PowerPoint Presentation

[Audio] Welcome to module 2 "Digitalising the agri-food value chain, from farm to fork".

[Audio] In Module 1 we saw that goods, money and information flow through the agri-food chain. In this module we focus on the third flow, information, and ask how digital technologies change it at every stage, from the seed supplier to the consumer's plate. A simple way to organise the technologies is as a layered stack. At the bottom is the sensing layer: devices that capture data from the physical world, such as soil probes, weather stations, machinery sensors, cameras, satellites and drones. Above it is connectivity, the networks that carry data from the field to the cloud. Then comes data storage and integration, where information from different sources is brought together on platforms. Above that is analytics, where statistics and artificial intelligence turn raw data into insights. At the top are applications and interfaces, such as farm management software, mobile apps and dashboards, which turn insights into decisions and actions. This stack does not belong to one actor. A sensor may be installed by a farmer, the platform run by a technology company, the analytics provided by a research spin-off, and the final recommendation delivered by an advisor. Every layer is a place where different actors meet, which is why this topic connects so directly to business ecosystems. A major review by Wolfert and colleagues found that big data applications in smart farming reach well beyond primary production and are influencing the entire food supply chain. The review also reports that several authors expect big data to shift roles and power relations among the players in food supply chain networks. Keep that last point in mind throughout this module. Technology does not only make existing activities more efficient. It can also change who has information, who has influence, and who earns what. Over the next ten minutes or so, we will follow the product through five stages: inputs, production, post-harvest and processing, logistics, and retail and consumers. At each stage we will ask the same three questions. What data is generated? Who can use it? And what value does it create, or at least promise to create? I also want to be clear about what we are not doing. This is not a catalogue of gadgets. Technologies matter only insofar as they solve a real problem for a real actor, and many digital initiatives fail precisely because they start with the technology rather than the problem..

[Audio] Everything digital starts with data capture, so let us look at the sensing layer. It is useful to distinguish proximal sensing, meaning sensors placed in the field, on machinery or on animals, from remote sensing, where observations come from a distance, from drones or satellites. Proximal sensors measure soil moisture and temperature, weather conditions, leaf wetness, nutrient levels, water flow, milk yield, animal activity, and many other variables. Modern tractors, harvesters and sprayers also act as mobile sensing platforms, recording position, speed, yield and application rates as they work. Remote sensing offers a different advantage: coverage. Drones can map a field in detail on demand. Satellites can observe entire regions regularly. The European Copernicus programme is a good example. The Sentinel-2 mission provides multispectral images with a spatial resolution of up to 10 metres, and the two-satellite constellation achieves a revisit time of five days. Closer to Europe's mid-latitudes, revisits can be even more frequent. Because these data are openly available, a small advisory company or a university can build crop-monitoring services without launching a single satellite. That lowers the entry barrier and enables many new business models, which we will discuss in Module 4. Data from all these devices must travel somewhere, so connectivity matters. Options include mobile networks, short-range radio, low-power wide-area networks designed for battery-operated sensors over long distances, and satellite links for remote areas. The right choice depends on how much data is sent, how often, and how much power is available. Now the honest challenges. Coverage in rural areas is still uneven. Sensors need power, calibration and maintenance, and a poorly maintained sensor produces misleading data, which is worse than no data. Devices from different manufacturers often use different formats, which creates the interoperability problems we will discuss in Module 3. And costs can be significant for small farms. For students and lecturers, one important lesson is that data quality starts here. A very sophisticated algorithm cannot compensate for a sensor installed in the wrong place or never recalibrated. Let me ask you to think about your own context. What is the most valuable piece of data a farmer in your region does not have today, and what would be the cheapest reliable way to obtain it?.

[Audio] Data has little value until it supports a decision. This is where analytics and artificial intelligence come in, and it helps to think in three levels. Descriptive analytics tells us what happened, for example a dashboard of last week's soil moisture. Predictive analytics estimates what is likely to happen, such as a disease risk forecast or a yield estimate. Prescriptive analytics recommends what to do, for example when and how much to irrigate. Machine learning has become a core tool for the predictive and prescriptive levels. A widely cited review by Liakos and colleagues, published in the journal Sensors in 2018, surveyed machine learning applications in agricultural production systems and organised them into four categories: crop management, livestock management, water management and soil management. That framework is a useful map. Within crop management we find yield prediction, disease detection and weed detection. In livestock management we find animal welfare monitoring and production forecasting. Water and soil management include irrigation scheduling and soil property estimation. Computer vision deserves special mention. Cameras on drones, tractors, robots, packing lines and even smartphones can recognise symptoms of disease, count fruit, grade quality or identify weeds, making it possible to act on individual plants rather than whole fields. Most farmers do not interact with algorithms directly. They meet them inside tools: farm management information systems that record operations and costs, decision support systems that translate data into recommendations, and mobile apps for advice, alerts and compliance reporting. The usefulness of these tools depends less on the sophistication of the algorithm than on whether the interface fits the farmer's workflow and the advice is credible. Now the limits. First, data quality: models trained on poor or biased data give poor or biased results. Second, local calibration: a model built for one region, variety or climate may perform badly elsewhere, which is why local validation by universities and advisors matters. Third, explainability: a farmer who does not understand why a system recommends an action is less likely to follow it, especially when the cost of a wrong decision is high. And fourth, trust: advice linked to a commercial interest, such as an input supplier recommending its own products, raises legitimate questions. The key message is that analytics is not magic. It is a way of using data to support human judgement, and its value depends on the surrounding ecosystem of validation, advice and trust..

