Session Information:

How to use a combination of process mining and automation to continuously discover, automate and monitor your business processes

McKinsey estimates that $3.6 Trillion of work is automatable (Developed Countries)*
But how do you go about finding that work?

Process Mining enables you to discover the best automation opportunities hidden within your processes. Our integrated process mining and automation solution help customers get from discovery to implementation, through to measurement, and back again.

Key Takeaways

• Learn how to leverage process mining for better automation outcomes
• Why process discovery from all angles accelerates the automation flywheel
• See how organizations are putting this into practice

*Source: McKinsey Global Institute, “A Future that Works” –All US non-physical automatable wages, extrapolated to include Japan, UK, Germany, France, Spain, and Italy.

Session Transcript:

Us today coming directly from Frankfurt in Germany. We have Rudy Kahn, who is the vice-president of process Mining for UI path.

And you could not be better served Dann by Rudy because Rudy's one, I'll ask you one of the fathers of process mining and you'll get to hear directly from him, he was probably one of the first to recognize the potential for process mining. After 10 years in business consulting at IBM and other companies rudi founded Process Gold in Frankfurt, Germany in January of 2010, It was the first consulting company specialized in process mining.

And then, in 20 16, Process gold publish, it's also a solution, which is used by E Y, Worldwide, as a standard for process audits. At the end of 2019, process gold was acquired by UI path, where Rudy's supports, as vice-president for process mining. The over 8000 customers of UI path, in the integration of process, mining, and robotics, process automation. So, Ladies and Gentlemen, let's learn from one of the creators and founders of this incredible technology of process mining on how it works, how it creates value for all of us really, We're super excited to have you with us. Thank you for taking the time to share your expertise with our global audience today.

Thank you so much again for for having us. It's a real real pleasure to be with you guys today.

Well, there's not much to say about me in that, because basically Josie said everything.

You know, I'm in this business for quite a long time and in the last, you know, few days I Lawrence, 1 or 2 things about process mining, and it's my great, great pleasure to share with you today.

So, I can tell you when I met, but it's fun to us for the very first time in 2009 and you know, if you are into process Monday, you have heard this before, because we'll release the godfather of process mining. You know, he had this crazy idea more than 20 years ago to reconstruct business processes only from data at that time. Nobody believed it would be possible. I didn't believe it is possible when I met father, you know, for the first time in the Netherlands in 2009. But, he showed me, you know, what, what he can do, And I thought, oh my God, this is so amazing, if this will work, it's going to re we, or at least I am looking at processes forever. Because before process called, also, consult with IBM, and some other companies for about 10 years.

And, no, but I would talk about this, maybe in a moment. It's not about me. It's really about, about process finding, today, and it's about our customers. So, today, we will look at process learning from a slightly different angle. You know, of course, we're going to cover some of the basics.

But, we want to show you really like the next emulation of process mighty. Because we believe that process mining needs to be more than just process discovery. And that's the reason why you iPads, acquire process goes back in 2009. And, you know, there was a really funny story I just want to share with, You know, I was, At that time, I was in California in Silicon Valley working on the international expansion. And one day morning was very early morning, because they thought I'm in Europe, but I was in California, and what that exactly the different time zone. So they woke me up at five in the morning, and we had a very good chat for about an hour on them.

I was asked to reroute my flights to come to New York because we really need to talk. So I did two days later. And after a recession of 14, 14 hours, I think. And the second bottle of red wine and a steak house next door. We actually decided that process mining and RPA, it would be really a mentioned heaven.

And but we should join forces. It took us six months to figure out the details. And then, in October, 2019, we signed the contract.

And 10 days later, I have been onstage with our management. And we announced the acquisition. And since that, I can tell you to also roller coaster ride.

It's amazing what process mining can do for RPA, and I'm very, very glad to have Neil with us, so new comes from from grow mark, one of our customers right now. And Neil is the Director of Enterprise Continuous Improvement. New wife got wake up and thank you for joining us.

And you know in case you don't know Goldmark, I'm sure you can tell us a lot about it but very short, it's a leading agricultural co-ops in the US and new Over there is the six Sigma most about and is focusing on leveraging technology in order to really improve process output.

So, Neil, Welcome, So happy you. are you, our customers, so happy you are with us this morning, and I know you're on vacation, so apologize, you see your February, I think, for you. It's, oh, what time is it for you? At 7 0 AM in the morning?

Anya, OK, good. So, after breakfast, let's see. Neil, I would like you to introduce yourself, and maybe tell us a little bit about the challenges you are, you know, you are facing at .... So, what, what's your, what's your current situation?

Sure. So, as I said, I'm Lean six Sigma Black Belt been doing process work for about a decade now.

Really, just focus on getting teams involved in the process improvement. Getting the black belts and green belts out of the way, with the Lean six Sigma methodology and allowing the teams to be able to see their process. And, if teams can see their work and see what they're doing and, you know, kind of a mantra of Lean is to make problems visible.

