Mikkel Svold (00:13): Hello and welcome to the Food Tech Podcast. Obviously, we talked about this in some of the last episodes, data is very important in the dairy industry, but just how important? What is actually driving dairy development within data and how can you use it? That is what we are talking about today. And with me, I have Erik Vedfald who is the Chief Architect for Production IT at Arla, which is, for those of you who don’t know, Arla is the largest dairy producer in right now Scandinavia, may be coming up even more. But at least as we speak here, a very, very big player in the dairy industry in Scandinavia and in Northern Europe. You know what, last time when we talked, when I called you up last week, you laughed when I asked you how do you stay up to speed and we were talking about data, we were talking about how to use data, we were talking a little bit about AI and all this. You laughed and you said, “Well, you can’t really do that.” What did you mean? Erik Vedfald (01:18): I think this about being up to speed is a little of an illusion because the world is moving so fast. And if I just go two years back and we talk about AI, and Copilots, and ChatGPT and how that is evolving. And if I see how my own kids is actually taking those technology in, it’s look amazing how much they’re just adapting and accepting this as part of the journey that they are. Well, I don’t have to translate that into our production space, this is a slow evolving change that’s getting there and it really has to be linked into the production. You have to make sure the data quality is there. You have to make sure the decision is taken there where the consumer part is much more easier, but there is this interpretation that part of you get something introduced to your daily life as part of your iPhone and part of the data journey that you are. Mikkel Svold (02:21): Part of what your kids come home with. Erik Vedfald (02:23): Yes. I also saw the new AirPods that’s coming that where the Pro is coming with generation 3, has automatic translation in the ears real time as part of the function that you can get. So you can- Mikkel Svold (02:36): So it’s processing in ear? Erik Vedfald (02:38): Yes. Mikkel Svold (02:38): Wow. Erik Vedfald (02:39): So getting that and then looking back and how do I actually translate that into my production environment? That’s a huge challenge to get there. So we will never be able to keep the speed that the world is evolving around into our production environment. What is important is that we find the right cases that we do and actually lift technology as part of the business cases that is being deployed into the production. And those will take huge steps into our production environment. And that will also give us challenges about people have to adapt into more or less you do a quantum leap in the way that you work in the past and work in the future. Mikkel Svold (03:22): But now looking at the tech, and data, and AI development on consumer products, we all know ChatGPT, we all know AirPods, those kind of things. Do you think at Arla right now, I guess, well, I don’t know if you can say anything about that specifically. But do you think the dairy industry then, do you think the dairy industry are developing too slowly or is it just you can’t do anything about it? Erik Vedfald (03:55): That’s actually a good question. I don’t think we are developing too slowly. I think we are developing the speed where we can because it comes down to making money at the end. So creating business cases where you can really infuse technology into those business cases. That’s been cases about, for example, power consumption, how can we utilize, you can say adding to the grid and removing for the grid, how can we adapt and looking into AI models, optimizing how our production facility runs. Those is quite mature, but it still goes around the business case and the technology stack there. So it’s like, you can say have a big black room where every time you get a great idea you more or less light up candle. And that’s turning a small area with light and saying that’s now the new, you can say the new era of technology where we can really push the boundaries. (05:00): But the other areas didn’t become impacted on that small business case or use case that we had. And that’s how it goes all the way around and we are not lifting, you can say continuous, most of this is coming in steps. Because there is a big gap or there is an opportunity that we can look into so we can really, really get that technology in as part of the business case that we are driving how to save energy, how to track more raw materials. Mikkel Svold (05:30): But I know this may be production managers sitting out there listening, they may be shaking their heads because of this question, but anyway. When I look at the data that we have in my personal household, what I would do today is I would take that Excel sheet and then I’ll probably copy paste it or just drag and drop it right into ChatGPT and let just kind of deploy AI right away, right off the bat. Why is that different from the data that you have? I know you have more data, but is it much more complex or what is it