Deutsche Post AG (DHL) Earnings Call Transcript & Summary

Hello and welcome. My name is Robert Schneider. I'm very happy to welcome you today to the second chapter of our Digital Fridays series in the name of the whole DPDHL IR team. We had our first session last week on DIGI x ESG. For those who missed it, the replay is still available on our website. Today, we will jump right into the core machine room of our operations and look at digitization in terms of automation and data analytics. If you have any questions, please use the Q&A box, which is right on the right side of your screen. The program today will have 4 high-level speakers from our right colleagues around the world. We'll start with Thorsten from our data science team, move over to Sally from supply chain, talk with Axel about Internet of Things and then finish up on the forwarding freight side with Stefano. Again, think about sending in your questions, we'll answer them in about half an hour. And with that, over to Thorsten, please.

Yes. Thank you, Robert. Thank you for the introduction. You already said where I'm from. I'm from the Data Science Department of DPDHL, and maybe to quickly mention what I'm responsible for, I'm looking after the implementation of large-scale AI solutions for our business. And by the way, I'm also Head of GAIA, what GAIA is, you will learn later today in a couple of minutes, actually. So let's jump straight in. Digitalization and automation are really core for our business. It's not like we are doing this some way in an ivory tower and it's not really relevant for our business. Instead, it's really directly touching our profitable core. And the landscape of things we are doing is really huge. We are really touching all parts of our operations, all parts of our business, not just in warehouses, not just on the street, also in our digital touchpoints to our end customers and in our back office processes, just as important for improving our business through automation. I will share with you how we are doing this, giving you an overview on this and diving a bit into artificial intelligence as my area of responsibility. The main sources for incubation of innovation for us are our group centers of excellence. You might have heard of them before, we had 6 of them. So we have topics covered like APIs, like blockchain, like automation of operations or physical automation, and also Internet of Things, IoT, what you will hear more about later today by Axel. The most mature COEs, we had were data analytics and data solutions, where we try to integrate or where we are integrating data analytics and artificial intelligence across our business in all parts of our operations. Now you might wonder why I'm talking about we had 6 center of excellences. Actually, now 2 of them are gone because data analytics and data solutions have transitioned. We are beyond the initial state of incubation, beyond the first improvements in that area. And instead, we now have transitioned the organization into a service line in our global business services, which really reflects the maturity of these topics. Already for roughly a decade, we have been working in those COEs on those topics and just have seen a massive growth of the same. We have seen a growth of the footprint in all our bottom lines and also a growth in the headcount of the departments, growing from a small team to a 3-digit number nowadays. In this service line, we are helping our business to implement high-impact AI use cases across all functions, across all business units. And in addition, we are providing platforms and value-added services across the whole group. I very briefly want to jump in 3 examples just to give you a feeling of what is already happening, what we already have done in the past years. First of all, we have PCT, the product classification tool, as we call it. So if you think about cross-border shipments and what this means for our business, taking care of restrictions of processes, for example, customs is a massive undertaking. And just the simple task of having a proper commodity code, as it is called, based on a product description, is a really intense task. Just think of millions of shipments every day and it's a huge problem for our business if you try to do this. We have built an AI model, which we have rolled out at scale across our business units to support this process, to accelerate all these processes and imply massive impact both on the financial and operational side, and automation of processes is just one aspect. We are also supporting customs audit processes and help in improving the master data quality. If you want to learn more about this and the high impact that we are generating here, you have a chance next week, 7 days from now, where we have a deep dive on PCT in the session