Unlocking Operational Insights: Process Mining at MOL Group with Andras Katko

Hey everyone, you're listening to the "Mining Your Business" podcast, a show all about process-mining data science and advanced business analytics. Joining me as always is my good friend and colleague Jakob. How you doing today? I'm doing fantastic, Patrick, thanks. Let me ask you, do you know what a turnaround of refineries for oil and gas companies actually means? A little bit. Okay, well, dear listeners, if you're anything like me and that has absolutely no idea what that meant, be sure to stick around to hear all about how process-mining is shedding light on plant maintenance and beyond in the oil and gas industry. Let's get into it. "Mining your business show just keeps going and shows no signs of stopping." But let's also thanks to you, dear listeners, and the fantastic community that process-mining has. Now, I don't have much of a comparison with other IT fields because well, I only worked in one IT field, but I still think that process-mining community is the best out there. Right, Patrick, I know. Also, today's guest, who was also introduced to us through a listener, a fellow process-mining enthusiast, and well, interestingly, also a previous guest of our show, Didrik Badon-Gibin from Sylvanis So, Didrik, if you're listening to this, thank you. Thank you very much for connecting us with Andras. And well, actually, I'll get to Andras in a second, but also for you listeners, if you have any further recommendations, or if you know about people who would fit in our show perfectly, do not hesitate and reach out to us on LinkedIn, or just send us an email on mining your business podcast at gmail.com and send us your recommendations because there's never enough of interesting people with an experience from process-mining or business process management industry who we could interview. But let's get to the start of today's episode, and that is Andras Katko. Andras, welcome to our show. Thank you. It's a real pleasure to have you here, and again, someone with a pedigree that you already bring in, it's for me, this is always the best learning experience. And currently, on your LinkedIn page, if a listener goes to your profile and checks it out, it says that you are a head of group, corporate IT, etmol group. And my favorite question at the beginning is, what does it mean? It's a little bit outdated, and I'm not as, you know, up to date there. I have a name as a head of IT group back office. It's even more interesting. So I'm working for MOL, which is an Eastern European oil and gas company, and I'm reporting to the head of the IT group. And the back office is really in short three different areas. One is the CTO, so technology, rather, the CDO, the data analytics, and the third one is the core systems. Though core systems are usually which is beyond any domains, as you know, in oil and gas, we have upstream for exploration, downstream for processing, and we have retail, and obviously the functional areas. So wherever you need to have a cross-domain expertise or services, it belongs to there. So ERP is one of them robotics, another one middleware, and all the integration is in my area. So the back office is really, I have my biggest customer is the shared services, the GBS, which provides all the services. And I have very less contact with the end users, as such because it's done by the front office, which the group means they are doing the all the account management, the strategy, the SOS, as well as the strategy, the the financing of the projects, the tactical things that we do to quality assurance of the delivery of the project, as well as, you know, all the governance is actually in the group level. And we have a GBS, which is a few times larger than the group itself, delivering the services for the whole group, in we have about 30 countries. Now, before we get into what you're currently doing, can you tell me a little bit more about what happened before you became the head of group corporate IT back office, and kind of summarize how you got to where you are right now? How far should I go back? Give us the main points. We like to always build up this expectation also for the guests, you know, or for the listeners who might have similar aspirations eventually. Okay, so actually, I have an interesting career in certain sense. In the early 1980s, I came across, I was in the U.S. and I realized that IT and computer science is so important and I jumped on it. Eventually, I was, I left Hungary and I lived in Canada, in Asia, in Singapore, as well as in Hong Kong, and later I was called back to Hungary for some larger projects. And I tried out myself in multiple industries, in transportation, business, in manufacturing, logistics, and a majority of my career was in finance. So I was a CIO for Budapest Bank, which was a G-Money Bank. Later on, I was the CIO for Keller and later I enjoyed more and I first of all was the functional areas and eventually my rule grew up and became the back office head. Now, I have a question regarding your current position and you mentioned that there is a lot of subparts that you manage the data, you the ERPs, middleware layers, and so on. It sounds like a lot of digital transformation topics. I would almost call it this way because they all come under this roof of how you can run the process efficiently and what to use, what to not use. And my question actually goes into this decisioning. Are you also the person who decides about the tools or the software that the group as a whole would eventually use? Like, you know, if you are deciding between, I guess it's not a really decision right now, but between like SAP or Oracle or whether you would use this business or business intelligence tool or the other. Is