29 June 2017

UiPath Innovator Interview Series: A Conversation with Boris Krumrey

29 June 2017

UiPath Innovator Interview Series: A Conversation with Boris Krumrey

UiPath spoke with Boris Krumrey, recently named Chief Robotics Officer at UiPath. In this role, he leads the product and integration design of the company’s robotic process automation and AI technologies. Prior to UiPath, Boris was Global Head of Applications Management Operations, Automation & Robotics at Atos - an IT Services provider with over 100,000 employees in 72 countries. 

 

We caught up with Boris during a rare break in his preparations for UiPath’s impending product release. Our conversation reveals what he sees coming up in robotic technology, glimpses into new areas of UiPath innovation, issues holding back RPA ubiquity (and what he’s doing about them).

 

UiPath: As head of global application management services at Atos, the UiPath product platform was part of your service offering. Now you’re UiPath’s Chief Robotics Officer. How has that user experience influenced how you view your new role?

 

Boris: At Atos, we were continuously driving quality and LEAN Initiatives in waves to improve our managed services and consulting capabilities. However, my global responsibility for application management operations meant that I wasn’t just focused on tools and methods for efficiency and cost improvements, I was also searching for innovative technologies and ideas to develop next-generation services and new business opportunities.

 

Of course, automation has always been a key enabler of application services efficiencies, so to explore beyond familiar technology and best practices we really had to think out-of-the-box. Fortunately, I was aware from earlier work at Accenture and Infosys that software robotics was quickly advancing. When we looked in those types of technologies, it became clear that only RPA had the broad capabilities needed to service clients with large numbers of legacy or proprietary applications.

 

Finding the right technology was one thing. Implementing it across sixty Atos service delivery organizations was something else entirely. We had to define a roll-out methodology and train approximately 40 LEAN process consultants. Then educate more than four thousand people worldwide about RPA, show them how to assess automation opportunities and build business cases for implementation in our sixty delivery units. 

 

So, what lessons did I bring from all this to my role at UiPath? Two things.

 

I learned RPA technology must enable customers to achieve real cost, quality and productivity benefits in a sustainable way. In other words, a technology may seem cool and great. But, if it doesn’t help you grow your business, improve your services and cut costs - why bother? Furthermore, even if RPA does help the business achieve these goals, but the implementation costs and effort are too high - again, why bother?

 

My product design is guided by these key questions:

  1. How is implementation risk reduced?
  2. Once deployed, how does the product minimize operational risks?
  3. Is it “future-proof” - designed to eliminate technology dead-ends?Will customers easily achieve cost reduction and productivity targets?
  4. Does the product enable customers to deliver services in a faster way?
  5. Will people with implementation skills be easily found at client locations?
  6. How will it help our customers and partners generate new business?

UiPath: Looking across the industry, RPA rate of adoption doesn’t seem to be matching hype and expectations. What’s going on? Do customers not have the advanced technology they need to tackle more automation? Or, does the technology they do have lack what’s needed to make process automation faster and easier?

 

Boris: I believe better automation and deployment tools are what’s needed to accelerate RPA adoption, not vendors bringing more AI/Cognitive functionality to the table. As Forrester and HfS Research pointed out in their 2016 reports, most clients haven’t even started complex, large scale transformation projects. They’re still piloting very basic and simple process automations.

 

Right now, AI/Cognitive technologies are incorporated into our product platform. And we have a roadmap showing where those capabilities are going in the future. I can’t imagine any UiPath customer is sitting idle at automation, waiting for upcoming releases so they can go back to deploying our robots again. Of course, those future releases will increase automation opportunities, but our development team isn’t holding any customers back right now.

 

In these types of conversations, everyone tends to focus on what technology’s coming next. And what’s almost always overlooked is the difficulty of efficiently transferring process knowledge to the robot. That happens when process subject matter experts collaborate with RPA developers to create workflow automation. Knowledge transfer is not an easy thing.

 

For example, in the outsourcing industry, knowledge transfers for business or IT processes is always on the critical path and often have contractual commitments to deliver on time and on budget. To do this requires very rigorous time schedules of interview sessions for capturing the knowledge. It’s a key focus area for every outsourcer, and where they put transparency, KPIs and so forth to document and escalate: "Dear client team, you're not fulfilling your obligations. You haven't given me these guys to talk to, so now I can't finish my knowledge transfer on time.”

 

What I’m saying is this. Whether people choose to believe it or not, one of the biggest inhibitors to large scale automation is the inability of organizations to manage the time of those SMEs and RPA developers. It’s simply the case that people are always busy with other things. No more and no less. That’s why outsourcers learned from hard experience to protect knowledge transfer with tough contract clauses.

