousands of transactions-
enabling greater capacity and agility.
• Quick deployment: An RPA tool can be deployment in 2 to 6 months depending on the complexity of the process.
• Scalability: RPA is code-free and thus easily implemented in departments because it does not require programming
skills and business users can be trained easily to manage the bots
• Detailed audit trail: Anything performed by RPA can be recorded consistently thus providing a solid audit trail and data
pool. These key analytics can be used to optimize processes, improve the customer experience,
generate performance statistics and even use for efficient regulatory reporting.

With the above benefits in mind, banks have tested this technology across the various functions and with varying levels of
complexity. Early implementers were really focused on large enterprise functions, reconciliation, data management, high
volume repeatable processes. However, banks are now using RPA in more complex processes such as risk and compliance, data
quality testing, financial crime sanctions check, etc.

4
Applications of RPA

Challenges for Banks when deploying RPA

While robots in RPA are super-efficient they are not without flaw. A software robot can make errors when it receives the wrong data
from humans. The error will be repeated until it is manually eliminated. At the end of the day, robots are only instructions based on
a programmed algorithm to imitate humans by doing things in the same way; hence they cannot make autonomous decisions in
cases of unusual situations.

Therefore, RPA requires constant monitoring and control, corrections of the algorithm and completion of process paths. Any change
in the system in which robots move, directly affects the work of robotic agents. Banks must necessarily distinguish between processes
that can be completely automated and processes which require human intervention and apply the technology accordingly. An ideal
deployment will involve a mix of attended and unattended RPA processes.

Future state of RPA

The true goal of any process optimization is to achieve complete "Rules-based auto-
automation. This is the end goal even for banks, as they scale up mation is short lived;
their RPA technology across operations. However, as processes
that's not where the
get more complex, the level of decision making involved and
value proposition is. It’s in RPA
cognitive capability required becomes higher. Banks have realized
that the true potential of RPA is realized only when the technology
plus cognitive computing plus
is combined with Al and machine learning to make the advanced analytics plus work-
automation more cognitive and cross functional. Fintech force orchestration 1…
suppliers of RPA solution are already experimenting with Head of Innovation Lab
this model of combining Al with RPA. at Deutsche Banking

5
 Evolution of Robotic Process Automation

II. Artificial lntelligence (Al)

Artificial lntelligence and machine learning technologies are
expected to have a transformative impact on the banking industry.
The power of mimicking human thought has applications across
numerous business areas within banking, including customer JP Morgan estimates that AI
services, operations, risk and compliance, investment and trading, will account for 1.8% of
and cyber-security.
global enterprise IT
There is no argument that Artificial lntelligence is going to disrupt
the banking industry just as it is disrupting other industries.
budgets by 2021 up from
0.3% in 2016 2
Last year saw many of the large banks, both in Europe and the
US announce Al related initiatives and investments. JP Morgan
reportedly invested over $9.5 billion in technology in 2016 with a
large part being spent in robotics and AI to automate processes
and cut costs. Similarly, HSBC recently announced that it spent
$2.3 billion on improving its artificial intelligence (AI) and digital out as impediments to adopting this technology - the shortage of
capabilities around the globe. data science skills to maximize value from emerging Al technology
and the lack of the right infrastructure for deployment.
However, in spite of the hype and AI related initiatives making
the news, banks are still in the process. A recent survey by SAS on Another survey of the 200 global tier one and tier two banks
adoption trends of Al in Europe revealed that while organizations revealed that while 67% have actively deployed AI and machine
were optimistic about the benefits of Al, very few were confident learning the vast majority are still unaware of how to apply the
of exploiting the technology. Two reasons stood technology to solve business problems.

6
lt is evident that Al and machine learning technology is disruptive in its ability to provide cognitive capabilities to rule-based robotic process
automation technologies. Potential use cases of Al run across numerous applications including Al-powered chatbots to improve service and
operations, intelligent algorithms to bolster security, trading, business analysis, or the intelligent automation of manual processes.

Currently the disruption is most evident in the areas of customer service and risk management. lt is predicted that more than 60% of
customer interactions will be handled by Al in the future.

Some of the more popular applications of Al in banking are as follows:

1. Fraud detection
2. Chatbots
3. Algorithmic trading and robo-advisers
4. Recommendation engine

1. Fraud Detection:
Banks are increasingly using big data and Al to detect anomalies and fraud right from initial KYC checks to the transaction
processing stage. Most of the major banks across the globe are shifting from rule-based software systems to artificial
intelligence-based systems which are more robust and intelligent to detect money laundering patterns on a real time basis.