this also in your competence? Yeah, definitely. I have the recommendation, but obviously, since in our case most of the cost is actually paid by the business, obviously there is a certain type of agreement on that functionality, the cost, and we have we are very sensitive on cost increases and we have to put together that the the increase is actually benefiting as much as, you know, business can bear. So in case of let's say choosing the the tool, for example, currently we are using Azure CNAPs for an X and we did the decision and as well as for process mining we chose Salonis, especially because at that time there was very little competition, the competition was significantly less functionality and less not reliable, but definitely we had less confidence in those and we feel very comfortable the the road map that alone is put together. The RP was already done, that's already in our realm for over 20 years. Yeah, yeah. I mean, it's process planning in general, still young discipline, so there's still going to be a lot of competition coming up, but what I wanted to also ask as a follow-up question is how does this how do you generally manage all of this because I assume that there are hundreds of tools that you can first of all have, you know, using in mall, but then still quite a bit of tools that you eventually have to manage under one roof and would there be any silver bullets for people who are in similar position to yourself who what you can tell them as an for again, for inspiration or some best practices on how can they manage the tools within one company so that you get the benefits of the tools, but don't also on the other hand overload the users with, you know, dozens of different things that can be used often very similarly. Let's take step by step. There are certain things that we already started to do. First of all, that proof-of-concept type of thing. So we do have certain tests on the what do we really need, what type of functional, non-functional requirements we have, and we have a short list and we usually do some parallel testing of multiple tools. So let's say that currently we are using certain tools for middleware and integration, and we did POC with three different tools and eventually we chose MuleSoft. And obviously now we put together the how do we retire the old one, how many years it will take to be replaced, and you know the saying that if you change something more than 20%, you should actually do it right. So probably we will have some very opportunistic approach, how we'll actually roll out certain tools. So obviously it's a truly, truly green in terms of, you know, we never used it before, like a process mining, that was slightly different. But I would say that we go with the Azure Synapse as well, we did certain things and we did testing or test automation. We had obviously we were looking at SAP as well as known SAP applications, how we can actually not only automate the testing but automate the evaluation of the test results. So it takes time and especially in the company with this size. Yeah, I can imagine. And that actually brings me to the next question, which is the introduction of Selon, it's a process mining in general, and you mentioned that for standard things such as a middleware or this data transformation and so on, those are generally established markets with established vendors and you go into POC, you probably try out multiple things, see what works, see what doesn't and so on. But what do you do when somebody comes to you and introduce you something new? In that case, it would be Selon is a few years back, but you know, this is a rapidly growing industry, not only process mining but everything in technology. How do you even decide whether you go for a certain technology that has a very little trek record and I'm sure there will be a bunch of people coming to you now and let's try this AI helper or whatever, you know, AI is very sexy word these days and I want to know what goes into this decisioning for an organization at the size of mall and how do you even handle this influx of new technologies? Look, it's an interesting question. Obviously there are certain questions where we are approaching our potential, you know, suppliers, but we do usually have a multi-year plan what we are planning to do, so there is a 2030 strategy for business and we see what type of skills, what type of infrastructure needs to have. So in that case, we are not, you know, a startup type of thing that suddenly we do certain things, but we do relatively conservatively in this term, you know, we are not actively looking for and we have every year at the beginning of the year we have a vendor meetings and we are actually telling them what is our plan for this year and what is the longer term and we frankly tell them don't bother certain things, we are going out, we are looking at so our, you know, guys are, you know, we don't want them to leave us because we are not technically advanced but we are, we have to think about, you know, how important to have a 724 availability, we cannot jump to any new technology without, you know, really considering the impact and my security now is incredibly important. So we do fast, we do POCs, we learn, we go out and learn, but we are not always jumping immediately but rather than when the certain product reached a certain level of maturity. Let's assume that we have a roll out of some new initiative, let it be process mining right and we, let's assume it's a successful POC, like what are some of the planning horizons that we're looking at from like first being, like first hearing about the tool to actual implementation, like how long does this take in a place like more? How long does it take? Well, yeah, like on average, if you had to guess, like just from