It's amazing how much problem solving and improvements teams can do all on their own, once they really realize how their processes work and who's doing what. And start to ask them, Why are you doing that? I thought you were doing that. People really don't know how work flows through an organization.

It's just as eye opening to start to show them how it does and then let them no problem solve on their own, as I said.

So by, by profession and passion, I'm a trainer. So I've been coached to have short responses here but I just want to re-iterate what Joseph said. Please use the questions and ask us questions. Rudy broke the ITU chart a couple of years ago. So he's great to provide responses as well. But we'd love to go deeper on anything that, you know, the panel would like to hear more about. So, in terms of challenges. That grow marks, and we're a 100 year old organization almost, and being from agricultural that's very conservative, it's, you know, you know, we work a lot with farmers and people out in the field.

So, it's becoming more and more technology driven, everyday as all industries, But, historically, it was very manual, not a lot of technology and digital footprint there in this industry.

Just recently, we've, we recognize the need to really jump into the 21st Century, and, at Grow Mark, we've implemented a new ERP system. We're now an sap shop all the way through, and through our enterprise.

And part of the promise of that was to provide a rich treasure trove of data, which would provide us robust analytics and understand, Well, like most organizations that jump into an ERP, or you're faced with the chicken and the egg situation of, you don't know the technology fully that you're jumping into. And you don't know how your processes today are going to fit into that. So you take what you have, and you try to fit it in as best you can.

And what happens is there's there's opportunity in the new technology that you didn't know about, and so you have old processes and new technology.

So, we're, at that stage now, got the new technology, implement it. We're running on GCP. Everything's turned on, and working, But it's not optimized. We really don't know how it's working, and who's doing what, and we're still doing the processes.

We did, Priya, Sappy, kind of, you know, round hole, square peg, type situation. So, process mining was really, really powerful for us, to be able to just plugin to that data, and be able to visualize our process in a very, very quick manner.

A lot of our processes are siloed.

And a lot of organizations face that where I do my job and you do your job and never to try and intertwine. We have siloed IT requests. So every different area is hitting IT with their own request for updates, changes, bug fixes, whatever it may be.

But nobody has the end to end process perspective, to say, which one of those should be prioritized in terms of throughput for the whole process, Rather than just improving your process, whether you're the constraint or not, Nobody is really looking at it through that lens.

So, that's what we're trying to do with process mining, is, to make those processes visible, put those problem solving in the hands of the teams, and then let them determine, Well, know, if you improve your staff, that helps me with my step. So why don't your IT requests go before mine? It helps with that battle of prioritization that every organization fights with who gets the limited IT hours within the organization.

Um, so, a lot of our systems and our processes are what are called tribal knowledge. I get a ton of kudos for getting things done once you peel back the onion and actually see how a process works and how convoluted and complexes. It's really amazing than anything gets done at all. But it's through this year, through sheer willpower of our employees.

And the knowledge that they have, you know, often when I get into a team, I joke with them that, more often than not, I uncover that a team doesn't really have a process, as we would think of it. I say they have a series of exceptions that over time, they've learned how to manage expertly. Dealing with that, but it all comes from experience.

You know, in terms of training and new employees coming on, it's very hard to knowledge transfer that information when employees change jobs, it's great for the employee but it wreaks havoc on the process. So we're trying to shift from managing people to what I call managing processes.

And listening to the process and, and having that be a main driver of making sure that that our value gets to our customers in the most efficient manner and kinda getting the people out of the way. So, so that's a little bit of background about our organization, and where we're at, and some of the challenges. And why I'm really excited that, you know, we have process mine in house, and we're starting to leverage it, and starting to see our processes come to life, and it's really, really exciting.

Yeah, that, that, you know, that's really great, and it's, I'm sure it resonates with a lot of people, certainly. It does, is, it does resonate with me. Because I have been, the situations are. Very, very, you know, many, very often, In many, many times. But, I think you're, you're probably not a process mining customer of your iPad, your RPA custody rights, or you're also using the RPA technology.

That's correct.

And the way that came about it was kind of interesting because I felt so passionately about wanting to let employees see the process, and understand how things were working, I fell in love with process mining through UI path, We had a separate team in our innovation area that was looking at RPA.

And we had decided during our review that we were, we were comfortable with going in separate directions. So they wanted RPA. I wanted process mining.

And we weren't committed to having a single vendor that we both had to agree to.

Now, after reviewing both areas through, you know, our in-depth study and due diligence.

We did both land on UI path, which was exciting, because that provides some synergies, but it really has best in breed products for both of those spaces. But, it wasn't just the, you know, the in house opportunity with UI path that we're looking at. Both of them were selected independently of the other, and we both landed with the right path, so, that's pretty exciting.

For me, process mining, and making those processes come to life, You know, historically, and previous, You have to do workshops to map those processes and understand the base flow.

You generally have to do des site observations, because employees don't always know what it is they do. They know they get it done, but when you ask them, and you sit with them, like, Oh, yeah, I do do that. I forgot about that.