that, why can’t you just flick on the switch and say, okay, welcome AI and here’s our production data? Erik Vedfald (06:12): Yeah, yeah it- Mikkel Svold (06:13): Is that a dumb question? I don’t know. Erik Vedfald (06:16): No, I think it’s the scale and magnitude that makes this hard. So if we take an example, you take one tank and we put some raw milk into it, we pasteurize it and put into another tank. And there we have it as pasteurized milk ready to put in a carton. Putting that as part of an AI model would be quite simple. Now I add another line, and another tank, and another tank, and I add some chocolate flavor that we want to do or some mixing or we want to drive some powder. So the complexity of the site and the possibilities that the product can flow makes that quite difficult to have the- Mikkel Svold (06:57): Isn’t that perfect for AI? Lots of data, lots of possibilities, lots of good outcomes. Erik Vedfald (07:03): Exactly. But there has to be a model behind it. So you have to tell them that those key point how to take decision and that’s actually where the modeling becomes quite difficult in setting it up. Because you have to make the rule set around AI where the model really becomes self-training in the setup. So the complexity makes it a little hard, but you can really get there if you have the data quality there. And then we come back to the quality and the sensors and how this is being captured, and the system that’s tapping into it. Mikkel Svold (07:36): Yeah, I just recently had another interview with a company called OMRON and they produce sensors and they talked about this dumb and well, we talked about dumb and deep data basically, which is the sensor you mentioned, is that a one or zero? Is it true or not true? True and false? Or is it a visual sensor that can actually see things and give a lot of different data? And I’m figuring that that’s also what you’re talking about. You also need some hardware out there. But I actually want to come back to the data that you find the most important. In your opinion right now, what is that data? And not necessarily just for putting into AI, but just in general, what kind of data sheet are you looking into right now? Erik Vedfald (08:24): So I think the data is not the most strategy that we’re driving. I think more the quality is more the question that you have. Mikkel Svold (08:33): So the quality of data? Erik Vedfald (08:35): The quality of the data that we want to build into it. Most cases it’s driven by the use cases or the business cases that we run where we identify the data that’s being needed to fulfill that business case. And here the key element for us is quality of the data, not just having data or making sure it’s there. It’s making sure that we actually have a good enough quality to run it for analytics. And that in some cases is quite difficult to have the set of data. That could be production data combined with the laboratory data. So starting combining those data. I think a good element for this one we had for a long time there was this, you have a data lake, you have one big lake that you’re putting into it. Mikkel Svold (09:21): Yeah. And then different functions in the company, they can kind of detract or distract the data they want. Erik Vedfald (09:29): And I think when we have discussed this, they also talk about data pawns in it. But if you see data as a big skeleton that every data source is a bone that you have in your body, then you need different kinds of bone. And you also have, you can say the fiber that’s running between the bone, that’s also in some cases referred to a digital thread that’s mapping the different data sources in. But you have to have a whole body or you can build an arm but you can’t, if you want to attach it, you have to have the shoulder in place for example. (09:58): But you can build a hand and having that data running and you can take all the bones in the hand, get that sorted in and run data on this one. And then as you build, you can build different parts of the body and built that in. But you really have that insight that when you have the whole mapped out, you have a full skeleton that you see. And some of those when you go to the hospital there are something in there for you looking at, you can see all the bones in the body. Mikkel Svold (10:24): And just for me to understand it in reality, how does that translate over to what you do in production? Erik Vedfald (10:33): I think this about setting up small zones of where data is quality. So we have zones where we start getting more and more mapped out. That could be for example the intake where we take in milk. There we get quite good, we create more digital solutions, we take more samples, we make sure that all this data is being stored. And we start to having discussion, how can that actually impact the final product? And talking about how does that data come in and this is where we are learning into looking into it, how does those data model actually become active? So identifying the data points and making sure that those are key where we think there’s an impact for the products and seeing the outcome. You can say you can do testing and you can also blind