the introduction to it to finally doing the proof of concept, impact analysis and all these things until it's, we're actually looking at a roll out. Now, because the, in the case of, you know, trying out something new, we have dedicated budget. In that case, we have certain sandbox where we can play with. It might not yet any, any future relevance in our environment. However, we do pass certain things and obviously we are talking to the customers as well. So our internal customers, let's say about drones or, you know, voice recognition or video analysis and we do certain, you know, type of POCs. And again, since we are providing some backgrounds and we are, we have a certain framework, what type of technology we are able to easily embrace. We, we limit in certain sense the choices, but still open it. And we have a good, you know, relationship of the business. They want to, they have currently a dedicated team, it's called 3D, downstream digital development, or that's the 3Ds. And so they put up a plan for downstream, at least 20, 30. What is these stages that they would like to reach by every other year? They have 24, 26, 28 and 30. And obviously they are, they realize that, you know, changing SAP dramatically, it takes time. And, and we are helping them to, first of all, to figure out how to do that. And where we can actually utilize certain things that it's already available. But we saw that a lot of features are available and we are not using it. Or we are using wrongly. Process mining is mostly about, you know, what we do, inefficiently or under utilizing. And, and obviously we could identify those places where we could jump in and turn on certain features. And speaking of process mining, how did and roll out of this technology look in your company? Who did come up with the, with the technology in the first place? And how did you identify the two actually needed? And what is the general idea behind using it in the first place? Look, it was, we had a three day POC with Sloanies. So, and we looked at purchase to pay. So, we did the three day exercise and eventually we showed to top management, you know, the results. And it was very impressive that it, it was in 2019 when we did this. And we took the, the whole, all the data from 2018 for, for the whole company. And we analyzed it. And there was certain question about automation. And it was interactively in five minutes, we were able to drill down and identify certain discrepancies or non-conformance. It's mainly that we as a company wanted to maintain all the catalog for the products. Meanwhile, we never updated. So, all the orders were supposedly automatic. It went through, but since it was outdated data, it was always rejected. And instead of, you know, changing the catalog, we always changed the PO. And so, it was always inefficient. And the people learned very quickly that to take a different, you know, route, route, and not going to the catalog, but going to another means. And we saw that if you go to the catalog, it took 25 days, if you went to other way to call the five days. And it was very impressive. And the group CEO was there. And he stood behind and he gave a green light, more further. And that was a big quick, I mean, kick to the, to try it out in, in the fall of India 2019. Now, when you said you rolled it out for, you tried it for the entire company, can we talk a little bit about what that means for mall, right? Because I mean, oil and gas, I mean, there's fingers like in every pond all the way from like the planning of the well-dating all the way to actually driving it to the refining it and all the way to the consumer at the gas station. It's like, what does the entire company mean for mall? I mean, we are focusing on the three largest market, where we are market leaders. So it's Croatia, Slovakia, Hungary, and obviously in some of, we have three different ERPs for Hungry and Slovakia. And now for the most of the countries, but Croatia is in two systems. They are pretty much side by side. And all the data is actually fetched to the Salonis. And always when we are already completely a certain part of the developments, we bring the inner part, which is the Croatian company that we are managing, but we are not a minority, but we are only 49 percent and it's large because one of them, the original SAP is from in 2003, maybe the next one is 10 years younger. And obviously the processes, the data structure, and those are different. And the obviously some of the statuses are interpreted differently. So there are a lot of not only the combining the data, but to actually harmonizing the data. It's usually the, it's a difficult, but not stopping us to quickly adapt and bring some data into the Salonis and the analysis. Andras, what if you are an oil and industry company, oil and gas company, what are some of the processes that you are tackling that you want to actually implement within process mining to get an insight in your dimension that you started with purchasing or purchase the pay process. What were the other processes that followed and also why did you choose at the beginning purchase to pay? It's mostly because there was a new management in the purchasing department, so they have a lot of ideas how to, what they want to achieve. And this POC was very, we learned a lot, we recognized certain opportunities. And so we started with purchase to pay, but then following year when we started to really start it in 2019 as a POC, 2020, we did two projects, parallel the plant maintenance and the purchase to pay. So the purchase to pay, we started to roll it out further, deeper KPIs and more analysis. And we replaced our previous plant maintenance and service management software and we came up with the new processes. And so the management thought that