Artificial Intelligence and Robotic Process Automation in Finance, what is it and why should I care?

Steve Britton, Client Services Director– Global Accounts at CloudTrade

07.11.2018 06:45 am

I know what Artificial Intelligence (A.I) is and I see Robotic Process Automation (RPA) in the press all the time, I hear you say, but can these new technologies really make a difference to my business’s management of our finance processes?

Let’s explore the technology first, then look at the challenges in finance and accounting that A.I and RPA go some way to meeting.

The difference between A.I and RPA

A.I is a computer program that addresses a task that usually requires human intelligence (or at least that is the theory), such as visual perception, decision-making, completing complex, but repetitive tasks.

RPA uses AI to manage a complex set of decision-based tasks, such as reading data. Based on the required business process, RPA applies rules to ensure that the data meets the set requirements. If the data does not fit, RPA ‘discovers’ the required data to enable the task to be completed.

Examples of RPA would include, processing a mortgage application, an insurance claim, a supplier invoice, or a customer order. In these scenarios, data is generated, either by a human or a computer system, and must be processed by the receiving system.

Any data must be acquired, interpreted and the information it conveys checked, to ensure it is complete and correct. Identifying where there are gaps or errors, and correcting those errors, means the task can be completed.

The term ‘Robotic Process Automation’ (RPA) is derived from the computer programs the process uses. Those programs are known as ‘robots’, and they can process data and complete complex tasks automatically, providing there is a logical or a rules-based set of steps to follow.

The element of human intuition is still preserved by a human’s cognitive ability to combine reason and logic. It’s worth noting that we have not yet been able to evolve computer programs to be able to think for themselves. Well, at least not in the day to day world of finance processing, but I am confident science fiction will soon become science fact and we’ll see logical programs very soon.

How to navigate your way through all of the RPA programmes out there

The reality is that the massive demand, and the clear benefits that RPA technology can bring to data processing tasks, has given rise to many RPA type applications. As with all computer programs, some of those programs have been specifically designed to address a market vertical, like manufacturing, pharmaceutical, engineering, logistics, hospitality, or banking & finance. Others have been designed to be more ‘generic’ in nature.

However, it’s important to remember that an RPA or AI application isn’t likely to do very much ‘out of the box’. Any RPA application must be integrated and configured to address the operating and business environment of deployment. Therefore, the application provider and the implementation team need to be conversant with the industry and business process they will be addressing. This is often the first error businesses encounter when they are looking for an RPA solution. One size really does not fit all, and you need experts that understand the field they are looking to automate.

Before you start to review the AI / RPA platforms available to you, you need to consider whether the required data is available to support an automated process and whether you have the current baseline metrics to measure and evaluate its performance.

Remember, a robot is ‘dumb’ without data. That includes the quality of the data across the entire process lifecycle. You don’t want to automate 2/3 of a process and then find that the remaining tasks require human intervention and you end up with no cost or compliance benefits.

Baseline Data & Metrics

That brings us on to baseline metrics. You must measure the end to end process, touch points and resulting costs. If you don’t how will you accurately measure the outcomes and benefits?

So, why do we need automated processes in F&A? Well, a recent Hackett report stated, “Highly educated accountants and financial personnel spend 65%+ of their time on manual, low-value processes.”

This is clearly unsustainable and an imbalance that makes the business case for driving efficiencies and automating many business processes clear.

Choosing the right solution for you

The first question you need to ask yourself before you embark on any automation journey is exactly what business processes need changing and why? In my opinion, this should start with ensuring we are able to capture and record the baseline measurement of the process in question.

If we take the Purchase to Pay process as an example, this needs to start at the beginning of a process. The requisition, creation or at the very least the ‘decision to buy’. We need to know how this is communicated with the supplier. Not at the ‘Invoice Capture’ stage, as is so often the case.

Performance improvement claims made by RPA vendors and solution providers, must be translated into the quantifiable outcomes that will be delivered and form the basis of an ROI calculation.

‘Sales’ claims must be supported by relevant case studies, as this will ensure you are buying a product that is designed for the purpose and outcomes you need.

I wouldn’t buy a 4x4 vehicle if I was only driving on a motorway or use a pair of scissors to cut my lawn. Both tools can do the job but would certainly not give me a cost effective or efficient outcome.

The same applies to RPA integrators/solution providers, experience in the Insurance or mortgage industry will not provide the detailed knowledge required to deliver an effective finance and accounting solution.

Making sure you have the right partners

Once you have selected the right RPA vendor that has the relevant AI capability. You need to choose the right system integration partner to design, implement and support (don’t under estimate the amount of support that will be required) the solution, this must include the delivery of a tangible business case that is measurable.

The measurement of the ‘As Is, To Be & Run’ phases must be accurate and delivered in a way that can be easily consumed and understood. It must deliver transparency and the ability to govern the process, deliver performance metrics with the ability to interrogate data from a single UI to a granular level and be able to run what if scenarios that are actionable.

In short, the deployment of an RPA solution is a journey, not a destination, and your service provider must be prepared to work in partnership with you and your end to end process lifecycle if real benefits are to be achieved and ‘maintained’.