4#5 - Olga Sergeeva - Data and AI in Modern FMCG Supply Chains (Eng)

«We made a transition from being a company that produces a lot of data, to a company which has control over the data we are producing.»

Unlock the secrets of optimizing supply chains with data and AI through the lens of TINE, Norway's largest milk producer. Our guest, Olga Sergeeva, head of the analytics department at Tine, takes us on her journey from a passion for mathematics to spearheading digital transformation in the fast-moving consumer goods industry.

Ever wondered how organizations can successfully integrate AI tools into their business processes? This episode dives into the uneven digital maturity across departments and the strategies used to overcome these challenges. We discuss how data visualization tools act as a gateway to AI, making advanced algorithms accessible without needing to grasp the technical nitty-gritty. Olga shares how TINE’s data department empowers users by providing crucial expertise while ensuring they understand the probabilistic nature of AI-generated data.

Finally, discover how teamwork and a systematic approach can drive data adoption to new heights. From improving milk quality with predictive algorithms to optimizing logistics and production planning, we explore practical AI use cases within Tine's supply chain.

Here are my takeaways:

• Mathematics is a combination of beauty, art and structure.
• Find your way in data and digitalization before jumping on the AI-train.
• Ensure that people can excel at what they are best at - this is what Tine tries to do for the farmers.
• Data only has a value, when it can be used - find ways to use data from analytics to prediction to more advanced algorithms.
• Create a baseline through a maturity assessment to see how you can tailor your work to the different business units.
• Follow up and monitor the usage of your data tools in the different areas of your business
• Create a gateway into data for your business users: Once that gateway is established it is also easier to introduce new tools.
• Data Literacy has a limit - not everyone in the business needs to be a data expert.
• Yet you need someone you can trust to enable and provide guidance - the Data team.
• Business users need to understand the difference between concrete answers and probability.

• How do you transform a complex organization without breaking the culture?
• Your data/digital/AI transformation team is key in ensuring good transformative action without breaking culture.
• Ensure you have good ambassadors for your data work in the Business Units, that what to transfer their knowledge in their respective units.
• Create a network of data-interested people, that help to drive adoption.
• Engage people by showing an initial value.
• Offer courses and classes for people to learn and understand more, but also to spread the word about your focus points.
• Inhouse courses provided by your own staff can increase the confidence in your data team.

• AI can mean different things to different people. It is important to define AI in your setting.
• Don’t replace existing work process with AI-driven solutions, just for the sake of it. Find ways to focus on where improvement actually provides business value.
• When you think of a new AI project, you have several options:

1. Develop in house
2. Buy off the shelf
3. Do nothing

•  Option two should be your preferred solution
• AI strategy is part of a larger ecosystem, with conditions to adhere to.
• Data and algorithms should become interconnected, also visually represented.
• «Always remember your core business.»

Transcript

Data Management in the Nordics

Speaker 1 0:00

This is Metadema, a holistic view on data management in the Nordics. Welcome, my name is Winfried, and thanks for joining me for this episode of Metadema. Our vision is to promote data management as a profession in the Nordics, show the competencies that we have, and that is the reason I invite Nordic experts in data and information management for a talk. Welcome back to Metadata. I'm really happy we have a new episode. We're going to talk about a topic that is really interesting, that also affects a lot of people in their daily life. We have a topic where we combine data, we combine AI, we combine fast-moving consumer goods, and we have a fantastic guest on the show today Olga.

Speaker 2 0:56

Hi, indrid, I'm very pleased to be invited here.

Speaker 1 1:00

Thank you, well, thank you. So Olga Sergeyeva, and you're going to introduce yourselves afterwards, but you are working at Tine, and Tine is maybe one of the biggest milk producers in Norway and it's a company that is owned by farmers and is working on behalf of the farmers distributing their goods in Norway, and we are talking about, I think, 7,000 farmers in Norway. Norway is a big country right from north to south, so there's a lot of logistics involved there. There's a lot of production, there's a lot of interesting data that you can collect, that you can work with, and it gives TNN kind of a unique position where you can get like an overview over the entire supply chain, from raw milk production on the one side to customer improvements on the other side, and we're going to talk about how data can help in that entire supply chain to connect the production on the one side to the customer on the other side. So, before we jump into the topic, olga, please introduce yourself.