it would be good to track the progress and the differences compared to the previous version of the process and what we have in place. And it was incredible how to track the adaptation. And it was a daily stand-up meetings at the beginning and looked at that H.O. you are not doing the right thing, or that you are recognized that are certain things that from the past still capped and so they will be able to really quickly enforce certain ideas. And meanwhile, measuring the performance of the process and they were able to change it because they saw that if you are in ivory tower and you come up with a certain process design and you see that there are certain corners that you should actually cut, you might change the whole process. And you were able to as well to compare different locations and you were able to benchmark what happened with the maintenance in Slovakia versus in Hungary for this similar environment. And you learned as well. Speaking of maintenance, what are some of the use cases that you would look into in the oil and gas industry? And also, if you could explain us because honestly, I do have a little experience with oil and gas industry because I also had a customer a few years back. So I know some of the wording such as turnaround and so on. But let's assume our listeners probably hear first for the first time that there is something like a plant maintenance in an oil and gas industry. Okay, so you can imagine that beside we are manufacturing either gasoline or some byproducts later on, you know, petrochemicals or lubricants. All of them are, you know, large factories and those are really dangerous in terms of, you know, if you if you screw up something that you can screw up very badly. And it's it's health and, you know, you can have a spill, you can have a, you know, accident, very dangerous thing. So the first and most important thing to maintain the not only the availability of the of the whole manufacturing, but to keep it safe and and the truth is that you have predictive and preventive maintenance and we have emergency maintenance. We keep it definitely different. Obviously, if there is something happen, we discover today, we need to react very quickly. Those are rather the emergency one. But we have to plan the maintenance we had because a lot of cases we have especially you know, equipments that was, you know, made to this particular site and it might take, you know, a year to actually redo it. So we need it time. And one of the case was that we usually realized that we could start it late, the planning. And we left out activities, raw material spare parts. We didn't have enough, you know, suppliers to support us in this work. And the company itself, so the service, so what happened is we have a business, which is managing the manufacturing. They have the engineers, they are looking at how the equipments are performing. They have to plan and create a strategy. What is the next portion of the factory, which site, which particular equipment group has to be maintained. So they are already creating 2024 maintenance. And we have to start to plan those things. We realized that we had issues with data, we had issues with the norms. You can imagine that we have people from materials, we have task lists, we need more kind of activities. Those are very badly maintained in terms of information. So the first thing is to actually create a strategy, scoping, planning the activities. And what we are doing now is the planning, we are trying to use templates for certain activities. And we are always enriching those, so we keep norms for activities, such as, let's say, the welding or scuffling. So we know that type of work, it will take X amount of time and how many times you have to repeat those things. What's follow one step on the other. So we are creating those knowledge in our databases. So we do the planning, so we can figure out the lead time, the cost. And so we can better tender some of the activities. Because a majority of the activities are performed by external suppliers. And our team is doing the quality assurance, the reviews. Obviously they are accepting the definition. Now in plant maintenance for those of us unaware of what goes into that or what how process mining itself fits into this picture. Like what are we actually tracking as a case here in this type of process? Okay, so first of all, in our case, as I said, there is a business which is using plant maintenance. So they do the planning of the turnaround, turnaround is a shutdown of a certain portion of the manufacturing site. So they plan the scope of the turnaround. They are planning the activities and they are handing over to our single service company. The single service company is actually collecting all these plans. And they shake it up and they say, okay, for this period of 5 minutes, so much building. So they're actually collecting and they create a different site, which is the service order that they are tendering all the activities, many tendering, let's say, calling some of the pre-qualified subcontractors. And they are sending out and getting the order. So what we have here is we have the plant maintenance, we have a service management, we have a planning tool to have a proper set of activities. We have the order materials. So we are using NM from SAP, for example ordering raw materials or semi-finished equipments and we have a multi-event log. So we are able to look at activities by itself. But in fact, you are able to track the flow and the readiness of certain things. So prior to that, we were not able to say how far and how well we are in a certain turnaround. Are we ready? Did we send out? Are we already agreed? Business and the service, single service company, whether the scope and the planning is complete, correct