on e-commerce. Let's move on to the second example that we brought. And I guess all of you know Burj Khalifa in Dubai as the tallest building in the world. Well, there is something that is taller actually. Right next to it, you might see it, it's the pile of invoices that we are sending every year, sending to our customers and that we have to track where we have to work on processes after that. The pile you are seeing here is a realistic estimate just for one of our business units. And you can imagine how much work goes into this accounts receivable process, therefore. And we are supporting, based on an AI model, our processes, our AR processes and related things by providing an estimate and really accurate estimate of the risk that goes with each and every invoice that we have to process. This is being learned from historical payment patterns for additional information that we put into this AI model in order to support our business and help prioritize on the most relevant cases and not waste time on invoices that will be fine anyways. This has a high impact for us on our free cash flow and other financial KPIs. It increased the efficiency for our accounts receivable and help improving our interactions with our customers. Last example really going to the core of our business, talking about last-mile delivery. We have implemented already 3 years ago, a system that helps in the core planning process each day. So for certain districts where we are going to deliver parcels, the first planning steps are always like you have this bunch of stops that you have to do, not 10, like in this example, but rather 80, 90 or even more stops in a crowded -- densely populated area. And we are supporting here and finding the best route, not best in the mathematical sense, pure driving time minimization. Instead, really the most adequate tool based on all the details that are there in the district. But how should we know when stores are open, when certain recipients are at home and similar stuff, we know it from the experience of our courier. So we implemented an algorithm to learn the tacit knowledge of our courier in order to provide better routes and on top of that, be able to provide accurate estimates of the parcel arrival time fully automated, leading to higher satisfaction of our employees, higher satisfaction of our recipients based on the digital features we were able to introduce for the recipients and a massive boost in efficiency by avoiding nonsuccessful deliveries. So these are just 3 examples. We have hundreds of additional use cases where we are working on with all the business units, as I said before. And we are constantly scaling this, creating significant impact across all our bottom lines and scaling it year-by-year to new levels. In addition to the things we have already been doing for many years now, of course, we are also looking into generative AI. With the release of ChatGPT last November, the additional awareness this year that this technology has gained, of course, this is also a topic for us. But we definitely were not caught off-guard or anything. We are very well prepared to adopt this technology. The new chances that come off this, the evolution of the possibilities that are there in the field of AI. We naturally transition from the use cases we have been doing already using the experts that we have grown over years now and the technologies that we have established over the same time. And with GAIA, we have set up a cross-functional squad across our service line for data and analytics. IT services, corporate strategy and all business units have joined forces in order to get the best out of this technology, managing both the opportunities and the risks that might come from generative AI. And so we are looking into providing a secure access to generative AI tools for our workforce. We are driving awareness and communication enablement across the group because this is really crucial that our colleagues understand what comes with generative AI and what they can do with it. We are driving governance and strategy and, of course, implementing high-impact use cases just think of the opportunities in digital customer touchpoints, automation of processes, for example, and also getting a better access to the massive amount of information we have internally at DPDHL. I want to say it again, this is not a sudden surprise for us. We were well prepared and we are going full speed into a proper adoption of generative AI. With this, I'm already at the end of my session and beyond the topics of analytics, AI, that I have tried to introduce you now. We have really 3 interesting upcoming sessions now. First of all, Sally Miller, afterwards Axel Wiebracht and finally, Stefano Arganese. And with this, I would like to hand it over to Sally. Hi, Sally.

