It is undeniable that the many recent innovations in consumer payments have been driving the future of business payments, including big data, Open Banking, and a focus on simplifying the customer experience though the much-hyped Artificial Intelligence. In this post, I’d like to have a look at the wider impact of artificial intelligence in payments, business payments specifically.
As both terms are often used interchangeably for marketing purposes, let’s first of all talk about the differences between machine learning and artificial intelligence:
Machine learning is essentially a branch of artificial intelligence where computer algorithms examining and comparing large data sets to find patterns and derive insights.
Artificial Intelligence, on the other hand, has a much broader definition as “the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence.”
The following diagram is helpful in adding some context:
Simply put, unlike machine learning, which is very well defined, AI is more difficult to pinpoint as it is closely related to advances in technology, making it a moving target. There is much debate about what is and is not AI: several decades ago, an electronic calculator was considered a form of AI. Today, AI is symbolised by gadgets like Google Home, Siri and Alexa, by the machine learning powered algorithms behind Netflix, Amazon and YouTube, or by hedge fund chatbots bringing trading to the masses. In future, it is possible that these even these developments will seem as old-fashioned as the electronic calculator is to us today. One thing is certain however: technology pervades our daily business and personal lives. For the purpose of this document, we will use the broader term of “Artificial Intelligence”, on the understanding that this comes in many forms.
Broadly speaking, artificial intelligence can be applied to the following areas:
- Operational efficiency
- Growth and innovation
- Risk Management
In our complex regulatory, technological and socio-economic landscape, improving operational efficiency is perhaps one of the first areas where artificial intelligence has delivered substantial benefits. The ability to automate repetitive, rule-based labour-intensive tasks can improve the quality of outputs and free up individuals, enabling them to concentrate on the more complex parts of a given process or other value-adding activities. Indeed, Robotic Process Automation (RPA) is forecast to be worth at least $4.3 billion by 2022. For example, RPA can be used to derive treasury and accounting efficiencies through data management to improve reconciliation, general cash cycle and cash flow management effectiveness. In fact, the automation of accounts payable (e.g. invoice processing) has now become a common use-case:
From the diagram above, we can see that RPA is non-invasive and requires minimum integration with the existing systems infrastructure. This in turn delivers productivity improvement by replacing human effort to improve labour-intensive activities, where people are performing high-volume, highly transactional process functions. Other applications outside of accounting and finance have been deployed in payroll and benefits (e.g. personnel data correction, tax adjustments, etc.), application processing (e.g. data entry, integrity checking, etc.), and sales related clerical work (e.g. data entry, name identification, etc.).
There are many other applications of AI to improve operational efficiency, such as those used to improve procurement processes. One example is that of contract management, where AI is used to extract data and clarify the content of contracts. This can speed up renewals and renegotiation terms for large numbers of contracts, decreasing potential disputes and increasing throughput. In more generic terms, AI can be used for process improvement.
But using AI is not just about operational efficiency and cost reduction. It can be used to help a company innovate and grow. As previously discussed, the innovations seen in consumer payments have been driving those in business payments, and if we look at the current landscape, it will become apparent that small-medium enterprises (SMEs) present a tremendous area of growth in business banking. It is therefore unsurprising that young companies that originated in the consumer space, and therefore offered a very streamlined customer experience at the outset, such as Tide and Azimo (with their roots firmly in AI) are now turning their attention to business banking.
That being the case, traditional commercial banks can no longer ignore the customer experience as technology and demographics are evolving: those entering the workforce are now true digital natives and their expectations are different. For example, United Overseas Bank (through a fintech partnership) now applies an AI-based credit assessment tool to transactional data to speed up loan approvals for SMEs.
Today’s enterprise customer expects not only to be able to track what they have spent historically, but also want to predict what they will spend in future, and manage their finances accordingly. They also want their experience to be personalised to their needs.
AI will ultimately enable businesses to move from hindsight to foresight.
However, enhancing the customer experience is not the only thing that AI can help with. Sales forecasting can be enhanced with AI-derived triggers (e.g. M&A, headcount changes, etc.) enabling businesses to determine whether customers are likely to leave so appropriate retention strategies can be deployed. AI can enhance CRM systems by profiling a customer or prospect to suggest talking points. AI can also be used to monitor staffing trends and compare them against departmental goals, objectives and sales trends in order to deliver an optimum staffing plan.
In a previous post, I examined how machine learning and automation can help organisations fight fraud and cybercrime. Indeed, artificial intelligence, in its many forms, can be used very effectively to manage risk. Going back to the previous example of Azimo, among its AI deployments is an algorithm that can scan documents to spot whether they have been edited or whether bank details have been copied and pasted into its service, thus indicating potential fraud. In addition, with the increasing regulatory pressure brought by the General Data Protection Regulation (GDPR), the 2nd Payment Services Directive (PSD2) and the many anti-money laundering regulations worldwide, the need to automate compliance activities has never been more pressing. AI can be a very effective way to address regulatory risk, if the current market forecast for Regtech solutions is anything to go by.
Indeed, artificial intelligence is predicted to offer new efficiencies in compliance, driving the Regtech market growth. For example, regulatory intelligence can automate the arduous task of analysing and interpreting regulatory change, providing businesses not only with the insights needed to make better decisions to support regulatory adherence, but also to minimise non-compliance risk. And whilst large enterprises are expected to hold the largest market share during the forecast period, those businesses able to offer solutions to enable SMEs to derive similar benefits could potentially capture a very lucrative market.
In conclusion, the opportunities for artificial intelligence are numerous. But the challenges must not be underestimated. Data sources must be reliable in order to derive actionable insights, and they must also be compliant with existing regulations. Organisations must also have a thorough understanding of their business processes. To draw a parallel with the Business Process Outsourcing craze of old, the successful businesses – i.e. those that derived efficiencies and savings – were those who outsourced processed they had optimised to the best they could. After all, outsourcing an inefficient process only results in an inefficient outsourced process. And what is using artificial intelligence but outsourcing a process to a machine? That being said, the deployment of artificial intelligence in any organisation of any size goes hand-in hand with digital transformation and cannot be treated as an isolated technology deployment. The best strategies will be those that encompass the people, the processes, and the technologies used. Deploying a layered approach, where automation is used where the processes lend themselves to it, will free staff to concentrate on more complex tasks and value-adding activities. That is true digital transformation.