and doable? And whether it's already done, and after that we can track whether all the necessary tendering or purchasing is already started. Are we expecting that everything comes in on time? That we can start the work, let's say, first of July. And later on, we are able to feed back the planned versus the actual. And obviously, after the execution, we have a post turnaround phase where we go back and we do the lesson learned exercise. And you are able to see whether all the activities are done on time. Well, so the quality, the performance, the completeness is already there. First of all, this sounds like a very complex process to me. And since I do have a little experience with the build of materials and with turnarounds and everything, I know from data perspective, it's really, really challenging. Let's put it this way. But my question would be, have you noticed certain improvements once you introduce this process planning technology and how did you exactly benefit? Look, you have to learn that you have to start you have to start walking before you can run. Yes. So our experience is that, first of all, it was shocking to see the variances. And it was as well very useful to learn how the process is actually done versus what was expected or what was the theoretical way of doing. And the second one, we learn that the first half a year year, we needed to have the master data, all the data cleaned up, even be recognized that some of the events were not done. Let's give you an example from the purchase to pay. So we looked at the beginning, you know, the peos that were opened for more than a year. And it was shocking, large number. And what happened is, it was nothing wrong, but we did not close the peos properly. So we ordered a hundred something, they delivered a hundred, but we only took over 98. So we closed and we paid 98, but we never closed down the peo with the two. So the system gave you a very false positive based on wrong data. So the first thing was to really put the processes at least into a normal more than you started to clean up the data and you were able to see the real problems and not only fate problems. And so I think the point was that before we were able to realize any kind of benefit, we had to put the system in place that it's reliable and the data was not questionable. And now we reached the point when we say that the data is coming from process mining, it's already accepted. And if by anyone would like to challenge it, the tool gives you a lot of way of presenting the data in different angles, you know, by customer, by brother groups or whatever. And you are able to give an argument is not about the data, it's about the process or responsibilities, the compliance. So it's really something that we are looking at, you know, give a management oversight, root co-sanalysis, process analysis, we can prove that the standardization give a lot of benefit, especially when you have multiple sites and you are able to compare those. And obviously you are able to identify some of the benefit of automation. But overall, listening to you and you said that as a small group, you started with process mining, already 2019, it's quite a long journey, isn't it? Like to get from the point, okay, here's an interesting insight from the proof of concept to fully establishing the way of working, trusting the data, something that you mentioned as well, which I believe is huge to properly using it and maybe ingraining it on daily basis into your work. Look, we talked about only purchase to pay and maintenance, but we have a very good and very extensive use of for order to cash. We have for accounts payable, we did it for IT service management. And we just started to use it for logistics, the railway logistics. So we started with the POC, on rail. We have, we own about, we are managing over 4,000 baggains in Slovakia and in Hungary. And obviously we are each of the sites, the refineries, the one in Bratislava, the other one is near the Sassalon, but they are managing daily to 250 baggains a day. And obviously you have to know where these baggains are, are we having sufficient that we can always fill up, whether we have, you know, there is maintenance there as well. So you have to be very aware of what you can plan with. And so we just started and very exciting moments of, you know, discoveries, how the system works. Why do we do that? Do we have the right answer whether we have a good explanation? Why do we, why do we do certain things? And you ask for on broad weather, we are creating the values, but we, I think we already talked about this with you at Jakob, that having a turnaround planned for let's say two weeks. Obviously we are prepared to have a stop of production for two weeks, but if we are actually for some reason, we are over, you know, if we are not finishing on time, that would cost not only reputation, but a lot of money. So in fact, what we are doing now is to avoid any kind of unforeseen, you know, activities or we are not missing out anything. So we are, we can sleep overnight, you know, that we need whatever necessary to have an on time delivery of that project, which might cost you a lot of money and you can imagine sometimes the reputation as well. Now with such a powerful x-ray of your business like process mining and with oil and gas having such a big focus on safety with obvious ramifications, is it a big initiative at Mall Group to use Salonas as a way to check for compliance or anomalies in the process that could lead to some sort of risk. Could you phrase it a little bit different so it will be? Yeah, of course, of course. I mean, with safety in mind at Mall Group is compliance and area of concern that you were looking at Salonas to analyze compliance or deviations in the process that could