test with the end users to figure out is there actually a different on those two products. And then start looking on what is the key elements for making the perfect products and why is there sometimes that deviate a little into the products. Mikkel Svold (11:34): How far along in the implementation of acquiring this data and using it would you say you are right now? Erik Vedfald (11:47): I think we are quite mature. I think there’s still a long step to go. I think that IT is, it feels like sometimes IT is a little ahead of the rest of Arla because we sit and work with it on a daily basis. I think one of the key challenges that we’re facing in a day-to-day business is how to drive the business change into the users. Making sure that they use the tools that IT provides into the production environment. And I don’t think that’s just Arla, I think that’s a common, you can say challenge. You have engineers developing awesome tools or they think they’re awesome tools. And then the user start having that adoption into how to use it. So starting having more focus on UX or user interface that how the operator is actually working with the tools, making sure that we actually get the turnaround on business cases that we do. So we also harvest the value, not just IT deploying software onto production sites. Mikkel Svold (12:56): Now just to kind of sum things up a little bit, maybe mostly for myself actually. But so what you would recommend and what you’re trying to do is you’re trying to, instead of talking about the perfect data, you talk about making sure that the data you get is quality data. So it’s a high quality data, that’s more important than having lots of data. Erik Vedfald (13:16): Yes, definitely. Mikkel Svold (13:17): And where would you start? Where’s the most important place to start harvesting that data? Is that from the raw milk coming in to analyze that on an inline analysis or is it further along the value chain? Erik Vedfald (13:36): I think this is depending on the product that we’re running. So definitely a key element in this is the how mature the site is. It’s depending on what equipment the site has and how it has been configured. There’s so many aspects in this. From looking from a data point and a governance around it, I think it’s key when we start looking into the different sites to make sure where is there potential savings or you can say opportunities for optimizing the production. That’s where we start driving in data quality. And the good part is as soon as you have done one place, that data will be there and will be ready for the next. So it becomes this snowball effect that every time we do a dataset, it becomes quite good and it’s prepared so you can build on it for the next business case that’s come or the next opportunities. Mikkel Svold (14:33): And try and put on your IT glasses. How do you actually ensure that? What is that about? Is that about making it possible to add another custom field in the database or how do you do that from an IT perspective? Erik Vedfald (14:50): That’s a really good question. I think this is where the governance around it, making sure that the people that’s working on the systems today. Understand what does this actually mean, putting a new temperature sensor, or flow meters, or lab result in place, what does that mean to the system that we do. And making sure that the IT platforms that we’re running support that journey. And that’s part of the legacy that we just touched upon that getting you can say the site in a more structured about the process that they do, making sure those processes being followed because that also supports the data journey that we are coming in. So thinking at data in from the beginning in the processes and talking about the ambitions where we want to be in three years or five years is key element. And it’s also key that the users who use the systems understand where this data is being used and how this data is maybe not tomorrow but in half a year, in a year needs to be prepared. (15:52): In most cases when we talked about the AI, they need to go three years back to have enough data to run some analytics for it. If we start implementing some points now we can start using them in three years. So it’s really important that we get that mindset built into people that data quality is something that’s key for us to drive data-driven decision in the future. Else we will end up having half data set or the quality that’s only 80%. And of course the machine learning will start adapt and take decision on it and improve and take more decision into this one and slowly build there. But the better that we can do data quality from the beginning, the easier it is to get in the right direction. Mikkel Svold (16:32): I think we’ve all tried using some AI tool, ChatGPT, Gemini, whatever you use. But I think we’ve all tried using one of those and then just getting an answer back that’s completely hallucinating. And that of course comes back to the quality of the input. That’s basically because it has no answer in the data set or it doesn’t have a strong answer or it maybe has a wrong answer. Erik Vedfald (16:58): And I think that’s the key element in this case