So you usually have to do some gas side observations, and then there's team discussions to bring it all together. Make sure everybody's in agreement and has commitment that that is the process, and that's how it works.

And then once you have the process down, you switch to data exploration, finding out what systems, each step of the process lives, and where that data is, where the tables are, how they're structured.

Then, you have to go to ETL, and it's just this long drawn out process that can take 12 months ETL. And then once you have the data wrangled and cleaned up and available, then you have to figure out the visualizations. Put those visualizations back in front of the team, Have them do a review, and give you feedback.

There's back and forth to make sure that the data is correct, and accurate, and representative of their day, and means something to them, and then you have adoption of that. Then, then, you can finally get to insights, and then improvements. Start with a team.

You have many, many, many steps to go through to make sure that you actually get good, robust data that's meaningful. The team that they have, action, that they can take, that will actually move the needle on that. Most metrics that you see throughout an organization are lagging financial metrics that me, when I come in and do my job, and I sit at my desk for eight days, I can't move the needle on that. I don't work in sales. I don't work directly with customers.

So finding a metric and finding data that actually is reflective of the keys that I hit when I come into work every day, can be really, really challenging.

But once you find that, the engagement of the teams and the interests of the teams goes through the roof, that's fantastic. Thank you. Almost point. My complete presentation, because she wanted to present No. No, no, no, you're talking about how great process mining is, and what it does. But, you know, since this is a presentation, Or, this is a conference spot.

For me, I thought maybe it might be useful for our audience to give a short introduction of process, mining, what it actually is before we move on with our discussion. And then they also show you a real demo. So first, maybe some of the basics. And if you guys already know what process money is so lean back, there would be no use for you. But for everybody else, you know, this might be actually quite helpful.

So, well, you know, talking about about process, my name again, Rudolph .... I'm the VP and had Ambassador for process mining at UI Path. And as Neil already said, you know, we are not, basically we're not really a process mining company, so we are RPA company. So what is RP?

Know, we have these little nice bots you, I've also seen them sitting beside me or my bet, but we have these little boats, and they, you know, they basically emulate what people do. So they are reading screens, they are deciding what to do. They are interacting with applications. And they take away the robotics, repetitive work away from people. So people feel better.

People can really focus on their work, on creativity and solving problems, and not, you know, doing the day-to-day job, copying stuff from left to right.

And I can tell you, you know, and for that we need more transparency about the real processes.

And I can tell you that as a little boy, I was actually not very exciting about bots. You know, I was more into superheroes. I've always wanted to become a superhero. I always wanted to become Superman because this guy, you know, he was very strong. He could fly, and he had one very special superpower.

You know, I'm sure every everybody who's involved in process discovery could need, and that's the superpower of XYZ machine.

Of course, the knowledge, I realized, if I want to make some money, I need to pick a real job. So I joined UI path after university and I became another kind of superhero.

I actually started my career as a business console.

And as Neil said, if you ask people about processes, very often, they don't even know what the processes are. So a typical response to, you know, to a detailed questionable, says, it's really, honestly. I don't know. Often we do it. I didn't know how to, you know how long it takes. I don't know what their value stream has. No idea. We have a process documentation, of course. But this documentation is outdated at the moment. You print it.

And, you know, if you're a consultant and you get a response like this, New, What do we do?

We do workshops, and you have probably seen pictures like this before you in your life. You know. This is where you sit together with some people in the room. Everybody is tell you a different story, and at the end of the day, you have more processes that you have participants in your workshop.

Let you know, that was my life for a couple of years. And only then, in 2009, I traveled to the beautiful country of the Netherlands, and I met this gentleman I already mentioned to listen to us.

So he read is the guy who invented this technology.

So, what is process mining?

For me, it's still a little bit of magic. Of course, it's not magic.

It's a lot of math, but it's an auto magically automatically way to reconstruct and analyze business processes from data only.

The way it works is quite straightforward.

You know, because every transaction in an IT system is leaving some digital footprints behind and systems are storing this inflammations and database system. So, log files.

Now, imagine, if you collect all this information from all the different places, and you create this one file, we call it Event Log, with only three attributes, you can already visualize the processing.

So, the first, what we need is an activity like registered to find permanent or confirmation in this example. Now, if we take one single case and the case you know is, can be an invoice, could be a purchase, order can be a customer request, it depends on the process.

But if you sort the activities by the time, you're already get a process.

And now, imagine you do this for not only fall for one invoice, for example, but you're doing it for all the data. All the invoices in your system.

Then very quickly, you will realize there is not such a thing as a process, but there are many different ways how this process can be run.

And believe me as a consultant, not everything I have seen in my past was really efficient off. We'll frisk, and we need this transparency before we can decide, you know what to do. If we see a broken process. We need to fix it, if we see a process that is not efficient, and it can be automated.