Hello. Thank you. I'm Sally Miller. I'm the DHL Supply Chain Digital Transformation Officer. And I also am the CIO of North America. I thought we'd start out talking a little bit about supply chain. Supply chain is the largest -- the DHL supply chain is the largest provider in the industry. We're 2x the size of our nearest competitor. And we operate in approximately 50 countries worldwide. We have -- we launched and accelerated digitalization program over 5 years ago to focus on the key technologies that we believe will have the most significant impact in our operations. And I'll touch on a couple of these today. Assisted picking is one, unloading technologies, data analytics and indoor robotics transport. We have impacted over 90% of our sites with the go-live of over 6,000 projects, and we plan of having touched 100% by the end of the year. So we have a process to manage this because the rate of change in our industry is an epic pace. The amount of venture funding that has gone into the industry and solutions in the past couple of years has really sped up the need for these solutions in the supply chain industry. We've worked with over 100 partners, and these type of ideas -- innovation ideas come from existing customers, from industry trade shows, from start-up companies and the venture capital firms that fund them our associates. So we get all these ideas into a funnel, and then we do a proof of concept with those that we think will have an impact and improve the efficiency of our sites and then they move to the product stage and then we scale them from there across our sites globally. One example of a robotic arm, which is also in this use case an unloading technology is from a company called Boston Dynamics. You might know them from their dog and dancing robot in the beer commercials. But we deployed Stretch this year after codeveloping with Boston Dynamics the solution over a 3-year time period. We believe that this solution can do other things in the warehouse. So we're just starting with Stretch, but plan to roll it out to more sites in North America by the end of the year and then scale it to Europe and other regions beginning next year. What you see in the video here is from a company called Fox Robotics, and this is, again, unloading, but it's unloading pallets from a trailer autonomously. We have approximately 25 sites live unloading pallets with Fox, and we'll continue to scale this in North America and launched to Europe in the first quarter of next year. This is a great solution as well as Stretch previously because it's a very manual laborious part of the process to unload trailers not the most desirable position. So if we can automate it, all the better for us. Probably what has had the biggest impact in our business to date is the rollout of assisted picking box. These technologies have been more mature and earlier in the market. So we've been able to work with a couple of different vendors and scale this solution. So it's very attractive to our associates because it reduces the amount of travel time that they have. It reduces the training when we onboard new associates and it also improves picking accuracy. We've deployed over 2,500 assisted-picking bots across our sites globally. And these continue to scale well for us in sites where we're picking each, so individual units. Also, machine learning, AI, data analytics, we are using them significantly in our business today. As mentioned previously, ChatGPT has focused the spotlight on AI, but we've been using AI as an example to improve our safety in our warehousing and transport operations. And we've been using machine learning to take data from various process outputs and learn and predict where we're most likely to have an error so we can focus on those locations, improving the accuracy so that we have less error rates within the 4 walls and better outbound shipping accuracy. You saw the videos of the different assisted picking bots and robotic arm technologies. Those use computer vision and AI to navigate and to retrieve the item for that specific use case. And on the generative AI topic, we are finding several use cases in our back office processes that we can automate things that are pretty manual. One example would be potentially integrating files and exchanging data between applications and with our customers to be able to automate some of that as well as the documents that we produce looking at that as well. So of our 2,000 facilities that we have across the globe, we are using this technology and see it scalable and being able to apply to more use cases going forward. So the value that we bring to our customers is to integrate this technology with our core execution systems and get the most out of the technology and scale it. We believe that our leadership position in the industry and the fact that we have been the leader in innovation. We were recently recognized by Gartner as being the most innovative 3PL. We believe this is a growth opportunity for our business. It's much harder for companies to keep up with the technology and do these services themselves. We think there will be a rise in outsourcing because we continue to demonstrate that we can deploy the technology in the right use case and orchestrate multiple technologies in the same facility. So it's a very exciting time to be in the industry, and we're very pleased with the key partners that I talked through today, how our relationship advanced and how we've worked together to advance their products so that we can perform better in our sites for our customers. With that, I'll turn it over to Axel, who will talk about how we're using IoT within the group.