potentially have knock-on effects that could increase risk. Probably, I can tell you that those areas where we are using, they are totally hooked and they are really fan of it and they would like to maintain and increase the use of Salonas. Personally, I am not participating day to day. I am probably the, you know, the missionary and I'm trying to sell the idea and I'm very much involved as much of my time let me to participate in those meetings. So whether we have a certain risk or knock-off, you know, effect, I cannot tell you but probably I can ask some anecdotes from the others, whether there were certain things that we discovered but definitely we have, like in the data just when it's the rail, we seen a lot of activities. For example, that they already filled the wagon and they changed the purchase order. So suddenly they have to reshuffle the train itself. So it was never really visible for those salespeople that what is the impact if you are changing a certain order or a certain delivery after a certain time period. And we were able to show that, guys, this actually creates two, three day delays which is cost money, the rail, cost the customer because probably they would be very surprised and so on and so on. So if you are thinking such a thing that there is certain things that we can order to realize and definitely from this multi-vent loop, you are able to see some of the impacts. Definitely we have such a thing and it's trivial. So we have to go back to the salespeople and we have to explain to them what is the consequences. Or for example, another one, we have an application or mobile as well as on the internet that you can already feed your order electronically. And we are monitoring that and the engine is giving advises to the salespeople that you should approach a certain customer because they reach the frequency or the volume that it would be cheaper for them as well for us and it will be quicker the services if they change from the salesperson to an application and it would be better for them because they can track the estimated time of arrival, the completeness of the order. So there are certain things that we do based on the available data and we try to generate new leads for example. Now one of those things that I like hearing is when a company introduces process mining and the adoption is going well like in your words you said that the users are loving it. Now with them, I'm sure a lot of people would like to hear this, like is there anything that you did that drove this adoption or that had kind of invoked this motivation or this enthusiasm for the tool? Look, those groups that are using it, they see the direct benefit of this. And as I mentioned to you, one of the first things that I would say that management oversight. So if something you know, some simple KPIs and trends could be presented using the tool and they know that it's reliable data and if they need, if they had any question, they can drill down immediately and they can look at what is behind. Obviously, those are already, those communities who are using it, they get very comfortable and it's no question that they adopt. And interestingly, the real actually came up not from our side, but in fact the logistic people came to us that they would like to do certain POC and they will start with the real because this relatively less complicated than the normal traffic. And we have actually pipes, trucks, train and their barges on their new scope. So they choose the real to discover whether it is their potential. Plus, they would like to know the benefit of having a standardized process. But mind you, we have put together now tracking the felony success measurements. So we are looking at such as such a thing as enable data driven decision. The standardized processes such as the process effect effectiveness. So for example, quality, error rate, customer satisfaction or process efficiency as like cost, resource efficiency or process cycle time. And obviously, there is the process compliance as well. And whether we can reduce cost, enable process automation, for example, and whether can we improve the customer experience such as the usually time or our reduction, the reaction time is actually getting better or maybe proactive. Those are really where I think we can have a lot of games. Last question I would have for you. And this is something that you expressed when we hit our first touch point together a few weeks back. And that's also asking you from your position of a head of group corporate IT. And the question is, do you maybe have a fear of getting too obsessed with the process money and with illness in general as a tool now strictly speaking from your position as a, that you are managing all those tools next to one another. And how do you see this being integrated into the daily process that you're running? I mean, obviously my, I'm very excited about the tool. And you can tell that, you know, and but the company itself is very cautious to use it. They don't want to be addictive and addicted to, to process mining. A certain tool of process mining is on one hand. And saloonis is creating a ecosystem, not only process mining, but in fact a lot of intelligence, a new way of looking at this, maybe way of looking at the processes, the automation and some people has the reservation whether we should actually go that way. Because the more you use it, the more business critical, certain activities and it would be very costly to jump from one tool to the other. So we would be very vulnerable to, you know, price increases or whatever you licensing, you know, could come up. I mean, in the past, you've seen with large, you know, software companies change day, licensing and became pretty expensive. And you have to think how to re-shuffle