Indian banks adopted artificial intelligence technology only relatively recently but are catching up fast. The private sector leads in deploying chatbots that can handle many thousands of customer interactions, but where will this leave staff? Rekha Menon reports.
Technology is the cornerstone of India’s banking sector. And while Indian banks may have embarked on their technological journey only in the mid- to late 1990s, they have more than made up for this relatively late start. Unencumbered by legacy platforms, they leap-frogged ahead of counterparts in other countries, deploying the latest technologies from core banking and payment systems to risk management and, more recently, digital strategies.
A November 2017 report by analyst Gartner predicted that with the growing investment in digital payments infrastructure, the overall IT spending by Indian banks and securities firms would soon pass $9bn.
From human to machine
Today, as Indian banks look to gain a competitive edge by deploying cutting-edge technologies, artificial intelligence (AI) is gaining ground. Generally defined as a machine’s ability to exhibit human-like intelligence, AI helps machines perform complex and intricate tasks that until very recently were considered impossible without human cognition.
This includes recognising complex patterns, understanding unstructured data and forecasting. AI covers a wide range of technologies and application, such as natural language processing (NLP), or the interaction between human language and computers, which is used in virtual agents or chatbots; and machine learning (ML), which involves algorithms 'learning' without rules-based programming.
AI and ML have the potential to revolutionise the customer experience, especially at the ‘last mile’, by providing greater levels of personalised service and greatly improving back-office efficiencies at financial institutions, according to PwC’s Fintech Trends Report (India) 2017. Labelling AI and ML as among the “hottest technologies to watch”, the report estimates that more than 36% of the country’s large financial institutions are already investing in these technologies, while almost 70% report that they are planning to in the near future.
The rise of the digital assistant
India’s largest bank, State Bank of India (SBI), took its first step into the AI space when it launched its online chatbot, or digital assistant, in 2017 to address customer queries and help them with basic banking transactions. In recent years, the public sector lender has invested heavily in the digital arena to revamp its staid image, as well as to improve performance, and the AI foray is aligned with the same strategy.
SBI chairman Rajnish Kumar, in his investor presentation announcing the bank’s annual results for financial year 2018, mentioned that along with the chatbot, AI would also be used for profiling customers based on their mobile digital footprint, social profiles and bank statements. “AI will help the bank enhance the customer experience,” he said.
Indian banks typically start off with digital assistants that can converse virtually with both prospective and existing customers to provide information and conduct basic banking transactions. Rishi Aurora, managing director, financial services, at Accenture, says: “Chatbots are the entry stage of AI in Indian banks. AI solutions are also helping banks to improve their operational efficiency, while a future area of focus is using ML-based algorithms to build [decision-making] solutions.”
Private banks in the lead
While a couple of other state-backed banks are also evaluating chatbots, the lead in the AI space is unsurprisingly being taken by India’s top private sector lenders, which are usually ahead on the banking technology adoption curve.
For example, HDFC Bank, the country’s largest private sector bank, has deployed AI solutions in customer service, employee knowledge management processes, as well as in risk and portfolio management to provide deeper data insights. Its chatbot, Eva, has been operational for about six months and handles more than 700,000 customer conversations every month and can simultaneously deal with about 3000 concurrent customer conversations.
“AI is most effective in customer-facing workflow processes, as well as in processes that are repetitive in nature and where data science is important, such as in risk management,” says HDFC Bank digital banking country head Nitin Chugh.
It is an opportune time for the uptake of AI solutions in the banking sector, he adds, saying: “At the heart of AI is data science. Today banks have a lot of data, most of which has been created in the past two years. At the same time, algorithms have evolved and computing power has improved dramatically.”
Deepak Sharma, chief digital officer at India’s fourth largest private sector lender, Kotak Mahindra Bank, agrees. “AI is a transformational technology which depends on data. Today banks have huge amounts of data that is both structured and unstructured. The availability and richness of data has gone up in the past two years, while the cost of storage and computing has gone down,” he says.
Kotak embraces the bot
Kotak Mahindra Bank has extensively deployed AI in both its front- and back-end systems, from providing an automated mutual fund investment selection service and generation of wealth management reports for customers, to conversational banking and fraud, as well as risk management. The bank has also deployed a digital assistant for its senior management, which enables them to easily retrieve analytics reports in real time.
Kotak’s AI-powered conversational banking chatbot is integrated with the bank’s phone banking helpline and augments the traditional interactive voice response (IVR) system. Mr Sharma says it is bilingual, available in both English and Hindi, and combines conversational intelligence with human-like natural dialogue.
“We are convinced that AI is not a fad,” he says, adding that the bank has been able to see clear business benefits from the first day, from faster customer service, better insights, faster decision making, reduced fraud and automation of back-end workflows. “Our AI-powered process bot for working capital limit extensions, which monitors email requests and approvals to facilitate limit extension on working capital loans, has seen a 60% turnaround time reduction and full-time equivalent reduction by 50%.”
He adds that since launch, the AI-powered chatbot has successfully addressed more than 2.7 million calls, saying: “More than 70% of all customers calling the contact centre are interacting with the voice-bot every day. It has enhanced our IVR conclusions by three times.”
An AI architecture
Yes Bank, India’s fifth largest private sector bank, is taking a slightly different approach in its AI journey. While the bank has deployed a few AI solutions from external vendors, such as a bot for assisting customers on the Yes Bank website, and another for loan products, the bank is working along with Microsoft to develop an AI-based platform on which various services will be developed.
There will be three components of this architecture: a channel to initiate the interaction, an NLP platform to identify the intent and, finally, connectivity to various databases. “We are working very closely with Microsoft’s engineering team to develop our platform, which should be ready for both customers and employees in the next few months,” says Amol Pai, president, at Yes Bank. AI is still a new technology, he adds, hence one of the challenges banks face is that while many people speak about AI-based solutions, not many really know enough about the technology.
“This is a new space and banks are interacting with several small players. The positive aspect of this is that the amount of investment required by banks is relatively small at this point in time. However, the bigger challenge would be on how these small, point solutions scale up,” adds Accenture’s Mr Aurora.
Mr Chugh of HDFC Bank says a key challenge of AI projects is related to the sanctity and accuracy of the data. “There are several implementation failures because the data is not clean enough. Additionally, the pool of data scientists that understand data algorithms is very small. We also need to give time for the technology to evolve further.” However, he points out that Indian banks are on par or ahead of banks in mature markets with regards to AI adoption.
Reskilling staff
Mr Chugh believes AI will not lead to job losses, as is often the fear. According to analyst firm Gartner, by 2022 one in five workers in enterprises engaged in mostly non-routine tasks will rely on AI to do some portion of their jobs. “AI will augment capabilities and some workflow will be automated. While old jobs might go because of business process changes, new jobs will emerge,” says Mr Chugh.
At Yes Bank, where the goal is to replace 60% of customer service agents with chatbots in the long run, the focus will be on retraining and reskilling existing employees, the bank says. Mr Sharma of Kotak Mahindra Bank adds: “As businesses are growing, the focus isn’t on reducing headcount. AI will help in making businesses more efficient while automating mundane, routine tasks.”