Speaker 2 2:00

Yes, yes, my name is Olga. I'm working in TNF for already over 11 years and for the moment, I'm leading the analytics department. We are doing machine learning. We are doing different AI in our department. Our department is pretty fresh, but I was myself doing a lot of machine learning project conventional optimization models during my life. Also, I had to master degrees in industrial logistics and mathematical methods in economics, so I really like economics, I really like supply chain, I really like data and I'm really actually I'm coming every day with pleasure to work because it looks like every day I'm working with something I have passion for.

Speaker 1 2:44

Sounds like you are in your dream job.

Speaker 2 2:47

Yeah, it was a bit surprising because I never thought that I will work with data. I will work with mathematics and it was pretty occasional that on every turning point of my life I was turned towards the mathematics. After all, I decided that my fortune knows better what I should work with and it looks like it was really a good decision, because I really enjoy seeing the world from the perspective of data, Because for me it's like a new dimension or like seeing the world. You can see the world with the objects, but I see the world with the numbers and I really appreciate this mathematical beauty of the algorithms which you can apply for the real world projects, for the real world milk and I don't know if it was a part of your question, but I also like that. I'm working for Tina, where I can see the factual product we are producing, and I'm seeing the factory, seeing the actual milk cartons running around in the factories. I really love it.

Speaker 1 3:46

I really love, when I'm traveling in Norway, seeing the farmers which are producing the milk and I love to take the milk carton from the shop and have a feeling that I have something in common with yeah, this is fantastic when you actually can see the product and touch the product and and feel you are part of that chain, right, that chain that puts the milk carton on on the breakfast table um, really interesting. I've been working for years and all in the other I still haven't seen any oil, uh, so what? What are you doing in your free time, when you're not working?

Speaker 2 4:21

yeah, in my free time I I usually have some abonnements to opera and philharmonia. I'm a friend of National Museum. I like all kinds of arts and actually I see that mathematics is a kind of art that is the most beautiful of all arts. What I actually said to my child yesterday when he didn't want to do mathematics I like all possibilities, like art possibilities, what we have in Oslo. I try to use them. But I also do some gardening and now I'm very happy about that I can pick up some mushrooms in the forest Perfect time for this.

Speaker 1 4:56

Well, maybe a couple of years down the road you're going to get some truffles in Norway as well.

Speaker 2 5:01

I'm okay with the steinsop.

Speaker 1 5:03

So you had some interest for working with data that has been broken while you were studying. But what makes it so interesting to work with data?

Speaker 2 5:13

Data is beautiful, and I should say that when I was at school I had a teacher who demanded us to make all these mathematical graphs with different colors, and then I think I have a passion for art and passion for beauty and passion for structure, and mathematics for me is the concentration of this. But I think I wouldn't be that happy. So it comes from school. So my father is also doing some mathematical models in biology, so I was always looking at the world from the perspective of the models. I just find it beautiful. I find it exciting that with mathematics as a tool you can open the secrets of the world. But also I'm not that theoretical to do the science. Actually I was trying to do some PhD but it didn't work out. But now I'm very happy that this math has real world problems.

Speaker 1 6:06

But I really liked that view on mathematics as art, as an art form. That's really interesting and I think that we I mean we had a lot of different people on the podcast so far and everyone has their personal story about how they got to work with data. Where the interest comes from, it's always unique and that's what I think is so great about that data community we have in the Nordics that we have so many people with different backgrounds, with different views on data, that come together and work towards that data-driven or AI-driven future and the availability so really interesting.

Speaker 2 6:41

Actually, I really liked the question because it made me to reflect where does my interest come from? I didn't really think so much about it and, yeah, thank you for your question. I mean it helped me to understand myself better.

Speaker 1 6:56

So let's try to understand Tina and the drives in terms of AI a little bit better. So a question that always pops up when we talk about the AI hype is are we actually still going to become data-driven companies or are we going towards more of an AI-driven future? But what do you see? Is there a difference? Are there basically different terms to describe the same? Or where are we going in the sector, in the business?