Good. Thank you very much, Sally. Hello, everyone. My name is Axel Wiebracht. I'm part of corporate development, and I'm leading the Center of Excellence for the Internet of things for IoT at DHL. For us, IoT always consists of 3 components. We've got the IoT devices, the sensors, which are actually measuring, for instance, environmental parameters which are determining the position of goods, of shipments of assets and so on. The second component is connectivity is the network to actually automatically capture all that sensor data as well as to make sure we can make use of that data in real time. And number three is the Internet. It's more the application, it's logic, it's the software and platforms, which we are using to make use of that data to interpret that data into relevant business events. What I would like to do within the next few minutes is share with you how we are leveraging this technology to automate, to support, to improve some of our core operational processes in logistics. But let's start with looking at our logistics trend radar. This radar is published on a regular base by our customer solutions and innovations function and is giving an outlook to trends from the technology, from the business and social sector, which -- where we believe they will or are having and will more and more have an impact to the logistics industry in the near future. IoT was on that radar for quite a time -- for quite some time, but it's not there any longer, simply for the reason that IoT for us is something which we already use on scale on a day-to-day basis. So it's there, it's used by operations, and it has been deployed in scale. It is not a future trend any longer. However, you will find some trends on that radar, which are enabled by IoT or where IoT possibly will have the possibility to improve the footprint even further. These ones are here in bold and to give some few examples. There is the area of smart labels and next-generation packaging. The IoT devices I talked about are getting smaller and smaller, more capable needing less power and they start to be actually be embedded into labels and into, in particular, reusable packaging. So that can improve the possibilities of using IoT. Another example is next generation wireless. I mean recently, over the course of the last years, connectivity and IoT design networks have evolved quite a lot already. So that typically connectivity is not really a big challenge any longer. However, with the further deployment of, for instance, 5G and narrowband IoT with the further capabilities of BLE, Bluetooth Low Energy, it will be even easier and at lower cost to have standardized connectivity available in each and everywhere. So technology, devices, networks are improving, which will help us to further grow IoT. However, let's now look at where are we actually using IoT in logistics as of today. There's basically 3 main areas. There's the area of smart operations. To give you an example on that one, we have started to equip our sorting machines with predictive maintenance sensors. So a lot what we do in supply chain but also in post and parcel in global forwarding, in e-commerce is sorting. And in those areas where we are using highly automated sorting machines, we, of course, want to avoid any disruptions, any outages during our operation times. So we're putting predictive maintenance sensors on it to avoid these outages and to actually identify repair needs as they occur, which we have identified the noise, like with the engine of a cars typically a good -- very good indicator for something to be repaired or for something to possibly go wrong in the near future. So avoiding outages is one of the benefits. The other is we want to move. We want to shift from regular scheduled maintenance on each and everything rather to event-based maintenance and repair on these machines. Then in the area of shipments and customer interactions, we do, of course, have more and more customers who have a need, who have got a demand for real-time shipment tracking. They want to have real-time visibility of the exact location of their shipments and of the conditions. Customer with temperature-sensitive goods, with time-sensitive goods or with very high value goods who really want to see it from origin to destination. In that area, we have highly standardized the technology behind it, and we have integrated that data into our customer-facing portals as well. So supply chain customer for instance would be able to see in my supply chain, the real-time position of the shipments, where supply chain is doing transportation as well as -- in addition to all the other shipment relevant data as well. And then let's come to the area of digitalization of operational assets, which is probably the area of IoT with the highest scalability and the highest volume. So let's look at that in a bit more detail. What do we mean by digitalization of assets? While in very simple terms, we are moving, we are transporting goods from A to B to C. And of course, all these goods, these shipments are typically containerized. We're using totes, we are using stillages, we are using returnable transport units -- transportation units like roller cages. We are using the universal load devices like in air cargo and so on. We literally operate millions of assets in order to transport these goods efficiently. So it is beneficial and it is very vital to us in some areas to really have full visibility about where these assets are within our network. And that's why we've equipped several hundred thousands of these assets already with IoT trackers to have that visibility so that we know where they are and whether they are in use at the moment. That information helps us to balance these assets better across our network. It helps to improve the utilization of these assets. We don't need to invest anything into overcapacity or backup capacity, but we can better manage the existing fleet of assets. And of course, we possibly also avoid out of stock situations or avoid that we forget to pick up assets somewhere at a customer side or somewhere or have some other loss reasons. So managing the assets on a day-to-day base, in particular during peak time has become much easier with these real-time trackers on the assets. And to give you one more concrete example on that one, we can do that on a global, on a regional, on a country scale, but we can also do that on a scale, for instance, of a yard or of an apron like the airport of the Express Hub in Leipzig. We are using a few thousand assets actually to run the operations there. So to do the loading and offloading of the aircraft and all the operations around it. With equipping these assets with GPS real-time trackers, we and even the operators, the apron, have full visibility about where the next available assets like ground support equipment types like dollies, trucks, nose lifters, high loaders and so on where they are pick that up efficiently. So we use search and find times a lot, and we avoid that we are possibly having any delays because equipment is not at the right place at the right time. Implementing these solutions on certain facilities in certain regions is I wouldn't say easy, but is doable. Our challenge, like always is, of course, how can we run that at large scale? And in order to facilitate that, we have over the last 3 to 4 years, invested into building up a DHL IoT platform. This is a highly standard platform based on cloud technology, which we used to deploy these use cases in a standardize way. It is connected to a standardized portfolio of sensors as well. So sensors for all the different use cases I just described and we have an ongoing message stream of this real-time data about our operations into that platform. In simple terms, this platform is then translating these messages into useful business events. What is where, when, where do we have threshold breaches and so on. And then that data is further on being sent into our customer-facing portals like supply chain, for instance, or it is used by our internal operation colleagues into their internal systems. So that platform, together with the sensors, is actually something which helps us to have a standardized and hence, highly scalable approach for deploying IoT across the organization and across the countries. And that with that, I would like to hand over to my next colleague to Stefano from Freight for the next update.