 

In robotic knowledge transfer, people think it’s all done with agile, but that’s rubbish. There may be agile in the development process, but in RPA organizing knowledge transfer is as big a task as transition is in outsourcing. Many companies haven’t realized this yet and its why their robotics can’t scale big. If you look at the ones who’ve been slow to really implement, it’s because they're not organized to do it. They haven't put methods around the knowledge-transfer-to-robot process.

 

At UiPath we’re going to introduce tools and integrate cognitive technologies to make this knowledge transfer easier. Then, when our customers deploy their UiPath robots, the entire process of extending automation will be so well supported that it will make a significant difference.

 

UiPath: I can see how companies could underestimate the challenge of knowledge transfer in RPA, but service providers and consultancies must be aware. Do you think they see a good business opportunity here?

 

Boris: Perhaps. But keep in mind, by its very nature the target of the RPA approach is to automate the As-Is process state. And consider our ability at UiPath to record user activities and turn them into RPA. Hitting that target with our technology will always be faster. The key to long term success, however, is to implement a support organization that operates and maintains a company’s RPA. I’d say here is the real prize for the service integrators and consultancies - offering RPA as a Managed Service.

 

UiPath: In the RPA industry, and within other automation industries as well, you find vendors focused on developing specific products for particular industries. However, customers often resist investing in a wide range of related third-party applications. How do you see that playing out?

 

Boris: Well, we’re fortunate that analysts like Forrester, HfS Research, etc. consider us one of the leading RPA software vendors. So, customers know we’re worth a close look. To demonstrate why we’re the better choice than niche vendors, we’ve developed deep AI capabilities in the computer vision of our robot runtime and by integrating technology components from third parties like Google, ABBYY, Microsoft, etc.

 

There’s a race between product vendors in the areas of orchestrated automation, BPM, DCM and Robotic Process Automation. They’re all driving towards the adaptation of AI, fighting in the same market space- but taking different angles.

 

In my experience, most big automation initiatives are driven by leading service integrators and third-party consultancies. Like the companies you mentioned, they’d rather focus on a few technologies they can efficiently put their own expertise - best practices, RPA solutions and so forth - on top of it.

 

Another challenge for our industry is these consultancies and service integrators make a lot of money in the BPM space. When I was at Atos I was around a whole bunch of people operating in the BPM space. They looked at RPA and said, "Oh, it’s like BPM. It’s just a little different and quicker.”

 

They view RPA as an opportunity to not just fix something, but also a chance to extend the client’s scope of process transformation. Consultancies and integrators will also drive customers to the BPM solution because they have more skills in this area. All this makes them an RPA competitor.

 

UiPath: Where do you see UiPath’s technology roadmap taking its product?

 

Boris: Our product is moving to AI in two stages. The first is cognitive robotics, extending our robot’s computer vision capability with machine learning as well as incorporating cognitive services from IBM Watson, Microsoft, Google and other technologies. This stage, which is much closer than some may realize, will enable customers to automate unstructured data and complex rules. To illustrate, think of robots with sophisticated case management capabilities.

 

The second stage is authentic AI robotics, where robots reason and remember; learn from data; engage users interactively and proactively identify process automation opportunities.

 

As I said, all this is happening quickly. Which is why during the previous year we moved to two major releases a year, enabled free downloads of our Community Edition product and launched our online self-learning UiPath Academy. We want the proficiencies of our customer, partners and RPA professionals to keep pace with, and take full advantage of, all these innovations.

 

Of course, technologies and capabilities don’t count for much with our customers unless they translate into meaningful features and benefits. So, let’s illustrate these coming innovations from that perspective.

 

More and more people are becoming aware of a new channel for user interactions called Voice User Interface (VUI). Certainly, Accenture’s spotted it. People have been exposed to it in the form of Amazon’s Alexa or Google’s Assistant. Instead of performing actions with keyboards, they’re simply talking.

 

At UiPath we spotted it too, and we’re going to show customers how to talk to a Google Assistant and say, “Ok Google, I want to talk to the UiPath Bot.” “UiPath Bot, I have a problem with this, and this, and this.”  We’ll also show how we use other technologies to trigger a robot to perform actions and come up with information.

 

Imagine a full life-cycle example of a Dynamic Case Management System with machine learning; in which an email is processed, intelligently categorized and a response is sent. We’re working on this feature right now.

 

Remember my point that knowledge transfer was the single biggest obstacle to large scale RPA? Think of being able talk to a VUI robot on your screen, with the conversation going like this:

 

You: "I want you to record the following activities that I'm about to do."