Speaker 2 7:24

So, tina, in the age of AI, I should say that we are still. We need to take step by step and we should have first Tina. In the age of data and digitalization, I should say that Tina exists, that the farmers can do the things that what can do and know how to do the best producing milk the things that what can do and know how to do the best producing milk. And Tina exists that the farmers shouldn't worry about all the rest what's happening after the milk is being produced and how to bring this milk to the customer. So Tina is taking care about logistics, about production, about marketing, sales and supportive process for the farmers. So we are farmer-owned. We are working in the interests of farmers. At the same time, tina you mentioned, it's a very big company and we in fact have an overview and control over most part of the supply chain and overview over the whole supply chain. So we have a lot of data.

Speaker 2 8:22

When I started in Tina for 11 years ago, we were saying that we have a lot of data. When I started in TINA for 11 years ago, we were saying that we have a lot of data and we are producing a lot of data, but it took a while that now our old data are connected and they are all put together into some data platform where we can easily access it. The data quality is pretty good. So, since what has happened during these 11 years? We made the transformation from being just a company, a huge company which is producing a lot of data, to the company which has control over the data we are producing, which has control over the data we are producing, and now we are working very hard with increasing the maturity of our business units to use this data, to create value from this data and to use this data on more and more advanced level, from visualization to using the data for decision-making, plus having some predictions and some more advanced algorithms that are helping to act more efficiently. I don't think that Tina will be purely AI-driven, because our core business is to bring the milk from the producer to the consumer. Ai, for us, is a tool we are using to improve our supply chain, together with all other tools we can for this purpose. So we are yes, we are now data-driven, we have overview, we have control over our data, we are using our data, but AI is just one of the tools we are using to make us more efficient. So we are data-driven and AI-empowered company.

Speaker 1 10:06

That's a fantastic way of saying it data-driven and AI-empowered. I like that. There's one thing in what you said that I think is particularly interesting so I wanted to just emphasize that and that is that you talked about increasing the maturity of the business units to use data and create value with data, and that's what it's about, right, ai and data creates value in the core business, not in the data department. So, really, the work you are doing is for the core business of Tina and not for your department to shine, which is kind of it's obvious, right, but it's something that I think we need to emphasize more and more, especially during the AI hype. We build up units that are basically just data departments, blowing up in their budgets, in their complexity, yet you have to produce the value for the core business and in the core business. So that's where the valuation happens and that's what you need to focus on.

Speaker 2 11:03

I think it could be different cases. When I was doing some courses at MIT, they gave us some cases where the AI product became the main product of the company, but I don't think that for us that's going to happen. Still, the milk is our core business, but in theory, if we can imagine that everything can happen. I can imagine that tomorrow I will invent some AI tool that will create a hell of an amount of value that will still bring the value for the farmers. I don't think that we will say no to this, but for now, I still think that our core business milk, and AI is going to help it. It's the main purpose of AI to help the core business. Mel and AI is going to help it. It's the main purpose of AI to help the core business.

Enabling Data Adoption in Organizations

Speaker 1 11:47

Right. And then there's the classic digital transformation question. That comes right. If you are introducing new ways of working, you're introducing AI tools, you're introducing data observance or you're just basically introducing data to be used by the business units, that will eventually affect selfish processes. It will affect the way people are working. How easily are those things adopted? How much effort are you putting into the adoption by the business?

Speaker 2 12:19

We work pretty structured with this. So first, the business. Several years ago we made an evaluation of maturity of different parts of the organization digital maturity so we saw that it's pretty uneven. So we had departments that were very in front of adoption of digital tools. Some were more skeptical. We identified what we are missing data products and so on. I mentioned that we were transitioning from the on-prem to the cloud solution where we make huge effort to put together all our data. So in the parallel, we were working with every single functional area antenna, with helping to bring them forward in this maturity journey for the data.

Speaker 2 13:06

We are not finished. We are still working specifically with every single area in TINA and, yeah, we achieved pretty good results. We are following up the usage of our data tools in all parts of the organization. We have a permanent interaction with them for improving these tools and so on. Yeah, it's just our everyday work to increase the maturity.

Speaker 2 13:32

But let's say, now we are not at the stage when I can come to every single part of the team and say I will bring you a tool and you are ready to use it and you are ready to use it, but most of our departments are ready to adopt pretty good data visualization products and we are working hard and we got already on board pretty many departments that they are able to adopt more complex.