HSBC is incorporating technology from Ayasdi to tackle fraudulent activities such as money laundering
Use and KYC. HSBC is focused on gaining efficiencies from automating transactional-data analysis, and like
several other banking entities, it seeks to gain a significant reduction in cost by eliminating
Case #1 false positives in fraud detection. lt has already seen a 20-percent efficiency gain in initial
pilots.3

Citibank has made a strategic investment in Feedzai, a leading global data science enterprise that works in
Use real-time to identify and eradicate fraud in all avenues of commerce including online and in-person
banking. Feedzai’s technology is built using machine learning and artificial intelligence that adapts with
Case #2 the detection of new malicious threats at scale, helping business customers make data-backed
decisions and de-risk commerce transactions in real-time. 4

2. Chatbots:
This is the most sought after application for Al considering the benefits that banks will gain once a chatbot is successfully
deployed. Chatbots are a cost-efficient, yet powerful way to provide basic support for customer queries all through the day
24/7 without the need for customers to be in queue. Chatbots excel at collecting customer data from support interactions
and allow live support agents to use this information to personalize their interactions with customers.

The Royal Bank of Scotland (RBS) is using a chatbot called Luvo to automate and streamline its on- line
Use
handles simple
customer service. Developed using lBM's artificial intelligence platform Watson, Luvo handles simple
customer questions, freeing up the bank's customer support staff to focus on more difficult
Case #1
customer
questions, customer issues. 5
freeing up the
bank's customer

Bank of America introduced its chatbot, Erica, in 2016. The chatbot leverages “predictive analytics and
cognitive messaging” to provide financial guidance to the company’s over 45 million customers. As an
Use integrated component of the mobile banking experience, Erica is designed to be accessible to clients 24/7
and perform “day-to-day transactions” in addition to anticipating the unique financial needs of
Case #2 each customer and helping them reach their financial goals by providing smart
recommendations. 6

7
3. Algorithmic trading and robo-advisers:
The success of Wall Street traders using algorithmic or quantitative trading models has been widely publicized. With advancements in
AI and machine learning, banks and financial institutions are taking it up another notch by replacing human traders with AI based
software robots. JP Morgan’s asset management business recently reported that it is working on a new machine learning software
model that the firm hopes will make more efficient and profitable trades than its human counterparts are capable of.

On similar lines is the introduction of robo-advisers in wealth management. While fintech start-ups have been offering this service for a
while now, leading banks such as UBS Group, Wells Fargo and Citizens Bank have also launched their own robo-advisory solutions
which give customers access to their portfolio and offer support and customer service through chatbots, again cutting the cost
traditionally associated with wealth management services.

Use ING unveiled a new artificial intelligence bond-trading tool, called “Katana.” The system will be rolled out
Use#1
Case across the bank in 2018, promising to cut costs and speed up transactions. Katana learns from the history
of hundreds of thousands of trades and translates this into a prediction or suggested decision
Case #1 for the trader when deciding what price to quote when a client wants to buy or sell a bond. 7

4. Recommendation Engines:
Personalizing the customer's journey is a highly effective way of improving their experience and satisfaction, showcasing
key initiatives and driving traffic numbers. With increasing digitization by banks and the explosion of data online, these
engines are able to utilize data science and machine-learning to offer suggestions for related products, new offers and
other 'discovery' paths.

In 2017, JP Morgan announced the development of a machine learning based application called
Use ‘Emerging Opportunities Engine’ to be used in its Investment Banking Business. The helps identify clients
best positioned for follow-on equity offerings through automated analysis of current financial positions,
Case #1 market conditions and historical data. The technology has proven successful in Equity Capital
Markets and is currently being expanded to other areas including Debt Capital Markets. 8

Challenges for Banks in adopting Al

While the huge benefits of using Al are well acknowledged by banks, they still have to navigate through a few hurdles
which could slow down the adoption process.

8
3 Key Strategy Considerations for banks
while deploying RPA and Al technology
Like in all new age technologies, while benefits of deployment are Al
immense, deploying the technology in the right manner is critical for
an organization to maximise its potential benefits.
AI isn’t something that
can be tacked onto the side of
Some of the key imperatives for banks to consider when deploying
RPA and Artificial Intelligence is as follows:
the organization or left with a
business innovation team to do
Identifying the right processes to automate: on their own. lt needs to be a
One of the biggest errors that financial institutions commit with their core part of any transformation.
decision to automate processes is to carry out blanket automation. It You have to move at scale,
is critical for banks for identify processes that will have the most right across the enterprise 9
impact from the automation exercise. Otherwise banks could end up Director of Digital
spending on automation exercise which may not yield the expected Development, Lloyds
ROI.

