RPA: Think long-term, not short-term

Suman Reddy, MD of Pegasystems India, feels that Digital Process Automation is the way forward. When we apply DPA, it selectively thinks about where you need an RPA, where you need a case management, where you need a business process management and where you need AI.

Which are the types of industries using RPA?

Industries have many processes across their entire supply chain, all the way from back end processes of automated operations internally along with customer facing application such as service marketing and sales. We are identifying areas where there is a lot of repetitive manual sort of a work and applying a robotic automation to automate those processes.

There is a demand from their customers and increasing pressure of digitization, so organisations are investing heavily in terms of automating some of the processes that they have through RPA. Take a use case like a car manufacturing company Ford. They have about 8000 internal users who are expected to do a lot of routing and researching, essentially the entire workflow from one point to another and they use Pega RPA.

Also, there are organizations where you have a lot of information spread over many many systems and evolved over many many decades. In places like that, RPA has proved to be a huge productivity tool. Because by applying some of the automation of manual things that you have to do in certain BPO and CRM setups, it has hugely increased productivity and will continue to do that. But they need to really think about how to provide uniform consistent experience to their customers.

What about Robotic Desktop Automation?

Banking, insurance and telecom are the three sectors where we have several customers who are using desktop automation. Large Indian telecom companies do a lot of desktop robotic automation. They have thousands of customer service users who are receiving calls or providing service to their end customers and have to toggle between 8-15 different applications on their desktop to service one end customer. If you have a nice desktop robotic process automation tool, with one click you can update all of your records across say15-18 different systems.

When you look at banking or insurance, during the process of modernization, they might have invested in a lot of modern tools. But they all sit with so many legacy systems and silos and each one of them to have some business value for them. So therefore, desktop automation becomes a big deal for any of these large organisations.

What about RPA, AI and ML?

When it comes to RPA and AI, we also call it cognitive RPA. I believe that's in the elementary stages. RPA and AI are two distinct things. AI can be performed to fill up the programme to perform certain tasks. You can do things like Natural Language Processing. However, the process as a whole itself doesn't learn. On the other hand, AI can help optimise all the information that gets fed into RPA. AI can make RPA smarter.

AI needs to be at the heart of the overall digital transformation. Things have to evolve from doing this digital worker, which is like a small siloed work, to what we call as end to end Digital Process Automation.

What about Industry 4.0 with actual robotic workers taking command from digital bots being run entirely by AI-ML?

Some industries have applied some of systems to help automate manual things. In Japan they opened a hotel where there is no human worker. Self-contained robotic systems allow you to you register yourself and get you cards to go to the room. Now can this be done at an industrial scale level? Can you do manufacturing with actual robots? The answer is yes.

Large manufacturing companies already have lots of robots assisting humans. How far can that be taken? With the way rapid innovation is taking place, it could be a reality in the future. How many years it’s going to take is anybody’s guess. In 5-7 years you could see some small sector industry having a fully automated robot driven industry.

One headline said that “40% of all jobs will be lost due to AI”. How do you handle all these fears that AI-RPA will take away jobs?

In the United States, where I think a lot of RPA implementations have taken place, unemployment is at the lowest, especially in the technology space. I don’t buy all these headlines which talk of people not having a job due to AI-RPA. But will it change the job profile? The answer is yes. Computers will take over a lot of manual jobs.

But I think it'll also give an opportunity to evolve the human roles into a different high calibre domain. We are currently tied up with a university academic programme where we are using a platform that allows you graduate to learn higher level of skills which are not necessarily going to be repetitive in nature. We believe you don't have to traditional programming anymore because we now have intelligent systems which generate support for you. So we're actually giving you higher level of skills which will allow you to be immune from some of the AI-RPA related job losses. So jobs will definitely evolve.

What about RPA stifling innovation and the fact that it might be difficult for a non RPA company to maintain these processes?

Pega conducted a global survey where 87% of the decision makers who are using some sort of an RPA today have also experienced failure in some of their projects. What we learnt also is that RPA over a period of time can create silos within your companies. If you’re not putting your entire end to end automation in place then you may face work disruption and project delays.

Sometimes RPA is implemented in some organisations by an incremental buyer. They may be investing in some low-end RPA tools. They come at a cheap cost when you buy more licences. They are building a lot of these bots across the organisation or in silos. There are hundreds and thousands of bots created and rightfully so, because they are trying to achieve some sort of business game. But greatly organisations will fail when you’re not looking at it end to end. What we are saying is do not do it in a siloed fashion, which is what a lot of organisations are doing. That over a period will have so many disparate systems that it will result in lot a project delays.

Digital Process Automation is the way forward. When we apply DPA, it selectively thinks about where you need an RPA, where you need a case management, where you need a business process management and where you need AI. You need to actually apply technology elements in equal proportions. You need an all-encompassing approach, which takes the power of RPA, which takes the strength of case management, which is at the centre of automation of workflow, then applies AI accordingly.

How can SMBs/SMEs implement RPA? Is it necessary for them?

My question: Is RPA really a need for them? Don't go with the whole hype cycle that we're seeing right now where everybody wants to suddenly use an RPA tool to automate something or the other. There are some short terms gains, but more challenges in the long run.

Points to ponder over…

• The overhyped “RPA bubble” bursts: RPA users will enter the RPA trough of disillusionment. They’ve been told RPA is a cure for all digital transformation and process inefficiencies but are finding it’s not right for every situation, doesn’t scale, and hard to deploy.

• RPA becomes more than a cost play: Moving beyond the obvious cost benefits, organizations will focus more on using RPA to improve customer experiences by assembling disparate processes together to get closer to end-to-end automation.

• RPA today is looked at as the main step in the digital transformation journey: There is significant market hype in RPA right now, and our clients ask us to help them. There are a lot of great use cases for RPA to help fill automation gaps. On the flip side though, our clients who have used other RPA tools often come to us frustrated. In a recent survey we conducted, we found that global RPA users opine that RPA takes longer to deploy than expected – up to 18 months on average – and that these bots are prone to break over time. We advise our clients that RPA can be a quick fix, but that in the long run, DPA should serve as the core foundation for your automation solution across the enterprise.

• Mixing RPA with AI: We believe that cognitive RPA at this stage is mostly hype. RPA and AI really are really two separate things. Specific elements of the AI engine can be programmed to learn: Natural Language Processing, for instance, can learn to recognize new words while OCR can adopt new forms. However, the “process” as a whole doesn’t learn. RPA, in a similar vein, cannot “learn” or become smarter on its own.

• RPA 2.0 will sit on the sidelines … for now: True Cognitive RPA – where AI is embedded directly in RPA to add intelligence to processes – is still quite far from being a reality. Though RPA will remain a largely unsophisticated process in 2019, we will see more examples of AI and RPA working side by side next year.