Guy Kirkwood is COO of UiPath, a leading intelligent automation vendor. He has 20 years’ experience in outsourcing and a particular interest in the future of shared services and BPO. He is a writer and presenter on RPA and the future of work.
IRPA AI Founder Frank Casale recently had an opportunity to catch up with Guy Kirkwood, COO of UiPath. In his interview with Frank, Guy provided his observations on automation activities happening in the market and insights into how organizations are adopting automation.
Frank: How would you describe overall client activity in automation this year 2016 as opposed to 24 months ago?
Guy: There are fundamental differences between current client RPA activity and what we saw two years ago.
The most obvious difference is the volume of activity. There’s simply no comparison between the number of UiPath robots we’re currently placing into large scale automation implementations compared to 2014.
This spectacular pace of client activity is driven by an increased urgency for rapid time-to-market. Back in 2014 client activity resembled a form of haruspicy. A pilot would get approved and the results would be pondered for convincing evidence of cost savings and easy implementation. Then a larger proof of concept would be approved and another lengthy divination of savings and implementation benefits would take place.
Now client interest is driven as much by competitive advantage as cost and performance benefits. The business has realized this technology can bring dramatic benefits to its business model by making completely new consumer services possible – not just enabling better quality and customer service levels. This realization has led to a much greater sense of urgency.
Behind these important changes in client urgency and adoption lie several other significant differences between now and twenty-four months ago.
Credibility: twenty-four months ago RPA had serious credibility issues – not because it was a new and unfamiliar technology (though that certainly had an impact) but because what the technology claimed it could do was unprecedented. When clients heard or read that robots were: non-disruptive; blazing fast; error-free; inexpensive & quick to implement – and could reduce costs by double digits, it was hard for many of them to take RPA seriously. Over the past year, however, a rising tide of irrefutable use cases have emerged which put these credibility issues to rest.
Capability: two years ago robots were primarily single-user automation tools residing on the employee’s computer. Now leading RPA providers offer enterprise products capable of deploying autonomous robots in large scale automation solutions. In other words, robots – not humans – are managing robots. Single-user automation does provide high value, but unshackling 24/7 robots from 8/5 humans has allowed RPA to create absolutely compelling business cases. Additionally, these enterprise products meet client requirements for security, scalability, unobtrusiveness and compliance.
Business as champion: in 2014, BPO was talking to the business and RPA was talking to the IT Department – and IT tends to view robotic process automation as a low cost bandage for unintegrated legacy systems, and thus a competitor to their more elegant and expensive system integration projects. Once enterprise RPA solutions emerged, business recognized the technology as an excellent alternative to BPO and began championing it in their organizations, armed with effective business cases.
Frank: What processes do you see as most commonly being automated?
Guy: Certainly horizontal business processes such as FAO (e.g. invoicing, procure to pay, record to report) or HR (e.g. onboarding/off-boarding; compensation & benefits; position changes) are very common automation candidates.
However, it’s important for clients to understand that RPA excels at automating at the activity and task level within the process. This means the technology isn’t restricted to end-to-end implementations and can achieve significant benefits even when partially automating process steps.
For example, the insurance industry is an aggressive adopter of RPA technology, despite the fact that claims processing requires points of human intervention. Take the five steps of medical claim adjudication:
– Initial review
– Automatic review
– Manual review
– Payment decision
– Claim payment
The first two steps are driven by rules and easily performed by robots. However, only humans can do a manual review – checking medical documentation against claim records or determining medical necessity. In the payment decision steps, robots can follow rules to determine whether a claim is reimbursable or not. However, only humans can determine if a claim should be reduced, based on how well the billed services match the diagnosis.
While it seems reasonable to assume claims processing systems would have already automated these rule-driven steps, the reality is tens of thousands of BPO employees are manually compensating for poorly integrated legacy systems. By closing these holes, even partial process automation is delivering huge savings and performance benefits.
Frank: Where does one start if interested in automation?
Guy: The very first step is identifying an RPA Sponsor from the business who will establish the technology as an enterprise-wide strategic priority and secure required corporate resourcing. The next is naming an RPA Champion, responsible for RPA evangelization and tactical implementation steps – such as team formation; pilot and industrialization.
Role and training is often the area organizations find to be unfamiliar and challenging. The UiPath Client Services Group works with our clients in this area and recommends these core team roles:
– RPA Team Lead
– Business Analyst
– Automation Specialist
– Process Controller
– First Line Service Support
Each role requires specific skill sets and a training methodology that integrates both role and skill level-based approaches is recommended.
Certainly, as part of the initial pilot and proof of concept activities, the need to identify best-fit processes is self-evident and the qualifying characteristics of such processes are no surprise: stable, high volume, highly manual and repetitive; rule-based with low exceptions, and consist of structured data.