Will AI in Corporate Banking Result in Robo-Bankers or Cyber-Bankers?
Way back in 1956 the field of modern Artificial Intelligence (AI) research was borne, and like most new ideas, AI’s early advocates thought they would make extremely rapid progress. However, that early promise turned out to be virtually impossible to realize progress was extremely slow, and many people turned their attention to other areas.
More recently, several advances have come together to give new life to AI, including improvements in computing power, a focus on solving specific problems – rather than general intelligence, access to huge volumes of data – powered by the internet and a new approach – loosely known as deep learning.
The net result of all this is that AI systems today can do things unheard until recently – from beating the world’s best chess players to writing pop music. AI is being put to more serious uses, including diagnosing cancer, powering driverless cars and transforming education.
But what about banking?
The idea of a “machine” that does exactly what it is programmed to do, that never gets tired and never looks for a salary increase is incredibly attractive to many industries. When combined with the ability to absorb and interpret vast quantities of data tremendous quickly, AI offers tremendous potential to banking. However, while banks are one of the biggest consumers of information technology, the rapid advance of technology is leading to a new set of extremely agile competitors.
As 56% of global annual bank revenue is generated by corporates, it is little wonder that more than half of all the $78 billion invested in fintechs has been targeted at towards corporate banking. Sleek and efficient offerings around payments, foreign exchange, advanced analytics, and supply chain finance are redefining service, creating lucrative niches, and extending corporate banking activities from the corporates to the small-business segment. A recent BCG survey had more than 65% of corporate respondents stating that banks do not understand their transaction banking needs well enough.
But while the fintechs have some significant advantages the banks have many of their own, including established relationships and networks, trusted brands and a deep understanding of the complex regulatory landscape. The question is how can banks combine these advantages to ensure they retain their position as the customer’s principal bank.
Beat them at their own game?
Many people, particularly technologists, as espousing the view that data, or Big Data, is the answer to most problems – and banks have data in abundance. Corporate banks are now seeking to unleash the potential trapped in their customer data. They are using systems to analyse structured and unstructured data, turning raw data into actionable intelligence. As AI adoption gathers pace we believe there are four key areas of focus:
- Making the customer onboarding process faster, better and more customer friendly
Until now the customer onboarding process has been very cumbersome, error prone and time consuming. On an average it takes 1-2 months to onboard a customer, where the same information needs to be entered multiple times into different systems in different formats. Step changes in technologies such as Robotic Process Automation (RPA) and Natural Language Processing (NLP) are being combined with improvements in Optical Character Recognition (OCR) to deliver a frictionless experience. Benefits include the ability for customers to upload documents from their mobile phones from which the required data is automatically extracted.
- Delivering differentiated cash management products and services
In search of efficiencies, corporates are increasingly turning to new entrants – such as fintechs – for their day-to-day cash management activities. For banks that are comfortable with taking on a more advisory role, this can provide tremendous opportunities. Using AI tools and customers’ data, banks can carry out detailed analyses of corporates’ transaction flows, payables, receivables, assets and liabilities. Armed with this information, banks can provide advice on the timing of bill payments, offer discounts for early payments or provide investment advice. Not only does machine learning help banks improve their advice but it is also being used to improve product capabilities. Recently Bank of America Merrill Lynch launched a New Payment Reconciliation solution with AI capabilities which is designed to help customers reduce days sales outstanding (DSO) and improve cash forecasting.
- Increase agility in trade finance
The continued rise of open account trade is putting approx. $50 billion per annum in banks’ trade finance revenues at risk. To counter the threat, banks need to provide more competitive trade finance offerings which deliver what customers need at a price they are willing to pay. Technology can help – for example Citibank states that OCR could increase productivity by as much as 50% while also reducing compliance risk. When combined with machine learning the system could automatically transfer documents and feed data directly into issuance systems. Transaction costs could be dramatically reduced. By analyzing risks more quickly and more cost effectively, banks can offer more financing options to international businesses. HSBC has recently started using robotic technologies to analyze documents, digitize and extract the relevant information before feeding it into their transaction processing systems.
- Eliminate customer fears of foreign exchange trading
In 2016, five of the world’s largest banks paid fines of $5.7 billion in relation to charges of manipulating the foreign exchange market. This news fed into customer fears and led them to re-evaluate their existing relationships. CGI reports that 41% of corporate practitioners are reassessing their current FX relationships. AI-powered FX trading is expected to become main stream in the near future. Along with slashing costs, banks reduce the risk of price manipulation by minimizing human intervention. Recently, the US-based Fintech Investment Group unveiled its algorithmic forex trading platform – ART. The system uses 30 years of historical forex trading data to forecast future currency movements, automatically execute trades and eliminate many of the emotional or technical mistakes that forex traders make.
The future is uncertain
Despite the high profile campaigns – including Elon Musk’s recent warning about the dangers of AI, it is clear that AI’s advantages will result in ever greater deployment. Some observers warn of a future where humans are replaced by AI, in much the same way that robots have replaced many factory workers or ATMs have reduced the number of bank tellers. Other observers talk about how AI can be used to augment humans, to work with humans, making humans better – for example if an AI could be combined with human memory then people would never forget a name. If this latter vision comes to pass then rather than replacing bankers with AI-powered robots (robo-banker if you will), in fact what will happen is bankers will be augmented by AI (cyber-banker).
As is often repeated “nothing in this world can be said to be certain, except death and taxes”, but clearly change is also certain. The exact direction, timing and scope of that change are much less certain. The challenge is to anticipate change, prepare for it and when it happens – react quickly.