We want to automate it, and we need this transparency because of all the, you know, assumptions and some facts about processes. So, a typical assumption is that people believe that the process is straightforward.

Yes, of course, You know, if we move the slider in one of our applications to something like 20%, you will see that what we sometimes called the happy path.

But if you move to slide after something like 40%, if you want to actually get a different picture, you know, you'd get the same picture like frightening from a good workshop because people know that not everything is straightforward. There are some loops, There are some kind of monies and so on.

Neil, what do we get if we move to slide up to 100%?

You know, we get what's called a Big Bowl of spaghetti. Of course, the Bowl of spaghetti. In our application looks different. Maybe more like this, but it's exactly the same process like on the left, but this time we have visualized every single variation of the same process in the same picture.

And guys, you see the difference, you know? This is what's, maybe what the CEO thinks the process is This is what people know, what the processes are. And that's the reality. And this level of transparency, for me, it's always, you know, like an X-ray.

So like an X-ray of processes.

So if it hurts or if you want to avoid in the future, we take the data, we make this picture literally, what a lot of them? And we get full transparency. Of course. Always limited to the digital footprints if there's no data.

Well, we're lost, But, if we have data, we provide full transparency. And, compared to my previous work, you know, as a consultant, our customers actually save a lot of time. They save a lot of money.

No, but, maybe that's the best thing about it. We can repeat this over and over again. So, we use process mining before. We are, before we automate processes to understand, you know, what should be automated, what could, what can be automated, and what needs to be fixed.

And after we automate the process, we use the same technology again for the monitoring of the impact of automation. And, you know, we have the entire platform for today. So that's the UI path platform for automation.

And as you can see, process mining's only one of the, one of the, who is in the discovery box over here.

So our core business really is building, managing, and running the different type of bots we have.

We always start with transparency, because only if you understand what the problem is, you can solve it, otherwise, you can't.

Well, so we have this entire platform.

And I would like to show you how this really works.

You know, afterwards, I will show you our super power of process mining.

But before we go there, Neil, I have a question for you.

You know, out of all the, the, the, the the process mining vendors, and we, you know there will be some more in the next few days. Why did you select process code? Process got those goals before. But know your iPad froze a sliding scale used to the name. Sorry.

Well, we really like the output.

So, it says, you just showed in the slides, we did a proof of value with UI path.

And we had exactly what you just showed. We had the bowl of spaghetti.

So, for our proof of value experiment that we did, we did our P2P process.

And it was a simple five step process in the minds of people that, you know, spend eight hours a day, five days a week doing that job when we actually turned on the x-ray machine and saw the invoices flow through there over a six month period through those basic five steps, there was 474 variations of workflows through those five different steps. I can tell you, I've seen worse, but that's that's a good number. For us, it was jaw dropping. So, we really liked the visualization. We like being able to see individual work items move through. You know, for us, we have different products that we offer to our member companies. So, we do energy and agronomic, so seed and other nutrients that you'd put in the soil or farmers. And we could see it byproduct division. We can see here by individual product we could see those invoices, color coded move through the system.

Um, also it, Rudy I'm not sure if you have a Lean six Sigma background or any formal training or, OK, yeah, even, really tell them the system, in terms of a Lean practitioner and, and the things that we want to know about a process to be able to do Lean analysis and six Sigma improvements. It checked all the boxes, they gave us all the data, the lead times, the cycle times, everything that we would need to be able to plug that data into our formulas and be able to see where we're at today and then plan for some improvements. We also like that it was non intrusive. So, some of the other product we looked at, were installed on, on end user monitors, and did more of a recording of processes to create the process mining in the process. Maps. Still provided some of the data and things like that, but we'd like to, the UI PaaS solution, use the audit tables, Use the historical data that was already there, present.

We own it, and we can plugin, know, for the parameter value, Like I said, we did six months. But we can go back even further than that, to see how our processes have been running, rather than starting recording on day one. And not really getting to those insights again for the six months.

So, that speed to delivery was really important to us, and we really liked that.

Um.

That's, that's why we chose amounts, Yeah. What do you think? Should we show the folks how it actually works, you know? So they maybe have a similar experience.

Yeah. I think the visual and the output you guys created, this is sleek, I like it.

Really seeing is believing. So, guys, again, I let me share my screen and just show you, you know, how ... at UI Path looks like. And I have to say that's the current version, so we're already working on a new version, which will be released later this year. This will be, you know, in a totally different co-operate design, or user interface matching the rest of the UI path platform.

Maybe at the end, I can give you a sneak preview. But this is what we have today. You know, and I want to spend the next, roughly 10 minutes, just to show you quickly, how we use process money first, to understand the process, how to discover RPA potential. How to monitor automation afterwards, After we automate the process, I will skip this part. But today. And then we look again, at, you know, how we can maybe react with RPA. So, this is much more than just, you know, process discovery, but, of course, here, I have selected some data from one of our customers already in Germany. So, we have to 6103 cases and a case, in our example. You can see it from the activities, like receive an invoice, processing, voice, and paid invoice. Know, every single cases an invoice, and this company, in the first quarter of 2019, they received 6103 invoices.