Speaker 2 14:02

But I think here is also one of the interesting things what I experienced that as soon as the departments adopted the data visualization tool and started to use this data in the processes, when we are delivering them some AI project, which is pretty often delivered in the same interface as some Power BI report, let's say so it's very easy for business to adopt it. It's a big step for the functional area to make data as a main. It's like a part of the decision making and as soon as the data is a part of decision making, actually it doesn't matter where this data are coming from. Is it just like a historical data or it's a prediction from AI? So as soon as they are using data, it's okay to use AI to produce this data. So as soon as the initial adoption is at place, the AI adoption is going pretty fast, I should say Interesting.

Speaker 1 15:05

The AI adoption is going pretty fast. I should say Interesting. So what you are basically looking for is kind of an easy gateway into data, into using data in your everyday life, in your decision-making process, and once you have that gateway open, it's also easier to introduce new tools.

Speaker 2 15:21

Yes, absolutely. I think that in the strategical perspective, we don't need that. Every user of the tool needs to know how linear regression works. They just need to have a safe environment to get the safe data and understanding how to use it.

Speaker 1 15:41

This is really good because we've been talking about this for years now, about data literacy, and I think it's kind of a bit misunderstood.

Speaker 1 15:48

Right, the idea of data literacy is that everyone in the business can read, write, understand, work with data, and and I get it, it's, it's important? Um, it definitely is, but there is a limit to it. Right, at a certain point, the task gets so complex that you you as a normal user don't really need to know how it works. You need to know that it works and you need to have someone you can talk to who knows how it works. And that also changes a bit of the at least for me, a bit of the role of data department. So, if it's the business units that create the value values create in the core business, you take on a role of an enabler to help them create the value, bring them the tools they need to create that value, but also to bring the competency to understand the tools you have implemented. So if they have any questions on linear regression, they can talk to you and your team, right, and they get the right answer. And that's more of an enabling role than a role that is actually affecting.

Speaker 2 16:47

I just hope that they will not have a question about linear regression In my world. They shouldn't even worry about which algorithm we are using. They just should have the understanding. Like one of the core things that we try to explain to our business when we are delivering some AI or machine learning result is that it's not anymore the answer. They answer it's just a probability. So that's the step which I think is critical for data literacy for now. So you need to understand what kind of data you're accumulating, if it's the actual data or if it's a prediction with some probability. So that's the necessary data liter.

Speaker 2 17:34

But I'm pretty sure that maybe not yet, but in the future we will have the tools which will allow the end user to use pretty sophisticated algorithms without knowing how they are working.

Speaker 2 17:48

I know that maybe a lot of data scientists are not agree with me, and I was discussing it like a few days ago with one of the lead data scientists who was not agree with me but I think that the real power of AI is a tool which will allow you to use advanced algorithms in a safe way.

Speaker 2 18:08

Without that, you are really know how it works, but you are able to use the result properly, and I think that some of the tools, what we are using now, allow it, and I also think that some cases where we are speaking about machine learning and the production, for example, for sensors and so on should be done in a way that the person who is working with the production process doesn't really need to think that I'm using AI, but just see the forecast and understanding that it's a forecast, so not necessarily need to. I don't want to explain linear regression to ever. I can explain it to everyone who wants, but not for everyone who will use the data. I hope so and I hope I will make it safe enough for people to use it.

Speaker 1 18:58

Right and I agree with you. But that puts a certain responsibility on the data department right Puts a responsibility on and we've been talking about in the podcast before about responsible AI. We've been talking about explainable AI and all the methods, all the ways of understanding how AI works, how the prediction comes to, the result it comes to. Those are important and you need to have that knowledge somewhere in your organization but it doesn't need to be everywhere in your organization.

Speaker 2 19:28

Yeah, it would be very sad if it's only a data department can use the resources. Yeah, I pretty agree with you.

Speaker 1 19:36

So we are in the middle of one of the main topics here, right? The culture and data culture, and we talked a bit about technology, we talked a bit about process. Let's talk a bit more about people. What I think is particularly interesting about T-LINK is that, while you're looking at a large scale, at a complex organization that is farmer-owned but and you said it already it's not an IT company, right? There's a core business there that you have to serve, and the big question is how do you serve that core business? How do you transform such a complex company towards data-driven without breaking the culture, right?