your, your available services, especially now with the cloud services, the SAS type of thing. So we have a certain type of reservation about how far and how much we should adopt from the available tools in order to be more mobile in terms of, you know, we don't question the usefulness of the tool itself, but if, you know, it's like a nespresso is creating, you know, you buy for a few euros, the machine, because you pay for the capsule. So if the capsule price goes up, it doesn't, doesn't matter, you have a the nicest, you know, nespresso machine, it's, but expensive. So that's where the cooling factor from business, because as we said that the management is very cost conscious and they are tried to minimize the risk of, you know, suddenly we are totally dependent on a tool cannot really change. There is a danger, but probably we had to make the same decision at the time when we chose ERP system. So it's a question of how much do we believe that this tool should be with us for a long time? And it's a fair fear, I would say, your fair concern, because I've also seen it, and we've discussed it on the podcast many times before, especially with automations and what tool you use, and whether you use RPA, should you go for RPA in the first place, or should you just adopt a better process, or the same goes for automation and usage of tool S, salones, how much, or how deep do you want to go in automation within the tool, and does it, is it maybe worth it to consider automating it somewhere else? You know, those are all types of questions that the business, in my opinion, needs to ask themselves. And let's end. Yeah, you have a point as well there, Andreas. Please go ahead. But let me just comment on one thing that a few years back, I was probably very much on the same thought, but on the other hand, what I can see now is that we do not have people to actually fool for some of the jobs. So it's not a question we can actually reuse the number of people to perform the job, but we have no people to perform the job. And that's a slightly different scenario. So we have a, you know, some of the jobs are so dual and not so interesting that you don't get, you know, people to perform those things. So definitely, it's not a question to reuse the cost, but actually the question, are we able to perform the job at all? Makes sense. Makes sense. And Andreas, last question would be, where do you envision mall group going with process mining? So how would you, are there some topics you would like to tackle in the near future? And how do you see it operating in years to come? Poo. And that's, we have, look, if the already running, you know, areas, so the purchase to pay the order to cash, the maintenance, the AP and the ITSM, definitely will grow. It will not be so, you know, it will not be a leap step, but rather than it will be gradually and they have already planned how to do that. So for example, they would like to extend the cost mining, some of the activities. Obviously some automation should be done. And some areas, logistics will be the last, the next one. And probably if we are successful, we will continue the discussion with downstream. And I hope they mentioned that we have this 3D team, the downstream digital development team and we will identify the next possible areas where they see that because of the competition or because of the inefficiencies, definitely we will use the next area. And we need to speed up the knowledge transfer to business because currently we do not have really a center of excellence. We have a group of people and we are successful where the business already acquired the knowledge and they do they analysis and they do almost like a, you know, part of the IT democratization to actually enable them and power them to do the analysis. So we can focus on the data engineering part and they can actually do the analysis and they can spread the knowledge to the other employees. So for example, in order to get access to the process mining, you have to go through the academy. So you cannot just go and get access to the system unless you are qualified to be a standard user or you are an analyst or whatever. So we need to definitely widen the bases and create the whole culture and the mentality has to be changed as well. So a lot of change management has to be done. So I would not go for another five or two more processes, but rather than make it stronger, reliable and something that we can build on. Well, Andras, all I can say is good luck with that. I'm sure that there's a lot of work still ahead, but already I know that there's already things to be proud of and things that have been achieved. And you know, it's a never ending cycle of continuous process improvement. So best of luck to you and your team. Thank you. And if you have any questions, please don't be shy. Just reach me out. Yeah, you will also find a link to Andras LinkedIn profile in the show notes. So be sure to ask him any question that you might have regarding process mining or the general way of how an all-in-gast company introduces process mining in the way that they're working with. Andras, once again, thank you very, very much for attending our show. Thanks. See you then. Bye-bye. Yeah, see you then. And for you, dear listeners, thank you to listening to us. As usual, you can send us your comments, questions or just the general messages that you might have on us on our LinkedIn profile where we are predictive and also on our email mining or business podcast at gmail.com. If you like our show, please leave us a review. You can rate us on basically all the platforms that you might be listening at. And thank you very much for joining with yet another episode of mining or business podcast. So talk to you in two weeks of time. Bye-bye. Bye-bye. [Music]