And, surprisingly, they figured out 165 different ways how this process can be run.

So, what I want to see what the process is, all I need to do is to click on process and the system will automatically, you know, automatically reconstruct process. But, the process, you'll see, you know, we have these boxes. We call them activities. We have the connections. We call them Edges. And because both of my slide Estonia such a 0% only, we see only the most common way. It'll vary, number one.

So, if I click on very number one, because everything is interactive, and if I would approve my selection, we would start to analyze only the 2313 invoices. But, you know, that's not what I want to do. I would like to see what happens. So, when I move the slider to the right, the system will actually indicates that, you know, most of the time, we go straight forward here from process invoice to the final check of invoice. But sometimes, to be more precise, but 1796 invoices, the first request data.

We check the contract, and only then we can do the ... of invoice.

The numbers representing the number of invoices. So, You know, of course, we can switch to other metrics, like, for example, value stream.

So when I click on average invoice amount, I will discover that this, you know, this different process, probably is, is because of a bigger invoice, you know, in our company, if we have another, if we have a big invoice, the certificate approval process.

And you know, I can do the same for four edges. So I might discover that apparently the fine encircled invoice is not find all the time because we see some loops.

And if I add some more exceptions, the system will indicate that we may even have some compliance issues because cases are bypassing the approval step.

Know, when we talk about automation, This is probably nothing you want to automate you don't want to automate a deviation or antibody like this. So here we can we can click on it You know? While we could click on it, we could go into the details We put C with the suppliers and all of that. Guys, this is, this is just standard process model. You will see this, you know, probably from the other vendors over the next few days. So, let's skip this, and let, let me show you some more advanced.

So, when I add more and more activities, I will discover, you know, maybe a short, fast track, and my process.

And if I add something like 100%, I can see all my activities.

So, if I add all my edges, well, I end up with the label spaghetti, You know, that 165 different variants of the same process in the same picture?

Cool, yeah, general, guys, I'm pretty usually RPA family, but my colleagues told me that automating, this mess would be a real challenge.

You know, so many variations, loops, compliance issues, whatever, you don't want to, or you don't want to automate our CEO. So, we need to focus on the right stuff, and how do we do that? Well, you know, we have one of the 65 different variants.

But let's say we focus only on the five most common variations of our process.

Well, yeah, I will say this picture looks so much better, doesn't it?

Yes, so, Yeah, right, so here, I know, I'm pretty sure we can do something about automation here, but where to start.

Well, we could, Of course, we can focus on the on the loops.

But, you know, we can also look at the bottleneck.

So when I switch my to my average throughput time, based on the dark color coding, I was very quickly realized that I have a significant bottleneck here between receiving voice and processing voice of nine days.

Know, the system automatically discover this as soon as, you know, it's a bottleneck, because it's so much longer than the 15 hours, or one point four seconds I see over here.

So, we should definitely do something about it, or, you know, consider to automate this part of the process.

And now we would, you know, we would start some other tools. We have like task mining, where we look into the front end applications and maybe some tasks capture. So, we have some more tools in our box.

But, today, we focus on the process Mugging, so I will skip the detailed work. I will skip the automation part.

And I want to show you how we can use process mining, really, to monitor the impact of maybe automation or maybe of, you know, every change you do.

So, when I click to another dashboard called Automation, we actually get the same process, but the pattern looks the same color is different, because the color is representing the different rates, are, different, you know, automation rate, so, we can see that already in the first quarter.

This part of the process received invoice was, fully automate it, you know, it's also Scan, bought the company already at, there was a payment interface, or pay invoice, or pay invoices fully automated.

But, you know, really the big bottleneck over here, you know, the nine days between the receiving voice and process invoice, a fully manual, 0% automation, and, we can see the cost. You know, we also had access to cost information from this company.

We can see that every processing of the invoice and the most frequent variant number one, was pretty expensive. $98, Jesus was probably more expensive than some of the invoices. But, again, you know, there was a clear expectation from the customer that this will heavily improve, once we start to automate the process.

And again, guys, know, that's where, why we decided that RPA process mining it, you know, The combination makes so much work, so much sense. Because once you understand that you have a problem like, oh, you know, a broken process, a broken arm, x-ray, it's not good enough, you need the treatment and that treatment. In this case, you know, it was automation.

So, we started to implement UI path in the second quarter.

And in the second quarter, you see the same process, but this time process invoice as a slightly different color, because by the end of the second quarter, already, 24% of all our processes were processed automatically.

You'll see the throughput time dropped a little bit from nine days to seven days, and also the cause dropped by roughly 20% from 98 to 81.

But in the third quarter, this gets really exciting, because, no, here, we can see process invoice is not fully automated by 100%.

And the super time, though, really drove from nine days to seven days another 34 minutes only.

And look at the cost.

