Mapping AI to ROI for Banks and Financial Organizations

Artificial Intelligence (AI) can be categorized into three application domains from a banking and financial services perspective:

Cognitive Automation: Cognitive tools to develop deep-domain specific expertise and automate relevant tasks

Machine Learning: Analysis of customer activities, NLP and recommendation algorithms to provide insights to customer, drive engagement, offer new products and open up new revenue streams

Cognitive Computing: Extraction of concepts and relationships from various data streams to detect data patterns and relationships to derive insights which can be converted to action items

Financial organizations have a greater opportunity to leverage artificial intelligence as they have access to a huge amount of data for artificial intelligence algorithms. Also, consumers are willing to share personal insights if they can get greater value in return. And most banks and financial organizations are already utilizing some form of analytical tools, so relevant algorithms and processes are also well-defined and standardized.

Using the components of artificial intelligence, there are several use cases in the banking domain where AI can be applied:

Cognitive Computing: Algorithmic trading and automated investment management

Machine Learning: Next best action and next best offers, wealth management robotic advisors, fraud detection and anti-money laundering

Natural Language Processing: Chatbots, robo-advisors

The diversity of the use cases indicate that AI has the potential to impact multiple touchpoints of the banking system ranging from adherence to compliance to greater customer engagement. In terms of customer engagement there is potential to democratize activities like wealth management advisory which till date was available for only a select group of banking clients. Also, banks can have greater insights about customers, which can lead to faster decision making and better levels of engagement. This will help them offer new personalized products thus opening up new value streams.

Besides enhanced customer engagement, AI also brings along the opportunity of automating procedural and repetitive tasks. And with ample amount of data, it can also contribute to automation of support activities like performing root-cause analysis, defect analysis, automated query resolution etc. This enables support staff to concentrate on complex queries thereby reducing turn-around time, improving efficiency as well as enabling time and cost optimization.

Hence we can see that AI and its applications have the potential to impact the ROI for a financial organization in a two-pronged manner:

They influence both the revenue generation, and expenses of a bank/financial institution. Diligent and effective implementation of AI use cases can lead to greater revenue generation as well as expense reduction. The contribution of AI on banks’ ROI is expected to go up further as more data and related insights become available and new AI use cases come up.


Internet of Things (IoT) and Software Development

Benefits of IoT on hardware product maintenance is well known. Devices sending information via internet to data centers and receiving inputs which help product companies support customers facing difficulty in usage of products like cars, laptops or any machine. The information received and analyzed can be used for improving the product. But can this be extended to software development or is it applicable only to hardware development?

Software product development teams typically get requirements from both internal and external sources. Internal sources include teams like R&D, presales, implementation, and compliance. External sources include direct interaction with customers or hiring consultants who have diverse market, industry, customer experience. Additionally if we get requirements from the software itself based on actual customer usage of product features in terms of performance, security, usability, APIs, and digital tutor, we can incorporate those feedback to improve the product.

IoT via Internet of Services (IoS) will prove very valuable and address many concerns with respect to how software interacts with other software applications and hardware devices thus becoming an effective source of requirement gathering. Information will get communicated faster and will be available to product teams for analysis, and obtain faster response for critical items. Thereby time to market for new versions will be reduced.

Applications in IoT will understand issues in application based on customer’s usage, interact with multiple versions so that patches available in one version get automatically deployed in other version, thus leading to support effort reduction. Applications are connected to the Web in order to transmit, receive and communicate with other applications with or without interference from humans. Security and privacy issues need to be taken into account in this interaction.

Technical backlog (in certain cases functional) can be created based on real time inputs received – based on software usage. Combination of multiple sources of inputs will add value at various stages of software development. Additionally we can look at automating impact analysis (functional, design, test plans, test execution) for software development so that it is not person dependent. The groomed (frozen requirement, design, test plan) items will be available to any team to deliver based on priority.

Clearly IoT systems are capable of transforming the way we gather information, analyze and further decision making. It also presents opportunities to create smarter applications to be run efficiently and automate many routine tasks.