Management buy-in:
The most important hurdle when implementing any new tech-
nology is to get a buy-in from key function owners across the Selecting the right partner:

organization, especially top management including the CXOs and The right man for the right job applies most when it comes to
Head of critical departments such as IT and Operations. Usually such selecting a technology supplier. Banks must first internally establish
initiatives are faced with resistance from most departments due to the objective and areas that they want to automate and pick a
fear of job losses and the general inertia associated with any change. supplier who is strong in that particular area. Factors a bank must
In some instances, these initiatives are considered to be the domain consider while selecting the right supplier includes the geographic
of the IT department and don't get much cooperation from the other location of the supplier, quality of technical support available during
departments. In such a situation, technology gets developed in silos and post implementation, past implementation track record,
and setting up a Centre of Excellence (CoE) team, that is required to customer references and of course the cost of implementation
deploy the technology across the organization, becomes a challenge. amongst others.

Upgrading legacy systems: Regulatory compliance:

Most banks have the same problem when it comes to deploying new Automation of banking processes mainly involves customer data and
technology - the cost problem because of huge legacy systems which any action that impacts customer data must be re- viewed carefully,
require hardware maintenance, software licenses and integration as any misreporting or leakage of customer data can lead to
costs. The solution is a new platform with web scale features. A key regulatory action. Hence it becomes imperative for banks to ensure
imperative for banks to successfully deploy and benefit from that the data that is automated is com- pliant with current and
intelligence operations will be to first upgrade legacy systems. upcoming regulations.
Intelligent operations entail the use of advance technologies such as
AI and RPA and layering these technologies on legacy systems will
not allow banks to realize the true benefits of the implementation.

9
4 Outlook for Al and RPA in Banking

The goal of RPA and Al based technologies is to reduce the Robotics, AI, & Data are
repetitive work done by humans thereby allowing people to three technologies to watch out
concentrate on more complex and creative tasks that for. While these technologies
machines can't handle such as maintaining customer continue to mature it’s the
relationships. Banks have acknowledged this opportunity ability to leverage the
and are already moving in the right direction with the larger convergence of these
banks and neo banks leading the way. capabilities that will influence
significant business outcome 10
The future state of a banking operation is most likely to be Chief Technology Officer, Wells
Fargo Group
completely automated, with human intervention only
involved in critical decision making aided by insights and
recommendations from Al enabled technology.
deployment and evolution of Al technology. A key concern

While RPA has reached a relatively mature state of adoption amongst industry exponents and employees alike is the loss of jobs
within banks, Al applications are still the bastion of large due to these deployments. Banks and other organizations are
banks and of course the growing breed of neo banks and trying their best to allay these fears by consistently reassuring
challenger banks. For tier two and tier three banks, AI is still through messages that their goal for Al is not to replace human
in the ideation stage with many still unable to leverage the employees, but to help them do their jobs better.
technology due to legacy systems and manual processes.
With studies indicating that spending in cognitive technologies

But the path has been set, with many banks upgrading their worldwide will grow by more than 50% for the next three years,
core systems and serious discussions in the board room on Al is expected to be the secret ingredient for any bank expecting
investing in and adopting emerging technologies. to compete in the market. At the end, the most successful banks
will be the ones who are first in the market and deploy the
The industry wide uptake of Al however will take a while, solution in a holistic manner rather than for individual processes
considering the various elements required in the successful and build around a customer-oriented strategy.

10
References

1. https://www.cio.com.au/article/614823/rpa-proving-its-transformational-value-deutsche-bank/?
pp=2 [Accessed on: March 2018]
2. https://www.investors.com/news/technology/ai-in-business-future-of-artificial-intelligence/
[Accessed on: December 2017]
3. https://www.reuters.com/article/us-hsbc-ai/hsbc-partners-with-ai-startup-to-combat-money-
laundering-idUSKBN18S4M5/ [Accessed on: March 2018]
4. https://feedzai.com/press-releases/citi-ventures-makes-strategic-investment-feedzai/ [Accessed
on: January 2018]
5. https://ibsintelligence.com/ibs-journal/ibs-news/sibos-2016-rbs-taps-up-ibms-watson-for-ai-
powered-chatbot-luvo/ [Accessed on: March 2018]
6. https://www.cnbc.com/2016/10/24/bank-of-america-launches-ai-chatbot-erica--heres-what-it-
does.html [Accessed on: April 2018]
7. https://www.ing.com/Newsroom/All-news/Katana-gives-bond-traders-a-cutting-edge.htm
[Accessed on: December 2017]
8. https://www.jpmorganchase.com/corporate/annual-report/2016/ar-ceo-letter-matt-zames.htm
[Accessed on: April 2018]
9. https://www.i-cio.com/management/insight/item/banking-on-an-ai-future [Accessed on:
December 2017]
10. https://mill-all.com/blog/2017/08/08/interview-with-scott-dillon-cto-wells-fargo-keynote-
speaker-at-fsi-transformation-assembly/ [Accessed on: April 2018]

10