RPA or Robotic Process Automation refers to the element within Intelligent Process Automation that is centred around automating repetitive tasks.
By using specific software tools such as those made by UiPath, specialists such as the development team here at Evros, are able to help businesses not only streamline their processes but also get a significant amount of time back through automation.
In this feature, we speak to Head of Application Development, Trevor Dagg, about how he’s been implementing RPA to benefit our Finance and Accounting Department.
What did you originally set out to do?
“A lot of what we do here at Evros is about helping businesses work in the best way possible through technology. And it’s really important to us that we have all the right tools, skills and experience to enable this for our customers.
And certainly, as the Software Development Team, we’re already experienced in building software to meet the challenges that businesses are facing today.
So, getting involved with RPA, Business Process Management and AI, was already complementing the focus we have as a team.
Over the last few years, we have been identifying software partners who have the right level of expertise to advance our offering. And what we now have is Dell Boomi who are working with us on the integration side; UiPath for automation; Bizagi for Business Process Management, and our strong relationship with Microsoft means we already have a well-developed offering around AI and Cognitive Services.
When you bring all of those together, what you have is an extremely sophisticated suite of tools that are empowering businesses and supporting them in their digital transformation in new and highly effective ways.”
Why did you choose to work with the Finance Department in particular?
“We wanted to start implementing these tools within our own business so that we could benefit from them but also it was a chance to work innovatively, so that we could also discover the best ways to bring them all together.
We looked at a number of initiatives and ideas across the business including patch management, server ticket management, account activation and deactivation, and certain aspects of security.
However, it was clear that we would have the biggest and most immediate impact on the Finance Department, purely because of the number of repetitive, manual, low-exception rate tasks that they perform.”
How did you go about discovering what tasks could be automated?
“Essentially, the starting point is to ask the business, ‘where are your biggest challenges?’ because this is how you’ll discover the issues that are absorbing the most time and energy as well as creating the biggest amount of risk and stopping you from being able to spend your time on driving greater value.
The next step is to work your way down and find out where the problems actually start and map out the individual steps within each process so you can see in detail what’s actually going on.
One of the biggest advantages that we have here at Evros is that we can be technology agnostic. So, once we have identified the issues, we are able to choose the right tools to address the issue(s) rather than being restricted in what we can use.”
How do you examine these challenges from a perspective?
“When you look at these sorts of challenges, especially for some of the larger organisations, there are a lot more moving parts that need to be taken into consideration around the strategic direction of a business.
So, we also need to look at their objectives, for example, are they consolidating, are they growing or merging – all of these factors need to be considered.
There are a number of things that feed into that and you have to be able to plan the implementation holistically, so that it is aligned with the organisation’s overall goals. And that takes experience, a particular skillset and also the right set of tools to do the job.”
What were the initial steps involved?
“Firstly, we sat down and looked at the processes that they were spending a lot of time on. And we also gave them a high-level brief of the type of processes we were looking for.
So, essentially, these are processes that are: manual, repetitive, rule-based with a low exception rate and had some kind of electronic standard or readable input that we could source the information from – although, that being said, the technology is moving at an incredible rate and it won’t be long before we can start looking at converting handwritten documents as well.”
What was the solution?
“The result was, we discovered the team was spending 20 hours a week manually processing up to 1,600 to 1,800 purchase invoices a month.
So, what we did was:
- Map the business processes to an automated approach
- Using the invoice templates from our key suppliers, we built and trained the data models
- Following that, we set up and configured UiPath Document Understanding along with UiPath’s AI Fabric feature
- The next step was to design and implement the integrations with Microsoft Outlook and Dynamics NAV (Evros financial system)
- We then continued to improve the process through Machine Learning (ML)
- The solution was configured to rank each scanned invoice to have a confidence score of 80% or more to continue its Happy Path for approval
- The documents that didn’t achieve 80% would be passed back to the Finance Team for update and correction using the solution. This would allow the data model to ‘learn’ and improve future confidence scores and increase number of successful automations
- And we also set up alerts and individual reports for each automation as well as a central report to highlight the overall performance of the solution using a Power BI dashboard
We also had various considerations to make including:
- Each individual customer/client has their individual invoice template
- There were other document types mixed into the start of the process that would need to be filtered out (e.g. credit notes etc.)
- Large invoices could potentially be invoiced in stages or as good were delivered. These were indicated as ‘part invoicing’
- Each invoice was printed, reviewed and stored for audit purposes as part of the manual process
- We also needed to consider a new storage approach”
What technologies did you use to perform these tasks?
“We used a few tools to implement this solution:
UiPath Automation Cloud
UiPath offerings including:
Other tools we used:
Outlook, Microsoft Dynamics NAV.
We actually started off with Microsoft Form Recogniser but then moved onto one of UiPath’s latest offerings, Document Understanding with the AI Fabric feature, so we’re experienced in all of these technologies.”
What was the overall impact of the solution?
“We started with a success rate of around 60% which rose to above 80% following improvements made to the AI over a 6 to 8-week period.
That translated into an 80% time saving for that task (16 hours a week) that’s getting better over time.
In addition, the solution we’ve implemented allows us to scale and doesn’t require any more resources from the Finance Team.”
Is there a risk of bots losing money through automation?
“Like with any important process, you have to build in your checks and balances to ensure the bot is performing as it should. That’s why, for example, we’ve created a separate database with Power BI front end, to track the performance of the bots. This means things like paying an invoice twice just won’t happen as Microsoft Nav won’t allow you to do that, it will just fail.
We also perform a monthly review of all POs at the end of each month anyway so we can spot if there are any issues captured.
In relation to paying salaries I would suggest the appropriate oversight and governance is applied to the process. So if it’s critical you have a human approval on certain aspects, you can build that into your process at the point where it makes sense for the governance to be applied.
The bots only do what they told – they’re not going to go off and change the figures randomly. And that’s one of the strengths of bots – they do exactly what they’re told and they can do it over and over again, 24×7, while eliminating what we currently have as a risk which is human error.”
Do these types of solutions work with legacy systems?
“What’s particularly useful about this method of working is that the software is simply mimicking the user. So, we don’t need to go through the painstaking process of trying to get different systems to talk to each other. And this is incredibly useful for any business relying on legacy systems for essential processes as you can cut down on the manual labour without the daunting level of time and expense needed to upgrade or move to a different system altogether.”
Find out more
Read more about Evros Intelligent Automation Services in our feature or get in touch to speak to our experts.