By Steve Cox, Group VP, Oracle Cloud Business Group
Finance departments have been racing around for decades to try to keep up with the increasing quantity and complexity of information. Technology—particularly ERP and EPM cloud—has helped them keep up. But the data and workloads keep growing.
APQC’s recent annual survey, “Where Does the Time Go in Finance?” shows that in spite of significant success reducing costs, transaction processing takes up almost half of a finance department’s time. This could be holding finance teams and their leaders back from taking on a more strategic role in emerging digital business models.
As APQC put it: “This means that in an average work week, highly paid finance staff are spending the equivalent of Monday morning through lunchtime on Wednesday making sure that bills get paid, customers get accurate invoices, general accounting work gets done and fixed assets are accounted for, among many other tasks that keep the money moving through an organization.”
This leaves precious little time to analyze, for instance, the cost-structure impact of one strategy over another, or the revenue and operating margin implications of investment decisions.
Now, financial professionals have a new weapon: Adaptive and artificial intelligence, bolstered by machine learning (also known as AI/ML).
The sci-fi sounding technologies are moving into business systems and promise to solve many of the time crunch issues that finance professionals face—namely the quantity, complexity, and accessibility problems that make transactions so time-consuming.
Let’s take a look at how AI/ML solves those problems.
Quantity: By some estimates, data volume is growing between 2X and 50X per year, depending on organizational investment in the Internet of Things (IoT) and Big Data. These technologies are the wave of the future, so it’s fair to say the volume of data will grow at an accelerated rate as usage of IoT and Big Data accelerates. Addressing this deluge of data so that meaningful insights be drawn from it in a cost-effective way will require automation.
As machine automation did in agriculture and manufacturing, AI/ML promises to eliminate the manual, task-based, non-value-added work required of finance professionals.
For example, the UK’s National Health Service, which serves the health care needs of the country’s entire population, uses predictive analytics to help identify fraudulent claims. Such an effort requires review of a substantial amount of data. Previously, a clerk had to review claims on a computer, screen-by-screen and field-by-field. Filtering reduced some of the work, but it was labor-intensive.
Now, potential fraud is identified through a series of well-known criteria—not just an effective filter, but historical data and patterns—which are continually updated via machine learning. The clerk’s involvement starts with the claims identified by the system. The manual work of identifying the claims is eliminated.
Complexity: As it grows in volume, financial data becomes increasingly complex for several reasons. The proliferation of market channels, payment methods and product configurations adds to the variation in recording transactions. ERP systems must be updated to accept information from an ever-wider array of data sources with unique protocols, which is time-consuming and expensive. With AI/ML, systems could adapt themselves accordingly.
Accessibility: AI/ML confronts the data-accessibility problem in two ways: by making it easier to find and use information in the system, and by making the information much more accessible to a broader set of employees. As AI/ML capabilities expand into more and more use cases, the software can recommend actions based on patterns and trends using intelligent chatbots, or “talkbots.”
One consistent refrain among finance staff is that they know the data is in the system, but it’s hard to reach. The use of chatbot (“bot”) technology allows professionals to use natural language, rather than cumbersome search tools, to find the information they need. For example, as the NHS clerks review claims, they can ask the system, in natural language, “Can you show me other similar items?” This allows for continually more creative types of inquiry to get more and more pertinent information.
Bots also replace the “tribal knowledge” within finance departments. Currently, financial information is made understandable for non-finance professionals only when someone translates the information. With bots, the company’s collective knowledge is gathered and made available to a wider group of people.
While this may seem a little futuristic to some, most people already are using this type of technology through consumer products like the popular Alexa, Echo, and Siri on their smartphones, which are voice assistants that use contextual intelligence.
In the business world, Oracle is using its Data as a Service (DaaS) Cloud to embed adaptive intelligence into its cloud applications for finance (and other functions). Dubbed Adaptive Intelligence Discounts, the system combines Oracle’s data pool (using Oracle DaaS) with your company’s data and, using algorithms, identifies which suppliers are likely to take advantage of discounts in return for early payment, and when. To achieve this without AI would require, at minimum, a full-time person to analyze supplier behavior.
Two concerns tend to arise when companies consider AI: security, and the potential need for specialized training.
Regarding security, AI/ML reduces what is one of the biggest vulnerabilities in managing corporate data: human involvement. Traditionally, when a threat is detected, the vendor creates a patch and publishes it, then pushes the patch to a company employee or third-party contractor to apply. This process can take several days, and some companies have been known to let months go by before applying a patch—long after it becomes available.
With AI/ML, there’s very little “vulnerability gap” between a threat and a fix. As threats are detected, patches are automatically created and applied systemwide.
Another way AI improves security is by reducing the amount of data that is exposed to humans; AI delivers more targeted information. At the NHS, for example, the clerk doesn’t need to have access to and review all the data. Rather, he gets only the result of a query, which exposes a smaller set of the data.
As for training your staff how to use AI/ML, this varies with the technology provider. With Oracle Adaptive Intelligent Apps, for example, AI is embedded in the ERP and EPM cloud systems that business users already know and use every day. The finance team requires no extensive training.
Ultimately, AI/ML will help finance professionals to effectively process the fast-growing volume of complex transactions. It will free them from mundane, manual transaction tasks; assist them in decision-making; and allow them to contribute their human intelligence, creativity, and business knowledge to help solve business problems and chart the best business strategy.