No, we started off with 98, 81, and no less than $30.

So that's fantastic. But again, you know, there's this, this is standard process mining, or, no, this is not standard process, why does a little bit more. But, that's the power of Automation of RPA.

And find that with process mining, we have the capability not only to discover the process to, to analyze it, to see deviations, to see nobody's compliance issues, but we can also discover RPA potential.

And off that we have old methods process, we used our logs from our bots, actually, to get this information, and to show how successful the automation really is, no therapy and discovery for diagnosis and therapy, that's what it is. And that's one more thing.

And I say, that, it's so powerful you're out of everybody is understanding this level of insight, and how powerful it can be. But, you know, as Rudy said, there's different types of therapy.

And so the automation of that step is really powerful. And you can see, nine days, 34 minutes is huge. And we all know RPA is capable of that.

But one thing that we notice is, you know, our happy path was very small.

I think our, our happy path, our highest number of cases that went through one variant was less than 15%, because we just have so much variation in our process.

We all know variation is bad in and of itself.

So, some of it is, if you're going to automate a step in a process, how do you get more work to go through that step and not go around and have so many different variants and different ways of doing it. So, we could tackle some of those edges, as well as the activity, for automation and say, not only do we want to automate this stuff, but there's only one thousand cases that go through that step. We want to get 6000 cases going through that stuff.

So, the, the automation that we do put in place becomes even more powerful, because it doesn't just improve the cases that go through that step, we can get more cases in.

So, there's kinda two ways to attack it, and different therapies and different opportunities to, to leverage improvements in different areas, right. And to your point, I fully agree for me, as a consultant. They are always multiple ways how to tackle a problem. You know, it's or some people call it PPT. So people, process and technology. Clearly, you know, RPA is the technology part. You just talked about the process part. But it's also the people part where, you know, once you understand who does what, you can ask, why they do it. You can talk to the people who can think about training for them, really, to help them to do a better job, Right?

Andrew: The people part, and the culture part is often the most difficult.

Getting people to change behavior, but when they see when I showed that spaghetti bowl to the team, that did it, know their jaws drop, just like the CEO room, that really, that's what we do the best that we have to deal with. And they were more validated that they knew that they work really hard. But to see it and visualize it, that really gets them onboard with, something's gotta change.

We need to start doing something different, and they become far more change, malleable, for all your change management efforts when you're trying to work on process and technology.

Absolutely. Hey, Neil, I would like to show the third part, you know, the reactive power, because this is, at least as powerful. You know, we have discovered the issues, we have fixed the issues with automation, but, unfortunately, no, not everything will be. We will be Right from the start node on The boats, and other people are perfect. So mistakes will happen. And I want to show you how we actually checking the challenges with our platform as well.

Well, that I have a bookmark here or really for the last day of my data and so this is the 30th of October 2019 and we can see on this day we had only 14 invoices but one invoice for whatever reason we managed to bypass the mandatory approval.

So, yeah, we need to fix this problem. How can we do that? You know, of course, when I click on the path, I get to filter with my selection, I've only one case. I go into the details to better understand what happened, and when I click, you know the case, I can see.

Well, this was actually fully automated scanning about processing, both control bot, but the approval step is missing.

It's only a small, it was 450 bucks, and Rebecca ... responsible.

And, you know, normally, we would detect this deviation or this antibody every time we process the data. So most of our customers does it once tonight, but we only few minutes left. So we cannot wait for it until tomorrow.

So, for that, I edit another dashboard here.

It's called Depiction Center, and, you know, this button normally doesn't exist, but if I push this button, know this, but nobody happens automatically in the backend every time we are processing the data. But here, now, this case was sent over to another part of our UI path Platform, actually, the action center.

And the action center, if I switch over, is, you know, it's like my to-do list. So, you see it? It's, I have no unassigned cases.

But when I refresh my my page and hopefully it works you know, that's one of the stuffs online maybe I need to go to ending no, must be on the side.

Yes, So, we can see, you know that a few seconds ago, this one case was actually created an invoice, approve this missing.

So we need to do something about it.

And while still loading the details, I can see, you know, in my case, from process mining, Rebecca, off the 450 bucks.

And if you guys workshop, they have a document management system, we can, as a button where you click on, OK, show me this invoice, and this is just a dummy for demo purposes. But we see the invoice. We say, OK, that's good.

We should definitely pay then one, We go back, and now after I assign this case to myself, I can for example, click on, Come On, give me the button system, seems to be very busy, OK. I can click on open in sap, then what would happen? You don't want to say people would come up and I would say just invoice after being locked in and I couldn't prove it manually.

But guys, though, this is exactly the five minutes of work we want to say for everybody, for thousands of people, this and other test.

So instead of clicking and doing this manually, I can actually click on Approve and then one of my bots will take over.

Well, on my behalf, go into sap with approve this invoice. And all I need to do is just to confirm that this case is completed.

Can I come back to a beautiful, empty to-do list again?