Future “Branch”ing

For most of us, visiting a bank means visiting a branch. Obviously, we do not need to visit the head office of a bank for simple tasks, or for that matter even a crucial requirement. Branches play a vital role for customer interaction and ensure seamless service. Though there are various other modes like ATM, Internet banking and mobile banking including customer service through phone banking, most customers are happy to visit a branch as it has a human touch. With the advent of new technologies and innovations, the future of a branch will change drastically. Listed below are some of the expectations that will arise for the branch:

Automation: As many feared, automation and robotic processes will eliminate redundant jobs across industries, and this is applicable for the banking industry as well. Especially simple jobs like accepting cash, printing statements and passbooks etc., can be automated. In the future, when you visit a branch for any of these automated tasks, you may be assisted by robots! Also, branches use robotic and machine learning services to carry out jobs like lobby management, request acceptance and guiding customers to the respective service oriented counters/desks.

Electronic Branches: Even today, we can do several transactions at the ATM. Banks are extending their ATMs as electronic branches. At present, most of the electronic branches are mapped with the branches next to them. In an electronic branch as well you can carry out transactions like withdrawals, deposits, statement printing and video conferencing with a customer service officer. Perhaps in the future, electronic branches will be extended even to carry out specific services with the intervention of bio-metric authentication. Users can initiate a transaction through mobile or Internet banking, generate a transaction reference number, authenticate it in an electronic branch using bio-metrics, and complete the authorization of the transaction after uploading a specific set of documents. For example, a customer who wishes to get a demand draft payable at some location, has to visit a branch or place a request online, and same will be couriered to the address. In the future through electronic branches, the customer can place the transaction through online or mobile banking, visits the e-branch, after multiple authentications, the demand draft will be printed through a machine and delivered to the customer on the spot.

Analytics: Even right now, branches are using analytics for most things, however, going forward behavioral analytics will play a key role, based on customers’ transactions over time, branch employees will be advised to suggest suitable products like loan, mutual fund, insurance etc.

Conclusion: Banks are investing a lot in the digital space and other channels, but it doesn’t mean that the significance/number of branches will be reduced to a great extent. Though machine learning, automation and other tools help banks in reducing costs, human intervention in banking transactions is still required as branches contribute quite a lot to a bank’s revenue through cross selling. Banks do not want to lose out on those revenue opportunities. Even though the demographics of the branch might change, wealth management will be an area of focus, as most resources will be involved in non-redundant jobs, and will be able to focus on growth parameters.

Digital payments with Intuitive Interfaces

Post demonetization in November 2016, the Government of India encouraged the people of the country to take up digital payments in lieu of physical cash. Lottery schemes were launched to promote digital payments on the UPI and BHIM apps. This led to a significant increase in the adoption of digital payments, and players like Paytm hugely benefitted from this trend. Local kirana stores, that were known to accept only cash, followed this trend and started displaying QR codes. All this marked the first wave of behavioral shift in regular payments going digital.

Urban and sub-urban population was quick to adapt to the new change. Youth were quick to adapt too. Higher and middle income groups were quick to adapt. But there still was a large segment of population that was not able to embrace the digital payments revolution. For days together, long lines were seen in front of ATMs and banks for physical cash.

The reasons for not embracing the digital payments were many. Couple of them were:

Mistrust on digital payments – Many people had a big mental block against using mobile phones for financial transactions, even though they could afford and had access to mobile devices. People could not contemplate parting physical cash and having to transfer money from their accounts via mobile apps was not seen as being very secure.

Illiteracy – Illiteracy is not an impediment for business but it is definitely an impediment in transacting digitally. This segment of the population gets turned off at the stage where they need to sign in with a username and a password on their mobile apps, which is typically the first step to initiate any transaction.

To make digital payments reach the next billion, the banking community and the new players in the industry need to take significant efforts towards simplicity, localization and security. A step in this direction would be to make user interfaces intuitive and user friendly.

I have listed some ideas here:

App Login with Pattern Lock: Today we have logins based on a username and password or a 4/ 6-digit pin. This does not cater to the people who are not versed with the language. Enabling authentication in a local language might just do the trick, as most of the mobile phones support these languages. So an option of pattern unlock should be given for the app based on preset patterns. For security reasons, this authentication may be complemented with finger print or facial recognition.