Thank you, Axel. Yes, I'm Stefano Arganese, Chief Transformation Officer of DHL Freight. DHL Freight is the land-based transportation arm of the Global Forwarding division within our group. I'm a long time with the company, more than 25 years by now. So I have been in different roles, and now we're very excited to be driving transformation at DHL Freight. Today, we'll have a quick look, we can move already to the next slide. We can have a quick look at the digitalization and automation of DHL Freight. We do see automation as a game changer in our business. And helping streamlining operations, increasing efficiency and filing also allowing for growth of our business. I will talk with 2 examples about 3 key aspects to consider when implementing our automation in our business. And one is automation drives, of course, in an environment where business processes are harmonized. So that, you need to go through standardization processes and having the possibility to then integrate seamlessly integration into those existing workflow. Automation also heavily benefits from unified and standardized data. Therefore, accretive and integrated data ecosystem facilitates a scalable automation initiatives such as those that I'm going to talk about. Of course, also well-defined rules and algorithms needs to be integrated in our tools, and I think will be good examples of what I'm going to talk about right now, which will be -- 2 examples, 1 is EVO, our new transport management system, we are deploying across streets and RAPTOR, which is our transportation planning tool. So these 2 tools will help us automating freight. Now starting with EVO. EVO is, as I mentioned, our new transport management system. It covers all our end-to-end processes, starting with order entry, dispatching of all the transportation legs, customer and order invoicing. So it goes into admin as well internal cost allocations and also covering all the master data that we need in our entire network. This is the DHL Freight largest transformation projects in history and it's replacing a number of TMSs that we have inherited over the past years. So we are now live with EVO in 29 countries already. We have a high number of users and number of orders have been already developed in EVO until now, and this just shows that indeed EVO fits in our business and our business needs and is highly reliable as well. The introduction of one single TMS, which we also compare to like speaking one language across all the countries that where we operate in, comes with an introduction of harmonized processes, as I said before, and also data structure, which has been key -- which have been key elements, key aspects to be considered when we started with this project. And we have real-time transparency on all data entered in EVO. And as an example, when we enter a shipment in our original terminal, the destination terminal already can see the same and can consider whatever is coming maybe next day for their planning and dispatching of their local operations. We will operate on one database, and this gives us the possibility to indeed have the real-time transparency at any moment. So this is also positive because EVO is automatically pre-planning with the algorithm, all the transportation legs as soon as the shipment is entered based on the characteristic of the shipment and the preplanned routing, then of course, the tool can preplan also the shipment's destination. Of course, this needs to be reviewed, but this is a huge support to our dispatches in doing the planning on their own. We also developed our new standard application both for cross-stocking as well as for pickup and delivery and this application gives all the information needed to our operators in order to know where to position the shipments within the terminal, where to load it and then based on that automatically triggers loading list, for example, all the necessary documents in order to execute the transportation. The software is developed with 2 main principles in mind. One is that operation drives administration. I said before, with high volume that we handle, it is important that people who can make the decision in the terminals as quick as possible, and then the admin derives the contents of that by producing the necessary documents. By doing that, system is also producing a multiple of data that we used to have with the old TMS, which is then opening up very many new opportunities for big data and apply this in different AI projects, of which, one example, is my next one. Then as the second principle for developing -- while developing EVO is do it right the first time. So enabling data quality and proper data structure. And it is important and as an example, I can give you the -- when we get customers