Automating standard processes has long been the top priority for finance departments. Now, things are moving to the next level: Thanks to machine learning, intelligent software can now handle tasks that it has never been able to perform before.
Most enterprises deploy software that posts payment transactions automatically and ensures that compliance rules are met. Which means that finance personnel are only called on to intervene manually when exceptions occur – such as customers omitting key payment data, making a typing error, or paying multiple invoices in a single transaction.
“This is precisely where machine learning comes in,” says Robin Bau, director of Shared Service Automation at SAP. “The new technology can now be deployed in any scenario where additional knowledge is required.”
Machine learning-enabled software learns in a similar way to humans, that is, chiefly through experience, observation, and historic data. It uses self-learning algorithms to spot patterns; it recognizes contexts, and it makes predictions. And while the concept of artificial intelligence is by no means new, it’s only now that computers have become powerful enough to analyze sufficient volumes of data and to allow data scientists to develop corresponding models from it.
Using machine learning to identify universal processes
Of course, theoretically, you could create a set of rules for every conceivable error by hand. But self-learning software saves you the trouble. “The technology runs in the background and ‘observes’ human actions. When an employee in finance allocates a consolidated payment to multiple invoices, the software remembers the action and performs it autonomously the next time around – without being explicitly programmed to do so,” explains Bau.
After a brief “familiarization” phase, the software understands comparatively simple, universal processes. It is also quick to learn company-specific compliance procedures: “Each company has its own rules and business scenarios. All the software needs is access to historic in order to be able to respond correctly,” says Bau.
Built-in AI for SAP solutions
In the medium term, SAP plans to build artificial intelligence into its entire software suite and all of its cloud solutions. One key objective is to ease the workload of shared service organizations.
“Many of the inquiries processed by shared service centers are fairly similar: When was my invoice paid? Why did I receive a reminder? Answering them is a very time-consuming process. [Software] identifies the correct processor when a ticket arrives and can often answer questions itself,” says Bau.
Going forward, artificial intelligence will be used for strategic matters, to provide accurate forecasts, and thus to support a company’s growth.