So, guys, that are all perfectly on time, because we want to spend some time on Q and A So what I tried to show you was really know how we use process modeling today. First, to discover the real processes, to understand inefficiency, to understand risk, to Discover RPA potential. And then, after we have automated process, to monitor the impact of RPA, who have reliable numbers on how well we're doing.

And if we are not doing perfectly well, how to use this technology, again, to react on any type of issues, and fixing the problem that, you know, the final check of invoice, or the approval step, was missing. With that, thank you so much.

And, yeah, let's move on.

Nice.

Thank you very much.

Let me re join you on screen.

What a fantastic presentation. Thank you so much for sharing your insights and knowledge. There's been a large number of questions coming in.

So it would assume, I would assume from that, that the audience has enjoyed this presentation as much as I have.

And Grace ****, who is your colleague, Rudy behind you, has been I bring answering as many as possible behind the scenes to price graceful her efforts as part of this process and really sort of move swiftly as possible into the first kapo, if I may, um, I'm really set the tone for what we're doing as part of this process.

So I would ask that we start with a question from Michelle Catto, and please forgive my friend show Michelle. Not as good as it should be. Jean-michele says I work within the Context Center environment evolving process mapping projects.

Would it be feasible to draw furnace IP for every process that is mapped out so that the SOP document could be used for training purposes and as a follow-up to that are there any other alternatives apart from creating an ESOP document? So I don't know who wants to take that first, guys.

You look into that, of course. No. Mapping out every single process and say P is, Oh, wow, that's a big challenge.

That's probably a task that will take you a couple of, well, maybe not weeks, maybe even months or years. You know, and not every process is really, I would say, well, not every sap process is suitable for a process money because you need to case. You need some some activities that needs to be some variation. the process because only then it makes sense to look into it and to know what I learned in the last 10 years. The biggest challenge really is to get the right data and to get the data right. So today, we know process, you know, we, we know pretty well.

But sap actually is a system with up to 170 or 200,000 database team. You know, this is insane. Probably. only journalists can come up with a system like this, but really, my apologies.

rather than an S&P was SOP, O S O P.

As I say. That was entitled my pronunciation. OK, no problem, so don't forget arching, I said sap is great, I love Instagram. No, just kidding.

Know, you know, basically, so Jean-michele, sorry, I don't know SOP. So if you'd like to reach out to me, Rudy at UI path dot com, where you'll find me on LinkedIn, and then we can have a discussion about this. But basically every system that leaves digital footprints behind is suitable for process mining. And if not for process mining, we can use our task mining capability. That's a different topic, you know? That's something where we create our own event log from absorbing the screen's a little bit. What was also talking about in the beginning? So we have no process money and task learning, really for the comprehensive end to end view on processes, but definitely, you know, contact center along our customers that you aren't Bath.

We have probably hundreds of contact centers, because this is really where, also, automation makes a lot of sense.

So I'm sure we have some, some use cases. So please don't hesitate to reach out to me, and I never have a detailed discussion.

But basically, yes, we can Fantastic.

And Rudy, would, you, just said, is exactly how we're coming at it. So, a process mining is great for visualizing your process, and understanding the flow, and getting those right matrix, so that you can look at those metrics every day. You know, we use the monitoring tools that you just demo to do, like, daily huddles and things like that, so we can see if our process and improving them, be talking about our process on a regular basis. But, the other tool you just mentioned, task capture.

That is the kind of the screen recorder and it records each step that you go through and it puts the output into a Word doc as one of the options. But the task capture tool through UI path is what we use to create our SOP documentation.

Once we have, we have test capture, and now with our new latest version, 21.4 coming out next next month, we will also have task mining. So, that's the, you know, like to avoid it. Like, the next version of ..., where you do not record only one execution like the perfect process, but you can, you know, you can start this recording for, like, a week observing what people are doing. And then, with some very fancy algorithms, we recreate the work in the front end from it. So, it's really sophisticated and history the next level of understanding.

So, it's a combination of process modeling, that that really gives you the full insights in your process.

Those are nice because they take actual screenshots of the screens that you're on. So, to train a new user to hand somebody, they will be able to see the layout and where they're going and outlines navigation.

It's really not personal.

And now the next question specifically for you, and Brad Thompson, asks, can you talk about the value you generated from your proof of value?

Yeah, so the proof of value.

Well, like I said, we used the P2P process. And what we did was, we started attacking those edges.

So, we brought in the tool, we started looking at it and talking to the team, and understanding why so many of those different variants happen.

It was a great opportunity to institute some training programs, to introduce some Lean concepts, and some, you know, explain to them why variation is bad in and of itself, just so that they knew that not only is there a lot of variance. And they understand now why their day is so chaotic and hectic.

But understanding why, just doing it differently from person to person, is a big deal and doing it differently from day to day is a big deal.

