Just how low can the RPA ERP robots go?

If we were allowed to create new insertions to conjoin acronyms with acronyms, then our first mission would be to add ERP to RPA. Enterprise resource planning (ERP) augmented under (AU) robotic process automation (RPA) is of course ERP-AU-RPA, or ERPAURPA (pronounced er-paw-pa).

Sadly, this concept is almost certainly not going to stick despite the well-aligned fit for RPA application and implementation in ERP. But what are RPA software functions at the core, how do they work well (and not so well) with ERP systems… and how far should we push the engineering envelope in terms of their usage?

Somewhere on the spectrum

Although the notion of software robots (bots) and automation functions has been around since pre-millennial times, RPA as a defined sub-discipline of the IT industry has flourished in the latter part of the last decade. The challenge for many organisations approaching RPA is that they don’t quite know how to classify it, let alone know where to start applying it. Somewhere on the artificial intelligence (AI) and machine learning (ML) spectrum, RPA still (arguably) suffers from association with ineffective chatbots and a wider distrust of ‘robots’ in general.

Regardless of how we classify it, RPA has the potential to reinvent work processes and digital workflows inside organisations at almost every level. Key among the practices it can be used to target for new efficiencies is ERP, due to the high volume of repeatable and definable transactions that pass through a typical ERP system.

At its most basic level, RPA is just ‘screen scraping’ to copy and track what happens when a user interacts with an application. That software application action might be filling out details in a forms-based app, it could be a database entry action or some other repeatable but essentially definable task. In the past, this has been known as ‘instant record-and-deploy automation’ by some parts of the software industry. We’re now moving considerably past that point as RPA starts to get more intelligent.

Process mining opens up ERP to RPA

To make ERP RPA work in practical real-world deployment scenarios, business managers need to work with data architects and software developers to break down workflows and look at the actions that occur inside them. This occurs through a practice typically known as ‘process mining’. The constituent elements of the ERP system may have already been clearly defined by the ERP vendor, but they are less likely to show how they ‘offered up’ to automation through RPA.

What we’re looking to do here is re-engineering the user interface (or at least parts of it) that a human would have been interacting with and re-purposing those RPA-ready tasks as a machine interface. Software robots can perform a wide variety of ERP-centric tasks from business team and departmental analytics – and onward to industrial equipment and machine performance analytics. But this technology needs to robot-to-robot, so there is a certain re-engineering responsibility to shoulder at the outset.

While RPA can play a useful role in orchestrating multiple systems, when it comes to interacting with an ERP deployment, it’s always on the outside looking in.

ERP RPA kill zone

So what parts of the ERP ecosystem do we apply RPA to? Technology analysts like to talk about the RPA kill zone, swim zone and free zone. The RPA kill zone includes basic ERP-centric data entry tasks that can be easily automated once systems are deconstructed and interconnected. RPA is good at shouldering repetitive, non-intuitive, process-driven tasks that are characterised by a definable pattern that can be digitally-tracked. These are the ‘low hanging fruit’ for ERP RPA.

The ERP RPA swim zone will be the tasks that still require some human interaction, but where humans will be increasingly replaced. Risk and insurance roles are a good example of professional tasks that appear largely skilled but can still be shouldered by software as they predominantly involve data crunching and analysis – a job the bots love.

“While RPA can play a useful role in orchestrating multiple systems, when it comes to interacting with an ERP deployment, it’s always on the outside looking in. ERP systems are in a better position to do the process mining they need in their own logic engines. They are also in the best position to gain benefits from contextual data using techniques such as machine learning and to properly combine domain knowledge with new technologies in an out-of-the-box manner – this is really intelligent process automation or hyperautomation, rather than RPA,” said Bob De Caux, VP for AI & RPA at IFS.

ERP systems are in a better position to do the process mining they need in their own logic engines.

Bob de Caux, IFS

“That being said, there is a space for RPA and ERP to work together,” concedes IFS’ De Caux. But, he cautions, it is incumbent on the ERP system itself to offer up ‘intelligent processes’ through an interface for easy consumption by an RPA framework. “This interface is the key, both for where governance should be considered as well as for where new ERP roles could be created related to optimising delivery. However, many customers will want this interface to be via the robustness of API rather than through a GUI, so RPA will not be the only game in town and will need to adapt to face off against system integration platforms,” he added.

A natural fit, intelligent automation

Prince Kohli, CTO at Automation Anywhere is extremely upbeat on the confluence of RPA and ERP and asserts that these two technologies are ‘a natural fit’. He makes this claim on the basis of a bot’s ability to handle everything from user logins, to data movement tasks, copy-paste commands and correcting missing data details. Kohli points out that ERP RPA is ‘most potent’ when applied to processes where efficiency, speed, or scale are the most important metrics, or where human errors are too costly.

Automation Anywhere has worked with a large retail customer that uses bots to automate its journal reconciliation, with the bot performing data validation to deliver the highest levels of accuracy. Other RPA bots deployed by this customer speed up order and invoice processing – and many apply intelligent automation to employee onboarding and administration to free up employee time for higher-value work.

“Data entry, movement and rules-based manipulation are essential but notoriously manual and error-prone processes that many organisations accept as a cost of doing business. RPA software bots can perform the same functions as employees would but much faster with a near zero-error rate.

From experiences with a customer base spanning multiple industry verticals, Kohli says that possibly the most important impact of ERP RPA bots for IT is the fact that they significantly enhance but in no way impact existing systems. But there is an understanding here that the roles of those within the business will evolve as employee time shifts away from manual, repetitive tasks to make way for finance specialists to offer more value with analysis and planning or HR to focus on nurturing teams.

“As many enterprises plan migration and consolidation projects of major ERP systems, it’s worth remembering that RPA can deliver powerful punches for one-off projects too. Santander Consumer Bank deployed bots to handle a systems migration converting data from one system to another. The cost for an external vendor to complete the two year project was a projected $2m. With bots running 24/7 the migration was completed in just 12 weeks with 100 percent accuracy,” said Kohli.

These are just a few examples out of a myriad of opportunities where intelligent automation can play a critical role in transforming the performance outcomes of ERP systems and unlocking value for the business.

The hands-on ERP RPA free zone

No analysis of RPA for ERP would be complete without mentioning the human factor and the no-go areas for software bots. There’s an ‘old’ joke in RPA: the masseur will never be out of a job. Now, unless an ERP system is being used by an enterprise-size massage company the joke doesn’t work, but it still makes the point i.e. it’s the hands-on human roles that RPA won’t be getting rid of any time soon. Those employees that do find their roles displaced by ERP RPA can now look to more intuitive, human empathy-focussed hands-on roles, which should be ultimately more fulfilling anyway.

As we now know, to make ‘real’ ERP software robots work in demanding real-world mission-critical applications, a degree of pre-engineering (process mining) is involved. We will need to follow this process with an exhaustive approach to data quality (we could call this the oiling and maintenance processes that any machine needs) to ensure that our bots function in a fluid manner. Further still, we will need to keep an eye on the whole robot factory floor so that we know who (or, in this case, what) is being tasked with which elements of which workflow.

This is the rise of the ERP RPA robots. They’re friendly bots that aim to please – just be sure to keep them well-oiled and tightened.