Intuitive Icons for Functions: Most of the icons in current mobile apps are not intuitive enough for an illiterate person to operate. A pay button will have ‘Pay’ written over it and some icon which is meant to depict a payment action. Such buttons should be standardized and should be common for all apps.

In today’s age, even a toddler can operate a smartphone. For example, a YouTube button resembles a big Red Play Button. By observing, a toddler learns to not only open the YouTube application but also to swipe, select and play the videos she wants. Swipe and select are standardized actions. Similarly a play icon is standardized. By observing the usage of these actions and buttons, a toddler learns how to operate an application. Similarly, a banking/payment app should have standardized buttons and actions which enables consumers across economic strata to use it intuitively.

Selecting the Payee: With UPI or the many digital wallets present today, anyone should be able to make a payment by just giving the phone number in the payment app or by selecting a contact. It should be like selecting a contact to dial a phone number. The contact list may have a photo of the person along with the phone number and name. Most frequented payees can be provided the additional benefit of faster payments on top for a quick payment.

Denomination management: Not everyone is comfortable reading and understanding the large numbers on the screen. At most, they are comfortable in counting a smaller set of numbers. Today, many hawkers or vegetable sellers are well versed with counting currency notes and doing the basic calculation of number or notes and subsequently the total amount. Recently, the government released new currency notes in distinct colors making it easier to identify and distinguish the value of a note.

Similarly, the amount of money a person has can be displayed in an app either in the image of the currency notes or some color coded icons. For example, if a person has Rs 4850/- in his account, it can be displayed as 2000 currency note with a number 2 (count of the notes) next to it, a 500 currency note with a number 1 (count of the notes) next to it, a 100 currency note with number 3 next to it, and a 50 currency note with 1 next to it. Instead of currency notes, these could be some other colored icons and all these icons or images could be displayed like a thumbnail. This serves two purposes, one is that cash in the account can be easily counted and second is that the imagery serves as a replacement to physical cash in the wallet/batua which is more of a psychological effect of holding the cash.

Cloud Service for New Age Applications

Bob has an idea to launch a gaming application which can be accessed across geographies. But he has limited funds, so he prepares a business plan to attract investors. In order to avoid the initial cost of setting up data centers he decides to use a renowned cloud service provider. One that will take care of providing low level hardware and software infrastructure on a pay-by-use model to host applications on cloud. Low level hardware includes servers whereas low level software includes a database management system. The cloud service provider will provide necessary security of the servers and data so that there will not be any threat of losing customer information.

The decision to use cloud service is beneficial to Bob because the initial cost is significantly reduced. Now, he can focus entirely on building his gaming app, and need not worry about the underlying infrastructure. At this stage, since Bob cannot estimate the possible traffic to his website, using a cloud service would benefit him as it offers the flexibility to add more servers and more storage capacity dynamically. This ensures business scalability with service available round the clock. In addition to ensuring high data availability, data can be stored in multiple locations using distributed storage. Another benefit is, the possibility of validating applications with more number of servers which otherwise would not have been possible with a limited number of servers. Additionally, the performance of the application with any number of users logged in can be validated.

Another important aspect that Bob needs to take care of is security when using cloud services while designing the application. In addition, he will also need to ensure application security. A cloud segment PaaS (Platform as a service) like the Google App Engine could be a good solution for Bob as it requires less knowledge and skill to develop and deploy applications, thereby leading to rapid time to market. Moreover, load balancing is automatic if his request rate increases, and billing will be based on the CPU cycles he uses and the bandwidth. Bob needs to be aware of APIs (Application Program Interface) to access infrastructure components and follow protocols set by the platform. He can also collaborate with social networking sites like Facebook for his gaming application and leverage active users on the social media platform.

Clearly cloud computing and storage services has paved a way for Bob to market his business proposition to investors effectively!

Corporate Bank Branch of the Future

In the corporate bank branch of the future, the relationship manager will spend more time with corporate customers and less inside the premises of the bank. He will be well informed about the latest developments in the company thanks to the internet. While on the move, he will access the bank’s application to approve transactions requiring his go-ahead, as well as provide recommendations for any renewal of credit facilities.

New Customer: He will update details on potential opportunities in the customer management platform and inform the credit analyst to prepare a credit application based on the details shared by the customer. Meanwhile, he will also inform the deposit team about a cross-selling opportunity. The necessary credit approval request will be sent to the approving authority.

