AI in automation: the benefits of integration



Katie Sadler
04/23/2018

How can we deploy artificial intelligence? Use it to enhance processes when automating says Volvo’s Mats Johannesson

automation
“The opportunities are there when you make up your mind to automate.” Photo by Shirly Niv Marton on Unsplash


“We didn’t begin with the most sophisticated and intelligent automations, we gradually moved on to that.” Working with JiffyRPA (Option3’s intelligent automation platform for RPA), Mats Johannesson, Volvo Technology Director of Accounting Shared Services, explains why it is imperative not to focus on the simple processes in order for automation to make a difference.

Johannesson believes employees play a key role in making automation successful. He has learned that “to make any automation successful you need to have strong commitment from line managers and operational teams if you don’t have that kind of commitment you will never get efficiency or benefit out of automation.

Speaking to Katie Sadler of AIIA Network, he advises, “don’t look for an opportunity to use artificial intelligence (AI), the opportunities are there when you make up your mind to automate. Go after that project and then optimize it using AI.”

KS: In what way is Volvo embracing and implementing intelligent technologies?  How is this likely to develop in the future?

MJ: First of all, we didn’t begin with the most sophisticated and intelligent automations, we gradually moved on to that.  Our principle is not to deploy intelligent technology for the sake of it.  We don’t go into a project and think of it as an AI project, we still see it as an automation project, or analytics, but then within this we utilize AI technologies.

Looking ahead, we will probably do two things: Firstly, augment or support the automations that we have already developed with machine learning to help make them more efficient and powerful. Secondly, use more advanced intelligent automation to detect deviations and potential errors and risks in our accounting.  I think that’s the two main areas we will develop.

KS: How has partnering with Option3 helped develop robotic process automation in Volvo’s accounting shared services?

MJ: We have worked with Option3’s JiffyRPA for about one-and-a-half years to develop robotic process (RPA) automation in different accounting processes.  We have developed robots of varying levels of specification going from purchase to pay to OTC (Order to Cash) and record to report.

JiffyRPA does a good job integrating AI and machine learning in our organization.  This is one of the strengths of their product platform. We use the JiffyRPA platform for automation and implement a second tool to build strong user interfaces. The platform includes a component which nicely integrates the machine learning too.

“To build a robot you need [employee] input and knowledge—without that we cannot build on automation at all”

KS: How receptive have employees been towards the changes that have been implemented?

MJ: Yes, employees have been involved in the project. To build a robot you need their input and knowledge—without that we cannot build on automation at all. They play a key role in making it successful. We have really learned that to make any automation successful you need to have strong commitment from line managers and operational teams if you don’t have that kind of commitment you will never get efficiency or benefit out of automation.

KS: What key lessons have you uncovered on your journey so far? Is there a challenge that you’re still looking to resolve?

MJ: We have been very ambitious this year and that has been quite tough.  We’ve been running many projects in parallel and have pushed quite a lot to develop quickly in this area. With that comes challenges.

Don’t take the infrastructure requirement too lightly is one lesson learned.  Make sure that you get your software and platform on the right infrastructure from the start.  I think especially in RPA there is a tendency to run it on lighter infrastructure, like desktops but you must have the correct infrastructure. As applications and robots become more business critical we must treat them as such.

“Make a proper analysis of the process before you start automation”

Secondly, make a proper analysis of the process before you start automation.  Make sure you’re clear about the process waste and question, how should the process be standardized?  How should it be optimized before you begin?  If you don’t ask these questions it will be a challenge to maintain, expensive and difficult to manage.

Finally, make sure you have invested the time in your specifications and reviews before you start development. 

KS: How did you decide on the scale of your RPA activities?

MJ: We began by implemented multiple RPA activities in parallel with one another.  Together with Option3, we selected and worked on two pilots and then continued quickly with more.  We selected the pilots from what were not necessarily the easiest and quickest processes to automate but rather the highest priority from a business point of view.

When you select a pilot you should ask yourself:  What is the big pain point here?  What can we do about it?  And how can we automate it?

Don’t just focus on the simplest processes, that’s key. Automation needs to make a difference. In our first pilot, we really cut the lead times doing some business-critical processes which was really appreciated by the business and it was hugely important.

KS: Do you have any advice for companies starting out on their automation journey?

MJ: Don’t look for an opportunity to use AI, the opportunities are there when you make up your mind to automate or do some analytics. Go after that project and then optimize it using AI—utilize it as an integrated part of automation and analytics rather than saying:  How can we use AI?  It needs to be driven from a business value point of view.

You may also be interested in...

“We need to change the automation model”

Cognitive RPA is essential for a continuous automation framework

Three myths that prevent organizations moving into cognitive automation

OUR BENEFACTORS