CDI, customer orders via interfaces, those are always automatically geo-coded and flagged in case it's not matching existing addresses. Customer supplier tariffs are all in the system, which then also allows to automate invoicing processes or checking of the invoices coming in from our suppliers. The high degree of automation is only possible as the new transport measure system is made to perfectly fit our needs. It is done by us. It is developed internally by own team of developers. Coming to the benefits of EVO, as I said, one unique TMS. Of course, quite improved as there is more automation and less mistakes possible. The IT around costs -- are lower than what we used to have with multiple and older systems. Coming to productivity, it's also increased. And as an example, I can say, the EDI out of confirmed number of shipments is 10 percentage points higher than before now reaching 90% plus in total. Automated invoicing also improved significantly with more than 27 percentage points, now reaching 70% of completely automated invoicing. And this also allows them to reduce DSO because the more you automate invoicing, the less mistake you have and the quicker you can be in sending out invoicing and collecting money. Moving on to the second example, which is RAPTOR. RAPTOR is an application that is tailored to support the daily labor-intensive process of dispatching. Dispatching is a rather complex operation to do and our dispatcher while super good and very experienced. It is just impossible to consider everything in a very short time frame when it comes to dispatching. I don't know, for example, in the case of RAPTOR, more than 1,000 shipments carloads per day. Dispatcher needs to consider undocumented knowledge, truck characteristics, service hours, customer plans and ours, time windows, break regulations, stackability of the goods. So all these things have to be considered while dispatching in a quick manner, more than 1,000 shipments in this example. Relying on a dispatch only is obviously increasingly complex, and that's why RAPTOR significantly use this complexity of the task and support dispatcher by calculating and providing route suggestions, which, of course, then be huge by our own dispatcher and also allow us to visualize the mapping, which would restructure take, but also visualize with the 3D models, what we call, a bit of a tetris how to load the truck with different kind of goods and packaging. RAPTOR can easily be integrated into the daily dispatch standard workflow and in our TMS. And because of that, it also facilitates the change of management that it comes with the introduction of the new system. As it is normally the case with artificial intelligence, RAPTOR also relies on close feedback loops with the users. So basically, there are obvious generated and proposed planning, but then dispatcher use those and then finally come to the final. Those are fed back into the system, so the system can learn for the next cycle how to improve the preplanned 2 things. So with that RAPTOR, we have now implemented in Germany, and we are using it for one of our largest automotive customer and is used, as I said, for about 1,000 shipments every day and it optimizes standard level growth, which is basically [indiscernible]. So without the usage of our own terminal, but just having a number of pickups of shipment to do, a number of deliveries to do. When it comes to the next slide on the benefits of RAPTOR, one clear measurable benefit is time saved in the dispatching activities. So you see there, we reached 92% improvement of time spent when dispatching. And that is just time. Then it comes with, of course, better planning and more precise planning, which also allow us to have better loading factors so we can use less trucks or put more goods into the same number of trucks. It also improved the driving distance so by decreasing it in optimizing the routing and that also comes with a reduction of CO2 emission, which is rather substantial, as you can see here. Machine learning can be improved by applying continuously supervised learning as these data that we are using are label data. And we also saw that -- when it comes to which neural network we are using here, we're using the convolutional neural network, which seems to be the best model for this problem. By the way, worth to mention, this tool is also completely developed within our company with the support of our data scientist colleagues that were presented at the beginning of this call. We are very excited personally, and very excited to soon complete the rollout of EVO in our business unit and apply RAPTOR to a much larger scale of use cases going forward. And with that, I think we are at the Q&A session and -- will give it back to Robert now.