Existing Customer: He can query limit availability and inform the customer accordingly. The customer would also be able to access these details using the banking application. In the case of problem customers, a proposal for debt restructuring would be sent to the credit analyst to help bring repayments back on track. Early warning signs can also be arrived at based on any red flags during discussions with the customer. A report with the key details can be updated in the system after every visit.

The manager, despite spending a productive day with customers, will not feel out of touch with his colleagues and the bank’s operations. This is because he is able to focus on his core competence and take help from other teams wherever required.

Digitization has helped the relationship manager access the bank’s systems to inquire, collaborate, prepare visit reports, approve transactions and do much more on the move.

Can AI help lenders prevent financial fraud and protect personal data?

Preventing financial fraud and protecting personal data requires intelligence and self-learning along with computation power. Artificial Intelligence is the key to making a system an efficient e-auditor by removing manual intervention. In some recent cases, the involvement of bank employees in a financial fraud has been amply evident. Artificial Intelligence is needed to reduce the reliance on human auditors without losing the knowledge and self-learning capability of humans.

The system for e-auditing can be made artificially intelligent with the following concepts:

Decision Trees

Decision trees are a popular tool in machine learning. The first step in a decision tree is to create a knowledge base. Knowledge base is similar to the memory component of our brain. Knowledge base holds information like – “If a transaction is done from a terrorist country then it is fraud”, “If a transaction involves multiple currencies and the volume of the transaction is high then it could potentially be a fraudulent transaction”. Knowledge base holds information about the parameters and the decision logic. It could start with limited information and data, but the system learns and updates the knowledge as it goes along. Whenever an event occurs the system scans the knowledge base to arrive at a decision. The knowledge base is represented as a tree and the decision is taken by traversing the tree.

Fuzzy Logic

Computers recognize data as zero or one. Fuzzy logic is to have a representation between zero and
1. Some of the Fuzzy words that are difficult to represent in computer language are “Beautiful”, “Tall”, “Roast the bread till it turns light brown”. Fraud is also a kind of fuzzy word and we cannot directly identify an event as a “fraud” or “not a fraud”. A huge volume of transaction in a day cannot be considered as a fraud for a corporate customer but it could potentially be a fraud for a retail customer.
So the output from decision trees cannot be used as-is to arrive at the conclusion. Based on the decision from the decision trees a fuzzy set is formed and the events with the highest cardinality are considered as fraud suspects.

Neural networks

To make a system as intelligent as a human brain, we need to borrow some concepts from the functioning of a human brain. Neural network adopts the concept of self-learning and pattern identification from neurons and axons. Human brain remembers and learns the instance based on the impact created by dendrites. It is similar to an impact created by a stone thrown to a clear water. Learning will done based on the communication between neurons through dendrites1.
Once the fuzzy set is created and the fraud suspects are identified the system needs to be sent for learning through neural networks. Human intervention is required to correct the decision and based on the weightage assigned to different parameters the system learns by updating its knowledge base.
To conclude, with the help of AI tools bank lenders can prevent financial fraud and protect personal data.

Reference 1:

Personal data everywhere?

Early morning, I received a call on my cell phone from a consumer finance company explaining to me some consumer finance/EMI offers. After a few hours, I received a call from one call center of a bank selling credit cards about life time free card with other offers. In the afternoon, I received a call from an Insurance company about some insurance offers and insurance tied with investments. These are a few calls each one of us typically receives on a daily basis. I don’t consider myself to be a high net-worth individual/very important person/person in the public domain to be receiving these calls so often. Obviously I have not registered myself in the “do not call” registry website and immediately felt the need to do it.

In the afternoon, I checked on to a web site of a travel company provider to look for some options for travel during holidays and these were very casual checks, not planned ones. When I went to another site using the same laptop/IP address, immediately popups for travel destinations I had looked up earlier appeared. Yet another instance was while trying to make a mobile payment – checking the quick payment options on a google engine search returned multiple results about relevant websites. I tried opening a website I had visited earlier and immediately my mobile number and mobile account number popped up without any registration or login. Since the earlier payment was through credit card, the entire credit card information except for CVV was auto-populated into the screen though I have not enabled autofill or saving of credit card data on any sites I visit. Though CVV is captured in encrypted format and credit card payments are further authenticated using multi-factor authentication techniques, this is still a grave concern to me.