Driving Data Adoption Through Teamwork

Speaker 2 20:12

That's very much the art, and it's not just me who is transferring it. I would say that the success of the transformation, like a digital transformation, what we have is because we managed to create a very good team for this and this team was working very fine and was able to involve organization. The process was very structured. I don't know what's the secret of success of this team. I was just a part of this, this team, but I just know that without having such a great team, we wouldn't succeed in what we achieved.

Speaker 2 20:49

For now it's like from one side, so it's a catalyst that was the digital, but from the other side, we shouldn't underestimate people who are working together. So we see that people are very curious about the data. People want to work to do the work they are doing better and they see if they know they have the data, they have the tools, they can do the work they are doing better. And they see if they know they have the data, they, they have the tools, they can do the work better. And this initial desire people to make the work better and our preposition of these tools is working very fine together. And let's say, of course, we we meet some some situations when people are very happy in the way how they're working. But in majority we always try to see that people want to change.

Speaker 2 21:36

In general, the team like a catalyzator. We have a bigger, pretty bigger team which consists of the people from the business who is very data interested, very data forward thinkers. So we are working very much with these people and they are transferring this knowledge and transferring this interest and data to people they are working together with. So we have kind of a network of the data interested people who is helping to adopt what we are developing in the different business areas. But in addition to this formal structure, we see a big interest from the people who maybe has nothing to do with the data, for example, a driver or a warehouse worker.

Speaker 2 22:27

And when we launched some courses in generative AI, we were doing this in-house and we invited everyone and we saw extremely high demands for these courses. People were extremely interested and wanted to understand how does this work and use it in those everyday life. So let's say, in the first several months we got over 700 people who assigned force for the e-generative ISO. The people have wanted and we created some structure to satisfy this demand. Of course, it doesn't work always like we want. There are a lot of challenges with digitalization, but we have ways how to work with it.

Speaker 1 23:13

This is really interesting. I'm just summing it up a bit. The first thing I took with me is you engage people by showing value. Once you show an initial value, it's much easier for people to get engaged, to get an understanding of oh, this is what's happening, this is what I can do, so let's see what more is out there. I like that one. The other one I took with me is you need to find some champions, people that are excited about it in the business, that can really push your agenda for you and, with you, build a network around those people to engage more people, to get them connected. We had a previous episode where we talked about communities of practice. This could be a really good way of doing that building that network. And then you talked about the courses, and I found it interesting because you didn't use the courses for the course's purpose only, but also to really spread the word about Gen AI, for example, what you do, and get people engaged. I like that.

Speaker 2 24:13

Yeah, and actually about the courses and all this availability of education in this area. I think we were very lucky with deciding to do it in-house. So the people who are working in China are seeing that we have internal competence we are not asking someone to tell about generative AI for us so they see that we have a competence inside of the house and they start to trust that Tina is good in digitalization and just this trust also driving the progress very much.

Speaker 1 24:40

So let's get a bit more practical about it, a bit more tangible. I guess we wanted to talk about some use cases to really get a feeling of what AI, what data, can do in a business. So first of all, maybe before we jump into two specific use cases, maybe you can tell us a bit more about how do you initially find use cases for your work?

Data-Driven Strategy for Supply Chain

Speaker 2 25:01

Yeah, so, as I told you, tina is a company which has control over most of the supply chain. It means that we have different departments production, logistics, finance, sales and so on. In fact, I feel myself like an internal consultant who is just coming to different parts and has the freedom to come to these parts and to propose some AI solutions. We have potentially the full range of different use cases and we always try to think that AI should bring the value to the core business, to the supply chain. It should make it more efficient. But it's nice to think about. But when we are searching for the use case, we always, yeah, first identify the idea. It could be this, it could be nice to have a look at this, could be nice to have a look at that, and so on. But after we identified these potential areas, we have a look. Do we have enough data? Do our data good enough to explore this area? If we don't have enough data, we ask do we have good enough partners to work? Is the maturity of this area high enough that they are able to work and use the results to work with the results that we are providing? So it's like a top-down approach, but we also do pretty regular meeting with, like these active champions, data champions from different areas. They are coming with different ideas. I want to have that. I think that would be nice to have a look at this, it would be nice to have a prediction of that and so on.