And just knowing that, if I'm more consistent in the way that I do my work, processors can run more smooth. Added a big value and over just a short amount of time, we were able to eliminate over 100 of those variants just by bringing awareness to it. We didn't actually do any massive implementations or projects or anything like that. But just people knowing that.

I need to go steps: A, B, C, because Steps A, B, and C is the right sequence of work.

And even though I didn't eat my breakfast this morning and I want to do Vika, just because I feel like it, I need to do A, B, C, and just, same sequence of steps, getting two neighbors to work together. So, a lot of areas, you know, Billy gets hired, and Susie trains Bailey, and then six years later, Billy trains, Tommy.

The process has changed over time because you're getting trained by different people and getting trained the way they do it, that wreaks havoc. And that causes a lot of those variants to. So we started to tackle some of those systemic issues with those safety and with standard procedures to get standard work in place. And then we can see the output and by rewriting process improvements.

Fantastic. Thank you for that. Guys, I won't say in advance. There's no way we're going to have an opportunity to answer all the queries. I know that Rudy and his team will follow up with you afterwards so please, do expect that contract.

However, this very interesting one next, which I'll go to and most, and thank you very much for offering this one. I think I'll direct this one to you, Rudy. I'm because you probably would know the answer this is the way to add manually processes which are not extracted from the ERP data.

Well, we have two distinguish between manually executed processes learning like signing of invoices on PayPal, you know, approving invoices by with a signature on paper.

Then there's almost no way to do it in a good way because, you know, no matter, of course, we can generate these. These events artificially just adding the data. But you never know when it was really done.

There is no time-stamp if your signature, if you are lucky, you will have a date, but you'll have no other time and so on. And so, you will never know how long it took.

So, honestly, there's not a good way to act manual tasks, like, you know, signing of paper.

If we talk about manual tasks like tasks done with a different system, or even not in the, so maybe something like e-mail.

Then, again, we have this task mining capability where we try to create event logs from applications that do not create any digit footprints like Excel, like e-mail shows to better understand the combination of trends, Actual backend systems and front end systems, really, to get the full transparency.

So that's a different topic.

So paper, not really for the, for the other stuff. Yes. It's a bit harder, but it's possible.

Fantastic. Well, I hope that to provide you with your answer. I've got two final questions here.

one is not so much a question, but a request for a sneak peek of the revised process mining UI from Daniel Viner.

He was obviously looking forward to seeing this and we will with the rudy's permission to end on that. But just, before we go there, Rudy, I go. Can I, ... has a question how can value stream mapping be leveraged with process mining is in parallel or vice versa. Maybe Neil Rudy, Neil. Do you use that in terms of what you're currently using for, what's the best way?

So we start with process mining, we start with the data and understand where they're at. And then we, we validate the value stream map using the data that was produced. So there is still some back and forth with the business unit to generate that value stream map.

But it's generally easier to pull the data out first.

Fantastic, and now, Rudy, if we could get a sneak peek, I think people were very pleased.

Yes, of course. And also what you see is the current UI. It's the old process called user interface, of course. And we are working on this really seamless integration of process mining in the entire UI path platform, including AI Center, including Task Monday, including actions and everything. So, when I go over here, so what you'll see is really the, the very first alpha version. It will be released later this year. But here, you know, you see the, it's, we have the typical UI path task bar.

Over here with data services task binding, the studio, where you can create your balls, AI center, and here we have the processors, you know, I can I can work with with the filters hit or removing or adding some new case owners, and I didn't refresh this page for a long time. So OK, but slow.

But, you know, this really is the Like the new user interface. You'll see, of course, is the very, very first version.

But you see, part of it is already here, and we are, our developers are working on much more stuff, but most importantly, ... is the integration with the rest of the platform. So if you want to use AI functionality, it's only one click away. If you want to move this process, take this process and automate. Just a studio is just 1, 1 step away. If you want to go, let's say, we've invoiced and you want to better understand what people are doing, the task mining's, just one click away.

So, guys, this is going to change the way you are looking at processes and automation dramatically.

And again, stay tuned. It will be here later this year.

Thank you so much.

Bruno, Rudy, Neil, thank you so much, An absolute tour de force and a great opening session for the events. Thank you very much, Neil, for interrupting your vacation. Please apologized profusely to you'll find before us. Really, thanks for coming in. And meeting, again, much appreciated. And, Grace, thank you very much for your hard efforts behind the scene.

That's it, guys. Thank you very much. Look forward to catching up with you in due course.

Take care.

Thank you.

Take care. Bye, bye.

Ladies and gentlemen, we will be back up at the top of the hour with the second session of the event. And I'm very pleased to be able to say that as the next part of the process, as laid off from a fabulous opening session, we will lead directly from that into Ramaswamy filing Gu from Avaya. He is the head of Digital Automation and Enterprise Architecture, and he's going to be looking at process mining, and the next generation enterprise. That is a session you do not want to miss. I look forward to seeing you at the top of the hour, and we will take a quick break now. Look forward to seeing you back. Take care. Bye-bye.