Robotic Process Automation: what RPA can do and what to look out for

Robotic process automation (RPA) is making waves as a way for businesses to cut costs and improve process efficiency. It promises to reduce repetitive work, increase accuracy, and improve job satisfaction among knowledge workers. But what exactly is RPA, and what is it truly capable of?

In a nutshell, RPA involves training a software bot (that’s the “robotic” part of RPA) to carry out repetitive tasks in place of a human employee. RPA processes typically involve data-related tasks where information is transferred from one type of software to another. For example, imagine an employee in the accounting department. They get an email from a supplier containing an invoice. Now they have to open the email, download the invoice, import that file into a bookkeeping or ERP program and —depending on how sophisticated the system is — potentially enter the invoice details manually.

This type of work is typically repetitive and not particularly interesting. Most knowledge workers view it as a necessary evil that has to be gotten out of the way. So what if these repetitive tasks could be processed automatically? An RPA bot could be programmed to open the email, download the invoice file, upload it to the appropriate software system, and use optical character recognition to extract the relevant details and enter them in the ERP system. The company would save resources, the risk of human error would be significantly decreased, and the accounting employee would be freed up to work on more intellectually stimulating tasks.

This is exactly what makes RPA such a hot topic.

What types of RPA are available?

RPA is a rapidly advancing field, and there are various solutions available with differing levels of sophistication.

RPA 2.0 is the most basic of the modern RPA systems. At this level of automation, a bot can be trained to automatically transfer data from multiple input sources (such as email and spreadsheets, for example) to a CRM or ERP system. It typically “learns” by watching its trainer (a human employee) perform a task in a graphical user interface (GUI).

After it is trained, the robot is deployed on a central server. It is manually assigned a workload and can then go to work, taking over tasks from one (or more) human employees.

While that statement might sound ominous in terms of job security, these types of RPA systems are nowhere near sophisticated enough to fully replace an employee.  They are only capable of dealing with strictly defined processes and highly-structured data — exactly the type of tasks that most knowledge workers would happily hand over to an assistant.

Autonomous RPA, also known as RPA 3.0, takes automation to the next level by introducing some aspects of artificial intelligence (AI) into the mix. With this type of RPA system, there is no need for a human employee to schedule the bot’s tasks. The bot is capable of dynamically balancing its own workload, scaling up and down without intervention.

In RPA 3.0, the RPA systems are typically deployed to the cloud and are capable of processing more complex tasks. However, they are still limited to dealing with structured data.

Cognitive RPA, or RPA 4.0, is the most advanced form of robotic process automation. It introduces a wider variety of AI functionalities, adding a cognitive layer on top of a classic RPA process. This cognitive layer might take the form of an intelligent virtual entity, or IVE, which would be able to automate and schedule even complex sequence of tasks. With the help of machine learning algorithms, the IVE could even learn to work more efficiently over time. Depending on the exact type of artificial intelligence used to build the system, the IVE might be able to process unstructured data, communicate with users using natural language processing, plan tasks, make decisions, and more.

How would this look in practice? Here’s an example: an intelligent virtual entity with natural language processing and decision-making abilities could be trained to monitor social media posts for mention of different products made by a company. It could analyze the posts for sentiment (positive, negative, or somewhere in-between) and forward comments with strongly negative sentiment to the appropriate customer service representative, who could then respond to the comment appropriately.

Many other scenarios would also be possible, and a huge variety of customized solutions can be developed to suit the particular needs of a company or corporation.

What are the benefits of RPA?

RPA’s most obvious benefit is lower costs, along with improved efficiency and accuracy. For high-volume business processes, cost savings can be as high as 30% in some cases. RPA can also be a major boon for companies in the financial industry by making it easier and quicker to adhere to compliance guidelines. Additionally, robotic process automation often improves job satisfaction for knowledge workers, who are relieved of monotonous and repetitive tasks that they typically find tedious. With these types of tasks taken care of, they can move on to more intellectually fulfilling tasks that allow them to create more value for the company.

What risks might there be with robotic process automation?

The first risk that most people think of with RPA is resistance from current employees — after all, who would willingly put themselves at risk of being replaced by a robot? But the reality is that RPA systems don’t typically result in layoffs. They might reduce the number of new hires at a company, but they don’t replace the existing employees altogether. Instead, they help employees to work more efficiently.

Employees often view the RPA bot as an assistant who makes their life easier and their work more interesting. More advanced RPA bots, or IVEs, can interact with employees using natural language and can even be given a personality and artificial emotions. That means that they often endear themselves to their human colleagues—just like Poppy, an RPA bot who was invited to the office Christmas party.

Another risk that companies should keep in mind is that even automated systems can become bloated and ineffective if they are improperly designed. When RPA is used as a Band-Aid solution for inefficient business processes, companies run the risk of designing an RPA system that could be a liability in the long-term. Sure, robotic process automation might mask inefficiencies for a while. But if the underlying business processes are not well-designed, the problem will eventually need to be addressed at its root.

A few things can be done to avoid this. First of all, companies may want to involve the operations department in designing and implementing the RPA system, rather than handing it over entirely to the IT department. In addition, they should partner with an experienced RPA provider who can help them to develop an effective strategy.

Finally, less sophisticated RPA systems could run into problems when the software they work with is updated. Theoretically, something as simple as a change to a drop-down menu could wreak havoc, as the bot would continue to “click” the same button until it is reprogrammed to do otherwise. This would mean that the entire RPA system would need to be updated along with each software or OS update. However, with the help of artificial intelligence and intelligent virtual entities, these types of problems could be easily identified or even automatically solved.

How to make robotic process automation work for your company

If your company is interested in implementing an RPA system, you’ll need to have a very clear idea of where it would be most useful for you. Start by identifying bottlenecks in your processes that could be relieved with automation, then decide which tasks are most suitable for RPA. Work closely with your operations department to make sure that the underlying processes are as streamlined as possible.

Whether you decide to use RPA 2.0 or something closer to cognitive RPA, you and your service provider should carefully test and validate all RPA systems before rolling them out. This will ensure that any inefficiencies or unexpected results can be identified early on — before they impact your business operations.

Finally, if you need to automate more complex processes — such as those that include language processing or decision-making, you’ll need a service provider who can integrate artificial intelligence or intelligent virtual entities into your RPA system. RPA 4.0 is an emerging technology, but AI developers like Blackzendo are pushing the limits of what it can do.