[Audio] Now we follow the chain, starting with the actors that supply seeds, fertilisers, crop protection products, feed, veterinary products and machinery. Digitalisation is changing their role in a significant way: many are moving from selling products to selling outcomes or solutions. A clear example is precision application. Instead of treating a whole field uniformly, farmers can use maps and sensors to apply seed, fertiliser or crop protection products at different rates in different parts of the field. The supplier may provide not only the product but also the software, the maps and the advice. The business logic changes from "sell more kilos" to "help the farmer use the right amount at the right place and time", which can reduce input costs and environmental impact when it works well. Distribution channels are also changing. Online ordering, digital catalogues and subscription services are growing, but local dealers and cooperatives remain very important. Research on precision agriculture adoption notes that input dealerships are often farmers' first point of access to these technologies, and that the dealers shape what is locally available and what farmers experiment with. This is a good illustration of ecosystem thinking. Dealers, advisors and cooperatives act as intermediaries who translate between technology providers and farmers. Digital advisory services are another growing area. These can be public, private or cooperative, and they may combine remote monitoring with field visits. In many regions the advisory network is the main channel through which digital tools reach small farms, so the digital skills of advisors are as important as the skills of farmers. This is precisely the kind of capacity that TALLHEDA and similar initiatives aim to strengthen. There are also risks. Lock-in is one: when a farmer's data and machinery are tied to a single supplier's platform, switching becomes costly. Another risk is bundled data collection, where using a product means allowing the supplier to collect data that may later be used for other commercial purposes. And there are conflicts of interest when advice is linked to product sales. We will return to these governance issues when we discuss data sharing and codes of conduct. For discussion: if you were advising a cooperative choosing a digital service for its members, what three questions would you ask the supplier about data and independence?.

[Audio] Primary production is where most digital agriculture attention goes, so let us look at what is happening on the farm. Precision agriculture is the best-known approach, and it is important to understand that it is not one technology. Research by Lowenberg-DeBoer and Erickson, which reviewed adoption evidence worldwide, concluded that precision agriculture is a toolkit from which farmers choose what they need. They also note that use on non-mechanised farms is almost non-existent. That reminds us that adoption patterns differ strongly by farm type, size and region. Other studies report that simple, low-disruption tools such as satellite guidance for machinery are widely adopted, while more information-intensive tools such as variable-rate technologies have been taken up by far fewer farmers. Automation and robotics are expanding. Robots for weeding and monitoring, automated milking systems, automated feeding, and harvesting assistance are all moving from research to commercial use, especially where labour is scarce or costly. In livestock, precision livestock farming uses sensors, cameras and algorithms to monitor individual animals, supporting early detection of health problems and improved welfare and productivity. A concept that is gaining attention is the digital twin. According to Verdouw and colleagues, in a 2021 paper in Agricultural Systems, digital twins are very promising for taking smart farming to new levels of productivity and sustainability, and the authors propose a typology of different types of digital twins along with a conceptual framework for designing them. In simple terms, a digital twin is a virtual replica of a physical object or process, such as a field, an orchard, a greenhouse or an animal, that is continuously updated with real data. It can be used to monitor what is happening, to simulate what would happen under different decisions, and in some cases to control the physical system. Digital twins were also among the topics presented in the TALLHEDA brainstorming visit at the Agricultural University of Athens, which shows how this concept is entering education and research. Controlled-environment agriculture, such as greenhouses and vertical farms, is particularly suited to digital control, because conditions can be measured and adjusted precisely. It is also energy-intensive, so the economics need careful analysis. The main message of this slide is that digital production tools create value only when they fit the farm's scale, crops and skills. A technology that is profitable on a large arable farm may not make sense on a small mixed farm. That is why we need to understand the whole ecosystem of services, shared machinery and cooperation models that make technology accessible..

