Here’s why AI can make finance departments even more efficient

Artificial intelligence can streamline and simplify the work of finance departments, and it is neither scary nor difficult. To understand what is so special about this particular technology, we simply need to accompany a six-year-old to school.

Films, TV shows and books often portray artificial intelligence as something futuristic, incomprehensible and perhaps a little scary.

Although some people may find it scary, there’s no getting away from the fact that it will affect the way we work in the years to come at least as much as computers did in the 1990s, the internet did in the 2000s and smartphones did in the 2010s. 

This is because artificial intelligence already exists, and it’s a phenomenon that’s not that difficult to understand. In fact, it can be argued that there are some people who are better equipped to understand it than the managers and employees who work in accounting and finance departments.

The challenge of robots

Ever since computers became commonplace, the mindset has been the same regarding how they help us in our work: someone creates a program for a specific task – spreadsheet, word processing, financial system or the like – and we use that program to perform that task. Occasionally, an update may give the software some new functionality or make something simpler. However, the input has always determined the output. 

This still applies to much of what we use computers for, and we’ve grown accustomed to working on them. If you’re quite an advanced Excel user, you may have created an Excel macro to automate a task you perform often – and felt the wonderful feeling of watching your computer spend seconds on a job that used to take you minutes or hours. Of course, this can be done in other ways than in Excel – many finance functions already use similar functionality for tasks such as invoice processing using “Robotic Process Automation” or RPA.

However in Excel, it doesn’t take much for a macro to stop working properly, and the same is true for RPA: A number is incorrectly formatted, a value is entered in the wrong row or someone has moved a column. That’s all it takes for the macro to collapse and need adjusting before it can do its job again (until something similar happens).

Learning instead of instructions

This is where artificial intelligence, or AI, incorporates more functionality and requires a new dimension in our understanding of how computers work.

So how does AI differ from RPA? And why can artificial intelligence provide your finance department with so many more opportunities than these overgrown macros?

Imagine we have a robot and its job is to accompany six-year-olds to school, bringing them safely through the city. If the robot uses RPA, it must be programmed with the exact route from start to finish: 350 metres along this street, left on another street for 150 metres, and so on. But if suddenly one day there are roadworks along the route, the robot could get stuck – unless a wide range of conditions and alternative routes have been programmed in advance.
If, on the other hand, we use artificial intelligence, we don’t have to program every bend, street and intersection on the way to the school. Then we just let the AI accompany us on the journey to school a couple of times and perhaps use slightly different routes. It will see where we start and stop, keep track of landmarks and notice details along the way. In brief, it learns from history and acquires the ability to find its way to school that way.

The benefits of artificial intelligence

The difference is that, while RPA must be programmed using a precise “instruction manual” for each step in a process, AI learns where it is going (what the end result should be) and thus understands how to get there. This means that it is much more adaptable to any deviations along the way.

This kind of adaptability is necessary for a finance department that receives and processes invoices every single day. The invoices are from a wide range of suppliers, concern many different products that have often been sent to various departments or should be charged to different projects. VAT rates will vary, as will payment deadlines, cash discounts and a million other things.

If it is at all practicable to create rules that are detailed enough for an RPA to handle such complexity, it is not very viable in the long run. It would take virtually an infinite amount of time to create such rules, and at least as much time to maintain them, since parameters can change at any time: paragraphs are amended, suppliers are replaced, invoice setups are changed, projects are finalised or started.

Even the most minor change can be a potential source of error, and where an RPA-based system can make a mistake a thousand times without learning anything, a solution based on artificial intelligence will be able to extract line data even if the invoice suddenly looks a little different.

This is what makes advanced artificial intelligence a much more powerful tool. Such a system will have the ability to see multiple contexts early on, as well as learn along the way. Thus, not only will it extract data from the invoices with greater accuracy much sooner, it will also learn from its mistakes – and keep improving. This means it will provide ever greater value and quality through the inevitable changes of everyday life, while paving the way for automation that frees up a lot of time for the finance department.

All of this, in turn, means that if you’re serious about using technology to streamline the workflow in your finance department, you shouldn’t adopt an inflexible system that will require a lot of maintenance to deliver on its tasks, and you should be able to look past the simplest tasks you need help with. You should therefore choose a system that can also help the entire company with more advanced issues as the department and the organisation evolve.