There are other sites where one registers for making purchases/online shopping/ecommerce and the access could be browser based or through mobile. In most of these sites, apart from name and e-mail address, mobile numbers, residential address with PIN and other credentials are given. There is also an increasing tendency to source PAN details, Aadhar, KYC or other forms of identity details relevant to country which are shared for completing the registration process/formality. There is also a message to comply with KYC guidelines asking for mandatory documents to register users on many online shopping /establishments. In the urgency to complete the registration, we tend to furnish this information in an ad hoc manner without a recheck. People are also lured by the freebies /discounts being offered through the ecommerce platform for a particular event and are not sensitized by the confidential personal information being captured/shared.

A few social media sites are mediums where data once used/searched by an existing user gets stored and in seconds, the search data is shared or used by multiple other websites for sourcing some other information. Needless to say, there is an analytical engine running over these sites which enables collection of the data, analysis of the data, tracking the IP address and the next time when the user logs into a website it surprises them with the information one is searching. There is also a tendency to upload photographs/images, update of status by an individual for important events like birthday celebrations, awards and ceremonies or important places visited which could be misused.

The mobile numbers are possibly shared by service providers/agents of the companies to different sources with/without a fee. The data on mobile numbers possibly gets extracted secretly, gets shared with other like-minded companies and is used for calling purposes. Personal data is available on mobile/laptop and used/shared through the network making the data really vulnerable.

The extent of data that is tied to or associated with mobile/credit card/PAN/Aadhar is enormous. Some of the data may not be known to immediate family members but might be available to a select set of vendors/company, and is potentially at the risk of getting compromised for malicious intent.

The EU parliament approved the GDPR (General data protection regulation) in 2016 to harmonize data privacy laws across Europe, and to protect and empower the citizen’s data privacy. However, this is in a very early form and all citizens across the globe are susceptible to data theft/attacks/breaches in cyber world currently. It will take time to have laws framed in terms of controlling the data, sharing the data, protecting the data, using the data, processing the data and till that time, it is up to the individual to safeguard to the extent possible.

To conclude, with digitization and communication revolution rampant across the globe, an individual is vulnerable to a data threat/piracy of personal information – mobile number, card number, identification details, bank account number (in a few rare cases) and the like. Though there are in-built security measures for accessing sites and for conducting financial transactions that may require multiple authentication modes, the threat still looms and one needs to be wary of the information being shared in the cyber space. Till cyber security and protection reaches a stage of maturity where there is a collective rule/regulation encompassing the devices and websites with technology assistance for controlled data sharing and processing across countries/geographies, it is up to the individual to ensure adequate measures in using/sharing personal data over the web.

Building a culture of analytics for all

Knowingly or unknowingly, we all do analysis and decision making in our daily lives. It could be deciding whether to hit the gym, which doctor to go to, which school to put your kid in and even for that matter whether to reward a bonus to your maid or not. The only difference when it comes to corporate life, we somehow are trained into the habit of taking instructions and following the same.

Corporates are readily equipped to build analytics driven products, but to build an analytics-driven culture is not easy. Culture cannot be brought in from outside, it has to evolve from within. It has to be lived. We can start by questioning ourselves – is it there at my workplace? Are we enough curious enough to understand the what and the why of what we do on a daily basis at our workplace? No points for guessing. Yeah we are still not there.

Internal clients – “Ohhoh… I am always left out”

To put it in simple words, analytics means structured problem solving. Not many at workplace acknowledge in the first place that there is a problem at hand to think and solve. Most workplaces understand and are geared up to provide analytics solutions to their external clients. But what about internal clients and where does analytics fit in? Will try to put across by way of examples – Do we use decision models to decide which clients to pursue? What is the expected client traction for the current product roadmap? Was analytics used to determine which category of opportunities resulted in best results? Can I prescribe what product features are required if I expect a % of turn around? This is where leaders can make a difference and educate their internal teams to identify the problem statement and inculcate the habit of applying analytics in the daily jobs being done. Walk the talk is the mantra.