because as soon as you put it out to an operator who is, you can say, working as a daily process today, get a new tool, you have to introduce it. And you have to make sure that they adapt into this one. If that tools give the first three times the operator tries it a wrong answer, then that tool is going to go into the desk and it will require 10 times as much effort to take him and take it up of this- Mikkel Svold (17:29): Of the drawer. Yeah. Erik Vedfald (17:30): Yes, and then try it again. And if it then fails again, you have to invest 10 times more energy again. Mikkel Svold (17:37): It’s one of those where trust is built in centimeters and torn down in meters, right? Erik Vedfald (17:42): Yes. Mikkel Svold (17:42): So you have to start all over again. Yeah. Erik Vedfald (17:43): And that’s the discussion about the data set because there is also a list. We can just get something out there and we will adapt into it. But if you are working full time focusing on creating really awesome product for any factories in the world, and they really have pride in delivering that good dairy products, so adding IT tools that will not support the journey, then it will take seconds. And those tools will just be discharged. And you will not gain the value out of it, you’ll not start building the model and everything else. So that sentiment to really have a good data set from the beginning, the use of field empowered, they feel it’s going to help the daily work instead of you could say over-complicating the daily work. That’s a key element for us to drive digital mindset to the future. Mikkel Svold (18:32): When you look at the different data that you already have, what kinds of errors, or uncomfortable truths, or hiccups have you already come across that you caught with data? Erik Vedfald (18:53): We can take an example. For example, OE, I think every product site has a discussion about OE and how to optimize data for this one. Then starting looking at deep dive into this one, you see a lot of opportunities and also a lot of missed data that have a huge impact. And that’s quite interesting to see how much that is driven by you can say machines and how much that’s actually driven by not having proper systems supporting the operators on a daily basis. So some of those key learnings where we think we had good data because the system should be capturing that automatic, maybe be adjusted by the operator due to wrong implementation of user interface and everything else. (19:41): So I think this coming back and evaluating the data is also one of the key learnings that we have done that not just getting it out and getting learnings from it but really be good to have the dialogue with the daily users. Getting the feedback and evaluate that you actually get what you expect out of it. So the validation of the data is a constant improvement that you need to do every time you update something in a user interface or you give new features to the users. Make sure that they understand and use those features correctly instead of they have a misinterpretation of why they were implemented. Mikkel Svold (20:15): Is that sometimes an eye-opening dialogue? Erik Vedfald (20:19): Definitely it is. I will say some of those utilization of tools is definitely where we see something else. We never thought that it could be used that way. Mikkel Svold (20:29): How’s that? Erik Vedfald (20:33): We developed a tool to remove, you can say paper lockings and everything else that was deployed to a lot of our production sites. It’s a manual data into tools where you more or less has like a web page, you can set it up the way that you want. It’s a quite free and flexible tools and it capture data into our production environment to make sure it’s being stored and all the right governance rules is applied to it. Most sites has it as a standard implementation but there’s a couple of sites who have taken a little beyond what we expect it could be used to. But we can definitely see how that tool can be used for those purposes, we never just anticipated. So they have learned a lot from it. And I think that’s probably the biggest learning that we get out of this one. If I see that site, they used a tool that was not intended to be there. And from an IT and from an architectural point, we will probably make adjustment into this in the future. Mikkel Svold (21:40): I was just about to say, are you then going to take that utilization case and then roll it out on the other sites as well? Erik Vedfald (21:46): Depends. In this case, I think there is other systems that’s better to adopt into it. But the biggest learnings for me was see how the mindset of the site, you can say taking this tool fully under the skins and look how can I optimize the day-to-day business that we do and reduce the amount of complexity for the operators. I think that was the big eye opener for me. And then you can say we failed in showing them how many other tools that they have in the toolbox. They only had a hammer and every nail or every screw was then just a nail and they used it for this one. But the mindset about how they’re working on the daily day-to-day work, getting that in, getting input from the