One day, the head of Barona´s digital services Perttu Mäkimartti, was sitting at his desk and growing frustrated. He was trying to complete what seemed like an endless list of invoices to be posted.
Is there anything more useless than making skilled employee spend their time on routine tasks?
Who among us is not slightly frustrated when we receive a notification that we have new manual tasks on our list? And with good reason – we are used to today´s technology and services and so are loath to return to yesterday´s routines. Surely each of us thinks from time to time that there must be a better way of doing this.
Mäkimartti left his invoicing work for a meeting with an expert talent who had helped build Barona´s artificial intelligence recruiting solution, Jelpp.
It felt like I had moved from one century to the next within the span of a day. Perttu Mäkimartti
Barona was no stranger to artificial intelligence and the opportunities it provided. The Jelpp recruiting system featured artificial intelligence elements that had been in development for awhile, therefore the possibilities and limitations of AI were familiar. The algorithm existed in principle, but so far there had just been no discussion of its applications outside recruitment.
The meeting inspired Mäkimartti. How about taking advantage of this existing concept to develop a solution for automating invoicing processes? We already know how to take advantage of artificial intelligence in a complex environment with a lot of unstructured data.
Well, the basic coder has a mindset where they would rather spend a month thinking about how to avoid some routine work rather than performing it an hour at a time from one year to the next. Perttu Mäkimartti
At a back corner conference room the thought turned into a plan. Barona Group´s Chief Financial Officer Pasi Vilhunen, Futurice´s Antti Rauhala and Barona Technologies´ Perttu Mäkimartti decided to develop intelligent financial administration together.
We thought it was justified by the simple fact that many of our clients were wrestling with these same challenges. If we can first solve this in our own processes, we can then offer the solution to others. Perttu Mäkimartti
This started the building and experimenting. Other top experts joined the team. Since the algorithm and the understanding of how to apply artificial intelligence to this type of case already existed, the first phase took only a month. Now the solution could be tested in practice.
Well, the 80 percent success rate was heartening. It immediately raised the question whether the accuracy was already superior to manual work. Pasi Vilhunen
The trial was a success. During the first round, it could already be shown that the artificial intelligence learned how to post invoices and the correct approval flow. This is of course not enough. We know that there is much more.
The results of the trial were very encouraging. However we knew that they could be even better. Artificial intelligence is able to recognize the data on the invoice on an even more minute level, and this of course improves accuracy. Perttu Mäkimartti
Where do we stand now? Barona started implementing the solution in our own processes in September 2016. The big vision is maximizing automatization. Legislation poses more of a limit to what can be done than the capabilities of artificial intelligence do. We will automatize the process as far as the legislation allows. This sounds almost too good! We asked project leader Perttu Mäkimartti a question with societal implications: Does improving productivity lead to unemployment?
The people who work for us are all top-of-the-line financial professionals. Instead of using their time on routine tasks this allows them to focus on something with more impact. As a growing multi-industry business, we constantly have many different ideas in development. Who knows, maybe at some point in the future these same people will be consulting other companies on how to improve their financial processes. Perttu Mäkimartti