“Curious for data and knowledge… I might just burn my fingers”

Once we cross the first hurdle to bring the team on common line to identify the purpose of analytics at workplace, the next road block is when teams have to overcome their own perceptions and fears. Some f the common myths that teams hold are – “The data is not there “, “It’s a lot of junk data and results may not be accurate”, “I will lose the power if I share the data”, “it’s too time-consuming, and the cost required to analyze is way too much”. A lot of us ask the right questions but how many of us try to get the right answers? I think the best bet here is to simplify the problem statement and encourage all to find answers. Organizations must transparently reward people to encourage sharing and analyzing data. Once a few rewards are won and the process is in place, the skeptics start losing ground.

“Old ways are always the best ways… why are u disturbing the set up!!!”

Then the next hurdle, once everything is in place, is to implement the outcomes of analytics and to get the team to adhere to new processes. That’s when the trouble doubles up. At times mix and match of teams and cross-skilling can deliver surprising results in overcoming resistance. The changes may not necessarily be only in numbers or end results, but also in identifying training needs, staffing needs etc. Here team leaders and managers need to support the team by giving them the training required and should give a quick view into the results as well.

“Trust Anchor”

The quality of data driven decisions in itself acts as a trust anchor for everyone. Decision making process does not always have to be behind closed doors. It can be more transparent and should involve employees at all levels who embrace analytics. For example, base level employees can contribute to data collection and can apply the tools to come up with the diagnostics for the higher-ups to view. A sense of ownership and a view into the entire decision making process results in increased employee engagement. A data-driven culture evolves and matures through such engaged employees.

Cryptocurrencies – a fad or here to stay?

Cyptocurrency, a buzz word in the recent times has gained significant prominence since 2009 with the advent of ‘Bitcoins’. As most of us know, bitcoin was conceptualized by a person or a group under the name Satoshi Nakamoto. Following Bitcoin, many other cryptocurrencies have come into the market. These currencies work in an economy parallel to the existing economy which does not revolve around the Government, central banks, regulations etc. Cryptocurrency transactions are effected on block chain using the concept of a distributed ledger. Transactions are made secure using the concept of Hashing. The use of private key and public keys allows the transactions to be discreet.

In the absence of a regulatory body, these cryptocurrencies give a free hand to transact. The transactions are hashed and the private key concept makes the transaction viewable only by the recipient. The transactions are difficult to track. Due to no monitoring or guidelines, there is a concern that these currencies could be used for activities such as money laundering to evade tax, and other kind of frauds. Many countries / Central Banks across the world have either warned or banned the use of cryptocurrency. In the recent union budget 2018 in India, cryptocurrency has been classified as ‘not a legal tender’. But this has not deterred people from transacting in cryptocurrency. Cryptocurrencies have a high market value making them an attractive investment option. With the investor interest cryptocurrencies have enjoyed so far, one wonders if the cryptocurrency market has already reached a saturation level, and if it is advisable to make investments. With the volatile situation of cryptocurrencies, it may turn out to be a risk to invest given that regulations are not uniform across the globe.

Central Banks can look at considering cryptocurrencies as regular currencies and the cryptocurrency transactions as any other digital transaction. Just as multiple different currencies exist across the world with each country having its home currency, cryptocurrencies can exist in parallel to regular currencies. Central Banks can formulate guidelines to bring about regulations for the cryptocurrency market. There are advantages of having co-existing virtual currencies. Some of them are – cost effectiveness due to the complete digitalization, the market value of the cryptocurrencies gets regulated thus protecting the interests of the people. With regulations in place by the Central Banks, fraudulent transactions can be tracked and monitored. The underlying blockchain technology ensures that the transactions are more secure. Ultimately it becomes people’s choice whether to transact in regular currency or in cryptocurrencies. Also, if the cryptocurrency market is regulated, untapped population may invest / transact in cryptocurrencies.

The debate about whether cryptocurrencies are here to stay is on. The future for cryptocurrencies looks uncertain. And with their growing interest and popularity, how Central Banks will handle the volume of cryptocurrency transactions, is also highly unclear. At some point, a uniform and consistent decision on whether to accept or decline cryptocurrencies needs to be made across the globe. Varied stance and disparate policies across the world only serve to cause confusion and make people vulnerable to risk and fraud.