Thank you. Hello, everyone. As promised, a lot of information in a very quick time frame. So a lot to digest. If there is -- there are more questions, please keep using this Q&A button, which I see you are doing out there. And we'll go right away into your first questions. And the first one who would have expected it is on Gen AI? What's the impact of Gen AI in logistics in the coming years? And which part of your business will benefit the most perhaps, Thorsten, if you can have a word from the broader business perspective to get us started?

Yes, absolutely. I think the short answer is it's going to have a big impact overall. Yes, it's still a bit hard to predict where the biggest levers will be, but especially in the back office task, but also looking at the overall workforce, which are using, yes, doing office tasks, for example, right? I think we can expect a big lever, a big improvement in efficiency there. And this -- we are absolutely engaged in this. And as I said, it's a bit hard to predict exactly the biggest value will be.

Okay. Second question, what are the challenges when scaling robotic solutions further? So I think Sally, that's probably over to you. Where do you see the future of automation?

So the first question, some of the challenges that we run into when scaling is a lot of the companies we work with are start-up companies and don't have coverage in all regions that we operate. We also run a very diverse profile of operations. So not one solution fits all of our sites. So one use case might scale much more significantly than another. So I would say those are the 2 biggest challenges. And the future of automation that I see in our business is more flexible solutions that don't require significant capital investments and that have the ability to be deployed in several different use cases. Stretch is a good example of that. We can use Stretch on old cartons. In the future, it can palletize cartons. It will be able to pick cartons for orders and also load trucks on the outbound side. So products doing more things within our 4 walls will definitely be in the future.

Excellent. As we talk about challenges, another challenge, I think, well known on that is the war for talent and then finding employees in that space, IT specialists and reskilling existing employees. So perhaps also again, to Thorsten, perhaps also from your experience. How does DHL do in finding getting the right talent and reskilling the other ones? And perhaps one again, Sally, for you on how did employees take the experience of getting new technology into their business? You talked a lot about what you already did with the 6,000, I think, examples that we have live, how do the employees react to that?

Would give it a quick start. So I think we have the benefit that we have a strong training and development culture in our company. We have a whole certified organization that takes care of the upskilling of existing workforce and we leverage this in every corner of what we are doing. Just to mention for analytics and AI, we have set up training curriculum, dedicated to a large target audience. We have more than 40,000 alumni of dedicated trainings on analytics and AI already in our company ranging from managerial level over the broad workforce until over to really expert trainings on the tools and technologies that experts need. And I think this is the culture that applies to all the areas that we are talking about today maybe some of you wants to add to IoT or to robotic automation. I think this is the culture that we are living, which is a big benefit going forward.

Yes. To add on that, there is -- clearly, there is a war for talent and finding good talent in particular for new positions is, of course, always a big challenge. I think at this moment, at least for the IT part, we are well staffed. We found a lot of talent, I mean, particularly in the area of cloud and cloud-based development. That is a challenge, but also and what you just said with internal training and internal training and skill up programs, I think we are doing very well.

Excellent. Sally, on the experience of existing employees perhaps?

Yes, very positive. There have been a lot of studies that people like to work in environments where more technology is deployed. And the experience our associates have had made their jobs easier. I mentioned less travel time and being able to onboard faster. But it also gives us the opportunity to create different roles like managing robots, intervening when there's an issue, creating better roles and reducing the need for the more mundane, monotonous positions in the facility. So it's -- it's been great to hear people say that they go home and tell their kids, they're working with a robot. And we also think that the turnover in sites where we have technology deployed is less than a more manual operation.

Excellent. Next one is mentioning freight forwarding. Stefano, perhaps you can have a go at it from the freight side and as much as you can also on the forwarding side. I read it out, you see the role of shared service centers will reduce with the use of AI in freight forwarding, how do you measure throughput in a shared service center? So the impact of AI on shared service center, Stefano?

Yes, definitely, there would be an impact, particularly with generative AI on shared service center as well. I mean the throughput is pretty easy to measure. It is productivity, it is speed of service that we get from the shared services and of course, is efficiency of the process in there, right? So I think we have also our own large experience with these tools nowadays, and they are indeed go in the direction of eliminating the boring part and the most -- as a repetitive part of the work. And this is benefiting, I guess, I would say, from the business for forwarding of freight or any kind of business, to be honest. So I think there's bright future also from that point of view in terms of additional improvements that will help our business overall.

Yes. So probably can help shared service centers become even more efficient, but also allow us to put even more services into the shared service centers, I guess?

Exactly. So that would allow to, let's say, again, with repetitive and administrative task more into the shared service center knowing that we get a faster and more efficient way of producing those tasks outside the core of the business into the shared service centers.