operators, getting input from anywhere in the organization how to optimize this. That was really an eye opener saying they actually took it in that really the anchoring into the business is one of the key elements. (22:47): And taking that and looking back on some of the cases where we have been challenged in the past really make this, the end user engagements is really the key for us to drive digital. It’s not so much about IT can deliver a solution, it’s the end user engagements and the acceptance from the end user that this is an awesome solution that we’re getting and we really see value in it. So the value proposition that you’re presenting, so why should I, when I give you these tools, what do I get out of it? What is the gain for me and in my daily work? I think that’s the key element of selling it in and really get that adaption into the people. Mikkel Svold (23:27): I think it talks quite nicely into something you mentioned actually before we turn on the microphones where you said that sometimes the operators and production, they have a desire for some end goal. And then how can you as an IT department cater to that? So the discrepancy between what production wants and what IT knows they want and what they can deliver. And can you maybe explain how that works? Erik Vedfald (23:58): I think it’s key. I think some of the journey as part of the transformation in Arla that we have done. If I go five years back, then looking in how IT saw themselves, we created awesome solutions. And I tried to paint this picture that we had a shop, we put it in the windows and we didn’t understand why the customers didn’t just come and ripped it down from the shelf and want to buy this amazing solutions we had. We’ve just forgot to invite the customers. Mikkel Svold (24:31): Oh that’s bad. Erik Vedfald (24:33): But changing that mindset over the last three years actually being going out to the customers, going out to our production site, explaining them why they’re getting the tools and explaining about the changes that we’re getting. It’s still, I would say, everyday struggle that we do to make sure that we still cover all the sites, all the communication, and all of it. Communication with 55 sites, it’s still a huge, you can say communication loads that’s coming in. So definitely not perfect, but we evolve hugely into this space. And we really see how that changed the adoption and how standard is now being more natural checking in as part of the journey that we are doing. So a key element for the whole digital you could say we do is communication and that’s probably some of the biggest eye-opener that’s coming. It’s not about if we can build a tool in AI, it’s more about communicating that we have a tool in AI and that’s actually a key driver for us to succeed in this one. Mikkel Svold (25:33): It sounds a little bit like all of the development, all of the new systems and new ways of doing stuff comes from the top. It comes from some desire to follow a strategic line or it comes top-down basically. So it’s something that IT wants and I know IT rarely wants something in themselves. It’s usually because someone else wants IT to do stuff. But it sounds a little bit like it’s the leadership board, it’s the C-level people, it’s the other departments. They want something, then IT develops it. And then they implement it down in production. And then it becomes a communication task to do that implementation and make production aware that why it’s important. Am I understanding that correctly? Does all of these developments come from a strategic point or do you also flip it upside down and then stuff comes from bottom-up? Erik Vedfald (26:37): There’s a big flip on bottom-up that’s coming in, has a lot of amazing development is being done on different sites. And just catching that and getting it formalized and into something that can be reused on the other sites is also quite a big task as the most site is very different in generality. But there’s a huge appetite from the sites to actually drive this. And we have so many amazing peoples at the production site, so they come up with amazing ideas that they’re driving in in sustainability, in optimizing energy, all those more cases that we’re doing. That’s really also very interesting to see and trying to capture that and make a business case that can be lifted from one side to the global of Arla, it’s a lot of our work also. Mikkel Svold (27:27): Gotcha. Erik Vedfald (27:28): From the top-down, what we do, it’s more like we want to have traceability. There is some, you can say, license to operate decisions that needs to be done. Where for us to work in a strategic direction, we need to set some direction that we need to follow and help the sites in implementing those. So it’s a mix and I think that’s also key for just having a top-down driven way of doing IT is really not a good approach. I really think this that we collaborate together with our sites and I also think it’s very important that IT sees them as a support function. We are here to support the sites, create production, and optimize their production. We are not an IT company. We are still a food and beverage company that has to have focus