Speaker 2 26:48

So we try to merge this list of we name it brutal list of the ideas with understanding of where the potential profit can come from. Do we have data and do we have maturity level, if the maturity level is high enough? So it's both top down and down top approach. But so it's like how it's working now. But now, more and more, I see the stream of people who is coming to us after we build up some reputation. People know that these guys they can help me with some more advanced analysis. So they are coming to us and asking can you help me with this? I don't really know, but I feel that there should be a possibility to connect this data and that data and then we are sitting together and trying to figure out how much value, how much effort, all the rest, evaluation, what you need to have. So that's identification of the cases.

Speaker 1 27:43

How do we do Really interesting. Let's talk about two cases. We had a prep meeting so we talked about those already. But one thing that I found really interesting was milk is probably the biggest product you are selling on behalf of the farmers. So when we talk about quality of the milk, you were doing some work there. Can you tell us a bit more about that?

Speaker 2 28:00

Yeah, I should say quality and security is one of the main words here in TINA. We are producing the food so that we are safe and we are responsible all the way for the safety of the product and we have control of the quality on every step of the process. And we have control of the quality of every step of the process. But to help, and when we start with milk collection, of course we do a lot of laboratory tests of every single car which is coming with the milk. We have laboratory tests and so on. But sometimes we need to make more profound tests for the bacteriological profile of the raw milk.

Speaker 2 28:37

But this more complex analysis demands more effort and we cannot do it for every single farmer every single day, every single time we are fetching the milk. So we do some what we name in Norwegian stikprøve, so some random tests and based on the history, this test it can be some aerobic or anaerobic forest analysis. We are having some algorithm which is saying for us which farmer root do we need to check specifically a bit more and that helps us to identify the potential problems with the milk deliveries where we suddenly should throw a big amount of milk because there was a bit of bad milk which was blended with a lot of good milk. So we are having some prediction of their milk quality and it's really influencing very much the quality of the end product.

Speaker 1 29:34

But this is really interesting because we talked about this already, but now we can get a bit more tangible. Right, when you introduced those checks, how were people adopting their processes around it?

Speaker 2 29:44

Exactly, this project was not led by me, so I just know that we anyway have a plan of checks and now some checks. We still do some random checks, but some part of this checking plan is provided by the algorithm. So more how it was adopted, I mean how it was received, I don't know, but unfortunately I'm not the right person who can answer you here.

Speaker 1 30:12

There's another use case we talked about and this is maybe the core of what Tina does and that is how to make the supply chain efficient right. How can we use AI to help us with optimizing logistics, with production optimization the one side is just optimization, the other side, but also like classic road planning, for example. So how are you going about using AI in the supply chain itself?

Speaker 2 30:37

I should say that now we should agree about the terminology. What is AI? So, if you're referring one of the latest presentations I got from Gartner, the optimization, routing, simulation and a lot of other conventional tools were included in the concept of AI and a lot of other conventional tools were included in the concept of AI. If we're speaking about AI as that broad term, we are using heavily AI in optimization of our supply chain.

Speaker 2 31:12

But let's say I'm pretty confident that good old times, optimization has a right to exist and it's, for now, is a primarily tool for our optimization and production planning within our supply chain. And we're using, and we will use all good old times, heuristics for routing and everything. I think that all these good tools which were developed over time are we don't need to exchange them with machine learning algorithms, so they are just doing those jobs very well. But there are still some parts of the supply chain which are very difficult to model, for example, and to formalize as a part of the optimization, and there machine learning can help us a lot. And, yeah, so we are using a good blending of the conventional tools or a part of AI now and machine learning things.

Speaker 1 32:00

Machine learning is primarily for the forecasting, evaluation of the probabilities and so on, while the optimization for planning so a couple of years back, I had a great conversation with someone about something as interesting as distribution degrees of milk. So how much milk are you sending to each and every store so you can ensure that the milk still is fresh in the store? They don't pile up on milk supplies in one store and don't have any milk on the other store stuff like that, which I think is really interesting. But, as I said, it's a couple years back and it's been done on really classic terms right, using historical data from a data warehouse to predict milk sales for each and every store, trying to distribute the milk according, and I don't think you need any AI tool to do that. Yet there might be the chance to be even more efficient with that. So the question is really how do you balance the investment of building a new tool on the one side with the effect you're getting out of it on the other side?