Excellent. Next one is a technical on IoT as far as I can judge on it. What role does RFID play in IoT? What's the potential for the future? Axel?

Yes, very good one. I mean, RFID is first of all a very old technology, hasn't been picked up by the logistics industry for a while, but I think it's from the '80s. And within the recent years, it's more and more being picked up. We are using RFID since some years already. And this possibly needs to be seen in competition to optical barcode scans. So we use it in those areas where the RFID scans as opposed to optical scans are really far more efficient, which is not everywhere the case. But if you imagine, for instance, the station where you need to do a physical inventory count of order shipments which are in that stations and that we want to replace barcode scans in there, it can be very efficient to use RFID. We also use it in supply chain, for instance, with customers who come with RFID-labeled goods already where we can just leverage on that one. So it can be a good efficiency driver. We are using it already. Would I believe it will replace the optical barcode scan each and everywhere, I would believe not because the cost for labels and infrastructure needs to be taken into account as well.

Okay. Can you perhaps explain shortly how we go about allocating budgets to kind of new projects? If you come across that question on, do I apply technology, which technology, how to process works internally and if it's possible in the shortest answer?

Well, I guess, like each and everywhere -- it depends, be it goes through a calculation -- business calculation of at the end of the day, finding out in each and every deployment, whether we expect benefits from that deployment because like with RFID and optical scans, technology might be of good fit for a particular case. And you think it's -- well, it works the same in the next country that might be a bit different. And there you'd rather check again whether you would really get to the same benefits as we're the first deployment. So it's a case-by-case decision typically.

Perfect. One still for the Stefano and the classic one, which we probably expected. You talked about EVO being developed internally. While I think the financial community knows a lot about CargoWise being deployed on the folding side. Can you also, in the interest of time, quickly explain why freight, it was more useful to develop internally versus going CargoWise for the in-house forwarding?

Sure. So when we discuss internally how to -- what strategy we want to follow on these questions, we went through a cases from external suppliers as well as internal, different size of external suppliers, small, big. And finally, there is no real out-of-the-shelf product that would fit our rather complex business environment, a company that develops CargoWise itself and did not at that time at least having anything covering growth rates processes. And that's why we decided to actually buy a rather small TMS core of it and then keep developing it ourselves internally based on our own needs. So I don't think there is right and wrong, but it's more right for us. It was really the best not to waste time to do that from scratch, but to buy a core and then basically developing according to our needs, which we define internally with our own people and developing our own developers, which has been in Germany because we build that team of developers internally with talent that was mentioned before. But I think we are extremely happy now with that choice because the tool is really allowing our business to be perfectly automated in the way we want to see it and we can do it the way we like it.

Perfect. I think we have time for 1 or 2 last ones. I'll read out the next one, as hyper-automation can significantly improve the decision-making process by delivering real-time data, how is your company leveraging this potential? I think, from a Thorsten again on data?

Yes, absolutely. I mean this is a really good question, making decisions based on data is key and getting good decisions also going beyond individual knowledge. And we are very much invested in that area from all perspectives. Real-time data maybe also refers to technology where we are really investing and already have the systems in place to manage real-time data, IoT data just as one example, but also the other kinds of live data that we need. And then there are many other dimensions like enablement of our colleagues, driving data literacy in our workforce and also sharing things in a communication way. So many dimensions that we are tackling, we're also using the phrases of know your data, own your data and use your data in order to make clear to everybody, data is key for our business. It's a core asset that we have to grow and work on to make it more valuable going forward. And this is really part of our business, part of our undertakings and very much part of our future.

And actually, the perfect lead over to the session, we'll have next week because I know that 1 of the 3 sessions next week will be from the colleagues from DHL e-commerce Solutions who will talk then 9 minutes about how they use data with different real-life applications on their pricing and on the routing perspective. So with that, I say thank you to the 2 speakers next to me. Thank you to the 2 colleagues on the screens out there. Thanks for joining on a Friday afternoon. Thanks for everyone out there who has joined today and tune in next week again for the next DIGI Friday session. Thank you. Bye. Have a nice weekend.