on this as the primary, you can say, work that we do. Mikkel Svold (28:19): I know that this could fill up an entire episode in itself and our time is nearly up, but you mentioned the communication bit where it sounds like you’ve done some learnings, you’ve had some learnings. Can you maybe dig into one or two examples of how do you handle communication of these developments now compared to what you might’ve done you said five years ago? Erik Vedfald (28:46): Yeah, I definitely think one of the elements is that we really focus on sharing information. We really make sure that we get the right stakeholders involved and tell them about both the application and how the landscape is looking for what and also the value that they’re getting out of it. So really telling what is the value that this IT solution brings to the table as your daily work because that makes it so much easier for the operator who get that tools to understand why do I actually need to learn it, and what does it bring me beside extra work in the beginning. So it’s hopefully it’s extra work. Everybody know when you get something new. You talked about a new update in the iPhone that’s coming with a new OS that even gives work. Mikkel Svold (29:34): Yes, it’s a headache, right? Erik Vedfald (29:35): Yes. But on the other side, when you get to the other side, when you have learned those changes, there is a value for you. You actually see new functionality, you can see new features that you can use in your daily work. And that brings, you can say, happiness, a joy into your work life. And I think one of the key elements is having focus about work life that we actually have IT systems who support the people in creating meaningful tasks. Instead of just having IT who creates a lot of bureaucracy about that you have to fill out certain amount of fields, and you have to update, and you have to type in, and you have to register. So we really should try to get to that point where we are supporting the people in the daily work. And I know that no matter how many podcasts we’ll do, there will always be somebody saying, yeah, we are not there yet. Mikkel Svold (30:25): Yeah. And always someone saying it was better before. Erik Vedfald (30:28): Exactly. Mikkel Svold (30:29): Yeah. Erik Vedfald (30:30): So I think that it’s tapped in again, it has to look into that again, the data set we’re doing. We don’t see the data set on a day-to-day basis why it’s important that we do it. But when we start looking a year or three years ahead, we start seeing why we actually have to focus on the data quality. And I also think that’s one of the key elements in this one. That data is not about tomorrow. It’s about the long journey that we’re getting into. Making sure that the data that we’re storing and the data that we are focused on is also something that we have an idea that how we’ll be used the next three years. Mikkel Svold (31:04): Yeah, because I was just about to ask, I know that we may not actually have time to talk about this, but I don’t really care. I think we should. When you are planning, long-term planning, and I guess in today’s IT environment, three years is quite long time. How do you know what data to really work on getting high quality data from? How do you know which data to choose? Because it’s only in three years or five years that you can actually use it. That time difference must be kind of a headache for an IT person like you. Erik Vedfald (31:44): It is. And I think the data that we talk is not, again here IT is a support function. We store the data, we capture the data. But we have so many good dairy engineers who specialize in different fields, different areas that they want to do, let’s say, butter for example. So we will have dairy engineers who just focus on creating the most perfect butter in the world. Those people know the data that we talked about. The QUEST people that works with butter on a day-to-day about quality, ensuring that you get the same smooth silk really nice product when you buy a Lurpak product. Those people knows the data or have an idea on the data that they’re looking into. IT needs to facilitate them access to that data. So making sure that we capture the data that they want, facilitating that they can run different kind of analytics or machine learning on top of that one to try to see is that correlation. (32:39): You could take the idea, if I take the cows in the winter, they are inside, they get cut down grass that we have stored over time. And in the summer, they get very fresh grass. How does that impact on color and taste and everything else? Starting running to those analytics in the future is definitely something that could be very interesting to dig into and saying how does that impact, how does that impact how you perceive the product when you get it tasted, see it opening, and all that. I think those is the IT’s responsibility to facilitate that one, but it’s really the end user. Again, the end user is really the ones who have the know-how, they have the skill sets to actually- Mikkel Svold (33:22): And the end user here, not being the consumer, but dairy engineers. Erik Vedfald (33:26): Yes. Mikkel Svold (33:28): Yeah, so