Robot: "Yes, what do you want to call this?".

You: "I want to call it 'checking emails' ",

Robot: "Ok, tell me when you're ready."

You: "I'm ready now."

[You perform the activity] 

You: "I'm done now."

Robot: "Right! I've got your sequence down. Now I'll send this off to the RPA developer and schedule an appointment for you and the developer to review the sequence and finalize the automation."

You: "Ok. Thank you."

 

Can you see how we’ll turn cognitive and AI technologies into features that change the way people think about process automation?

 

UiPath: Someone once said, “life is what happens while you’re expecting something else”. Looking into your crystal ball, how would you apply that to the RPA industry? What are the “something else’s” we aren’t expecting?

 

Boris: It seems to me the market is waiting for a magic moment when RPA technology demonstrates a repeatable amount of large-scale, productive process automation by robots.

 

But what if it isn’t a magical technology moment that transforms the RPA landscape? Perhaps robotic history will repeat itself and depend instead on human, social and cultural breakthroughs?

 

Look at what happened when the first industrial robots were developed. They never really took off in Europe or America because people were too skeptical about robotics. But when they were introduced to the Japanese - who love technology and believe in robotics, it was a different story.

 

The Japanese looked at industrial robots and said, "Japan is a small country, we need higher productivity to have these robots to do our work."  Then, as soon as Japanese showed they could manufacture cars with robots at large scale, the skeptics in other cultures said, "It works, and here's the proof."  Today, no one has to convince anyone about industrial robots. Instead, people are coming up with stupid ideas like, "Shouldn’t we be taxing those manufacturing robots?"

 

This is what I’m talking about. When you listen to what’s being said about RPA’s future, the conversations are all focused on technology. We're not seeing the things that are likely to play a much bigger part in the “magic moment” breakthrough.

 

I hate the images companies are constantly putting up of android-looking robots typing on a computer. That's the worst kind of image because people see it and say, "Look what's going to take my job." They look alien. They look dangerous. It's like adding oil into the fire of fear for robots. 

 

That's totally against what the industry’s trying to achieve.  For example, at Atos I had a target to achieve specific workflow savings.  Even though I had technical application management support engineers, people who understand technology as well as programming basics - and should have been eager to embrace robotics - they didn't. They wanted to come up with different ways, they pushed back as much as possible. 

 

Why? They were all afraid of losing their people. No one wanted to give up people. The whole of middle management was against it.

 

UiPath: I can see headcount losses causing career concerns for managers. Maybe putting robots on the org chart might help. Sr. Mgmt. would get an accurate view of a manager’s true FTE count and also know who was embracing automation and who wasn’t.

 

Boris: Yes, it’s in the culture. Particularly in those cultures that emphasise status. Seeing, for instance, two hundred robots represented would give a feeling of significance. It would motivate them to stay looking good and incentivize them to do more.

 

You could also put “increase the number of robots” in personal performance assessment targets. Imagine if robotics was part of “exceeds expectations” evaluation criteria. In some cultures, it would make middle management - typically where you see the most resistance to change - drive towards these targets with their teams.

 

In my opinion, another unexpected part of RPA’s “magic transformation moment” is the need to develop trust - specifically as it pertains to AI technology. This is a very important point for everyone to consider.

 

Imagine a company really scaling up and automating their processes with AI capabilities. Then, with all the automation in production, not really knowing what their intelligent robots will come up with for workflow decisions. 

 

Remember, they can't control those robots because they’ve basically been trained to replicate human thought and behavior. So, when they learn from data, reason, remember, reach a conclusion and make decisions, the company can only put in break points and check the decision before any action is taken. Companies will have to do this. Only after checking and consistently validating decision quality will organizations be able to trust their virtual workforces.

 

So that will be a key - and I think, unexpected challenge in the not-so-distant future. How do we enable AI technology such that our customers feel in control of their robots? Which always comes back to the trust issue. UiPath will need to provide control mechanisms, which also means our client services team will have to provide guidance on governance structures and so forth.

 

So, much bigger questions are on their way to drive the “magic moment” conversation. If you ask me to look into my crystal ball and answer, "What technologies?" Then yes, I can say it's going to be Voice User Interface and these other things coming up. But they’ll very likely be overshadowed by far bigger human, social and cultural issues.

 

UiPath: Boris, you’ve brought a distinct mix of RPA implementation expertise and thought leadership to this interview. Thank you so much for your generosity of time and insights for the audience.


by Mina Deckard

TOPICS: innovator interview series, AI, 2017, Boris Krumrey

Show sidebar