SPIE works more efficiently thanks to combination of RPA and AI
Working smarter with RPA
Facilitating growth in a tight labor market
120,000+ reports per year
At SPIE Netherlands, software robots installed by Tacstone in two different departments are making work more efficient. The Finance Department can do more with the same people, in a tight labor market, thanks to Robotic Process Automation. At the Service Desk, RPA combined with AI is speeding up the 270,000 service calls per year.
SPIE is an international service provider whose activities include managing and maintaining buildings, and helping clients with their energy transition and digital transformation. It does so in forty countries with a total of 48,000 employees. SPIE Netherlands ‘ Building Solutions division also deploys smart automation in its own operations.
“Technology plays an important role in our core processes,” says divisional controller Jan Heinen. “First, in our digitization agenda. Because buildings obviously provide a lot of data. We want to be able to interpret that data, partly to achieve predictive maintenance, where you can predict, based on data from the systems, when failures are going to occur.”
Working smarter with RPA
In addition, internal processes are being digitized. “Those are the basic processes you need to keep the business running, like finance processes. It’s getting harder and harder to find people, so if we want to grow, we have to do more with the same people. This means working smarter. For that, among other things, we use RPA.”
He himself was already familiar with the benefits of RPA. The question he asked the Finance team was “What are the things you do that irritate you that could be smarter?
“We haven’t looked at where the biggest savings can be made, because it’s about getting people excited about the possibilities of digitization, and facilitating growth in a tight labor market.”
Jan Heinen | Division Controller SPIE
Time saved by robot
The software robot was built in collaboration with Tacstone Technology, which at the time was already working on a software robot for the Service Desk. “I knew they were doing a good job. They thought along in framing the process to come up with a script. We currently have four basic processes running on RPA. We do this with one robot, which of course you can fill up with processes 24/7. For now, everything fits within this one software robot, which can be used variably.”
An example of such a process is the review of projects. ‘Standard forms are prepared for these reviews that run on SAP. This process involves opening Excel, starting and saving a macro, which takes about ten minutes, and then it must be moved to a folder. This costs people a few hours a week. Now a software robot does this.”
270,000 reports per year
However, the most advanced software robot does not run in finance, but in SPIE Building Solutions’ Service Desk. Here, RPA is combined with AI. “We are responsible for processing all customer queries towards SPIE 24/7,” says Pieter Quanjel, Manager Servicedesk. “That’s 270,000 notifications a year, spread across various channels such as building management systems that generate automatic notifications, traditional telephony and emails.”
Team leader Ilse Dankers adds: “We work with a team of about 30 students who process these reports. Students are deployed based on the volume of tickets, with the peak number of reports being significantly higher during the day than during the night hours. Our staff has no technical background, so they have to work with the information shown in our application. Therein lies immediately the strength of a software robot: it can assist in creating a report in the best possible quality.”
Figure 1: SPIE uses UiPath Document Understanding to process semi-structured service calls in an automated way
Software robot seeks information and determines urgency
RPA was initially deployed to process some of the large volume of incoming emails, 120,000 per year, more efficiently. Quanjel: “Many of those emails are standardized. This gave us the idea to explore what we could do with this. We wanted to extract basic information and automatically determine the urgency of incoming emails so that low urgency emails could be picked up later. This need led to a search that eventually brought us to RPA.”
“We clearly identified our functional requirements: what applications do we use and what information do we want to extract from customer emails and tickets? From a vendor selection with several parties, Tacstone was then selected as the supplier. This was a carefully considered choice after evaluating several options on the market.”
Implementing AI
In phase one, SPIE and Tacstone programmed the software robot to extract information from the emails. However, that process proved time-consuming and complex.
“We wanted more control over the process, so we started looking at what else was possible in our software environment with AI, in the form of intelligent document understanding.”
Pieter Quanjel| Manager Servicedesk SPIE
A major change was the move from working in the front end of the application, where the software robot literally performed click actions, to communicating through the backend with data links. This improved not only the speed but also the stability of the process.
Figure 2: SPIE uses a Machine Learning model (AI) to make the robots smarter
Simple for users
Dankers: “Technically, this may be complex on the back end, but for us as users it is actually very simple and intuitive. Someone can quickly learn how it works. That is a big advantage of this solution. And we no longer need Tacstone’s help at every step; we can manage and adjust it ourselves.”
For employees, not much changes in their work. Dankers: “They only see that a ticket has been picked up by the software robot. It makes their job easier because many fields are already filled in, so they don’t have to do everything themselves. They only need to check that the information is correct. They continue to check the information and determine urgency, then quickly forward the report. This leads to less time spent looking up information and administrative tasks.”
Figure 3: SPIE employees can train the robot themselves and learn to handle new document types and service report layouts
“The goal is not to replace people, but to make their work more efficient.”
Ilse Dankers | Team leader SPIE
Time saved by robot
Quanjel: “With the new solution, we can also easily add a new customer and train the system. By adding more and more customer information, the system becomes smarter and should eventually even be able to extract the necessary information from emails and perform actions on its own.”
The time savings are substantial: “Processing a regular ticket or e-mail ticket used to take between three and four minutes. With the robotic solution, that is reduced to just 20-30 seconds to check the ticket and add some information if necessary. This is a huge time saver. Moreover, this solution also improves quality.”
Smooth cooperation
About the cooperation with Tacstone, SPIE people are very satisfied. Quanjel: “In making the robot, we worked with a passionate programmers, who gave good advice and were really involved. They are always accessible, and if you have questions a response comes immediately. They think along and want to take steps forward, sometimes even a little faster than us.”