they know. Erik Vedfald (33:29): They know. Mikkel Svold (33:30): They know. I think I just quickly jotted down a question coming back to the introduction of this episode where data-driven development is in dairy right now and I’m just thinking if I was the production manager at a smaller site than Arla, but still maybe a medium-sized dairy. What do you think is the most important takeaway from this talk? And if you want to start working with data, I guess, where would you start? That’s one question. And then I had another question which was how do you get started? But that I think is ask your dairy engineers. Erik Vedfald (34:11): Yeah, that’s probably- Mikkel Svold (34:13): But where would you get started? Where would you start? Erik Vedfald (34:14): So I think there’s two elements in this one. The one is IT systems actually supporting capturing data, making sure that we have everything in place. There are different systems in the market that can be used for this one. But having that baseline, not about their data, but we have a system to capture the data. I think that’s one element that needs to be in place because then you have enabled that journey that you are. And another thing it’s really important to start having that dialogue with the people that they’re using the production. So the engineers, the operators, the quality department. (34:50): And where do we see challenges today? It could be a quality issue that you’re having. Let’s say that we create strawberry yogurt and suddenly sometimes tastes something different or the taste variate a lot in flavors. So then we start capturing data here, we start talking about the QUEST department, we start talking about the operator that could be capturing raw materials coming in. So the strawberry jam that you put in, having those use cases where you see challenge or opportunities today, and this is probably the people that’s working on a day-to-day business. They have ideas and then the IT can support this, create the know-how, create the insights, starting to run analytics for it. But it always comes down to the business case. Do we see an opportunity? So I think getting those logged in is the key element to get starting on the data-driven journey. Mikkel Svold (35:49): And that all comes from walking the floor, talking with the engineers, and then of course also I’m guessing looking into the strategic direction of the company. Erik Vedfald (35:59): Definitely. And also get inspiration. Mikkel Svold (36:01): Matching those. Erik Vedfald (36:01): There is a lot of fairs where you go out where they present different business case that could be for energy. And then you can look on some of the ideas that’s there, taking those in, present them to the engineering teams that you have. They will come back with this fitting 100% into the way that we are working or it doesn’t make sense because we are not using, for example, gas, natural gas to heat up our hot water. We use electricity. Then it’s definitely a case about electricity. We need to look into it. So taking some of the market trends that’s going in the market and ask the engineering departments and ask the maintenance managers about how this is applicable to the way that you are working today will be one of the key elements. Because every site that we have in just Denmark or Northern Europe is different. They have different technologies, they have different standards, they have different maturity when they were built. So it’s so different from case to case how this fits us. Mikkel Svold (37:01): What’s your personal favorite data point right now? Erik Vedfald (37:07): That is really good question. My personal favorite data point? I actually have no idea how that… Probably will be the optimization are getting Copilot in. So getting Copilot into making my life easier. And that’s not a data point, that’s more a tool how to work on a daily basis would optimize. And again the fact- Mikkel Svold (37:31): But it’s enabled by all the data? Erik Vedfald (37:33): It’s enabled by the data that I have in the mailbox, but it reduced the amount of work I have to do. So the value, again, this is becoming from a personal driven of the value I gain out of it. Mikkel Svold (37:47): I think let those be the last words and also maybe come back to this question of how to stay up to speed. It’s not really possible, but you can get inspiration from what’s moving really fast and then seeing what you can implement and talk to your engineers. That’s also one big takeaway, I think. Yeah. (38:09): Hey, Vedfald, thank you so much for joining this interview. This is all we have time for today. And for you guys listening out there, if you have any questions, please reach out to us at podcast@au2mate.com, and that was au2mate with a 2 instead of T-O. So it’s podcast@au2mate.com, and we’ll be happy to answer any of your questions or if you have topics that you think we should talk about on this podcast, please send them along and we’ll look at it. Of course also, if you like this episode, share it with your friends or colleagues if you find it interesting. Yeah, I think with that, I just thank you so much for listening and catch you in the next episode.