Speaker 2 32:59

That's exactly what I was speaking with my colleague about this morning. So we are actually, let's say, when we are thinking about a new, let's say reasonably big project, we have always several options. We have an option to develop it in-house. We have an option to buy a specialized, or we have an option to develop it in-house. We have an option to buy a specialized, or we have an option to do nothing.

Speaker 2 33:20

So let's say, as I see it now, for us, if we see a very good product on the shelf which we can use let's say some AI planning this or that we will use this tool. We are not going to develop a very fancy tool and to top these specialized companies which are working with the development very advanced platform for prediction I don't know, whatever it is. So, if we see a good tool, we are going to use a good tool, given that investments are lower than potential income For now, we are just running some tests within one to two years and see how much do we save in fact, or we see how much would it cost for us to develop and to maintain if we're doing it at home. So, yeah, we are always pretty focused on the evaluation of all these three possibilities.

Speaker 1 34:16

Fantastic and I think just to add on to that, because we've been talking about AI strategy before and the question do nothing, buy, build. That's something you should have some criterias on in your strategy to work from, or you have to have somewhere in your organization something to fall back to. Once those questions pop up and they will pop up.

Speaker 2 34:38

Yeah, I'm now developing an AI strategy for Tina, but I see AI strategy as a part of the digital strategy. So, in fact, whatever we are doing should follow these criteria. So, for now, I'm just waiting to see what are we going to see in our concern strategy, how will it be reflected in our digital strategy and then including our AI strategy. For now, it's not formalized, we are just doing it. But, yeah, I think it's a very good point for our strategy. Thank you for the tips.

Speaker 1 35:14

I have a last question that well, I think it's really interesting because you see a fast-moving development in making fast-moving consumer goods as a sector more data-driven, and I think that's something we talked about already and I think it's interesting. Yet where do you see the future of that? Where do you see this journey going in the development?

Speaker 2 35:36

have a vision I even see it in my eyes how I want to have it. I want to see our supply chain in a like visualized way in real time and I want to be able to make a kind of simulation of if I change this or that, what implication will it have on our supply chain in real time. So, plus, of course, all this built in. So I want to have the data and algorithms optimization algorithms should be interconnected, visual and real time for our supply chain of fast-forwarding consumer goods. For now we are still missing of having the optimization let's say, I make it like in a very I'm saying it in a very broad way Blending of real-time data connected, interconnected data, plus interconnected through the algorithms and optimization. That's what we are missing now and I think it will take maybe five, seven years to develop it. So to see the supply chain optimized, plus possibility to make a simulation on it in real time.

Digital Twin for Supply Chain Optimization

Speaker 1 36:49

I really like that. I did. I have a picture in my head of like a digital twin of your supply chain where you can have like some digital triggers, like on the model. Where you can have like some digital triggers, like on the model, you can change something that has an effect on real life but also the other way around Interesting. I like it.

Speaker 2 37:04

And it should be done on like a Norwegian map, you know, yeah, a Norwegian map, and it should have different layers. On one layer, we have farmers. On the other layer, we have all our factories. On the other level, we have all our factories. On the other level, we have all our customers and we have all the transportation and the production facilities and warehousing. All this in the one.

Speaker 1 37:24

Let's say I will have a room for this and we will have this 3d model on the, on the table well, we are at the end of the conversation, so thank you so much for a fantastic talk and for guiding us through this topic and through the work you are doing in such a profound way. Before we finish, do you have any key takeaways or call to action, something you want the listeners to do or to think of or to take with them?

Speaker 2 37:50

I think, yeah, maybe I would like to share a thought which I finally formulated for myself that AI is just a tool. It's just a tool to make our business better, more efficient, more transparent. So I think we shouldn't overestimate the possibilities of AI and we should use it wisely as something some enabler for our core business. So AI will not bring milk for you from the farmer to the, but it can make it this week a bit more smoother. So always remember your core business and use AI where it brings value for you and for your owners, and for farmers in our case.

Speaker 1 38:33

Fantastic.

Speaker 2 38:34

Thank you so much. Thank you very much. It was a very interesting talk and thank you for inviting.