[Audio] After harvest, the product enters the part of the chain where losses can be large and where digital tools can make a measurable difference. FAO's State of Food and Agriculture 2019 reported that around 14 percent of the world's food is lost after harvesting and before reaching the retail level, including through on-farm activities, storage and transportation. The same report found that harvesting is the most frequently identified critical loss point for all types of food. The figures vary greatly by region, commodity and stage, but the message is clear: a significant amount of the value created on the farm is lost before the product reaches the shop. Here digital technologies work in three main ways. The first is monitoring storage conditions. Temperature, humidity and gas sensors in silos, cold rooms and warehouses allow early warnings when conditions drift out of range, so action can be taken before spoilage spreads. The second is sorting and grading. Cameras and sensors can assess size, colour, defects and internal quality faster and more consistently than manual inspection, helping to match each batch to the right market and reduce waste. The third is better planning, using forecasts of harvest timing and volumes to coordinate labour, transport and storage capacity. In processing, the food industry is already highly automated, but digital integration across the chain is often limited. A review on digitalisation in the agri-food chain notes that the food industry is highly automated yet has not fully used the potential of digitalisation to connect the different steps of the chain. This is a real opportunity. If a processor has access to data about how raw material was grown, harvested and stored, it can plan production more precisely, adjust process settings to raw material quality, and document compliance more easily. Conversely, if farmers receive data about how their product performed in processing, they can adapt their practices. This feedback loop is one of the most valuable features of a digital chain, but it requires trust, shared data formats and fair rules about who can see what. A processor may be reluctant to share quality data if it fears this would weaken its negotiating position, and a farmer may worry that detailed production data could be used against them in price negotiations. Reflection: in your own product example from Module 1, where do you think the largest losses of value occur between harvest and shelf, and what information would help reduce them?.

[Audio] Logistics is the nervous system of the chain: it connects production to markets. For perishable products, speed and condition control are decisive, which makes this stage a natural place for digital tools. The central idea is visibility. With tracking devices and sensors on pallets, containers and vehicles, actors can know where a shipment is and what conditions it has experienced: temperature, humidity, shocks, door openings. If a cold chain breaks, an alert can trigger action while there is still time to save the load, and the data provides evidence of what happened, which matters for insurance claims and disputes. A research stream from Wageningen has formalised this idea under the concept of virtualisation. Verdouw, Wolfert, Beulens and Rialland, in a 2016 paper in the Journal of Food Engineering, describe how, in an Internet of Things environment, food supply chains can become self-adaptive systems in which smart objects operate, decide and learn autonomously, and they propose an architecture for the information systems that enable this. Their case study concerned fish distribution. The intuition is simple: each physical product, batch or container has a digital counterpart, a "virtual object", that carries its identity, history and status through the chain. Beyond tracking, digitalisation changes the paperwork. Electronic transport documents, digital proof of delivery, and e-invoicing reduce errors and delays, and shared platforms can match available transport capacity with demand, reducing empty runs. For small producers, joining such platforms can reduce the cost of reaching distant buyers, though it also creates dependence on the platform operator. Why does this matter for food loss? The FAO analysis I mentioned earlier found that, for fruit, roots and tubers, packaging and transportation are also critical loss points. Better condition monitoring and handling therefore have a direct link to reducing losses. Finally, sustainability reporting is increasing the demand for logistics data. Buyers and regulators want to know the carbon footprint and conditions of transport, and a digital chain can provide this data automatically rather than through manual reporting. There is a recurring theme here. The technology to track goods already exists. The harder questions are organisational: who pays for the sensors, who owns the data, how do systems of different companies communicate, and how are benefits shared? Those questions will be central in the next module on data and interoperability..

[Audio] Traceability is one of the most discussed applications of digital technology in food chains, and it is worth separating what the law requires from what technology makes possible. In the European Union, Article 18 of Regulation (EC) 178/2002 sets out the traceability requirements, often summarised as the "one step back, one step forward" approach: operators must be able to identify who supplied them and to whom they supplied their products. The regulation does not specify exactly what information must be kept, and for food of animal origin there are additional rules in Implementing Regulation 931/2011. This legal baseline is deliberately minimal. It helps authorities contain a food safety problem, but it does not by itself give consumers or buyers a detailed picture of how a product was produced. Digital technologies make much more detailed traceability possible: batch-level records, real-time updates, and information shared across the whole chain rather than only between neighbours. This is attractive for food safety, for faster recalls, for proving origin and organic status, and for supporting sustainability claims. Blockchain often enters this discussion. In a widely cited review, Kamilaris, Fonts and Prenafeta-Boldú concluded that blockchain is a promising technology for achieving a transparent food supply chain. Other authors summarising that work point to barriers such as the reluctance of small and medium enterprises to adopt new technologies, the absence of suitable policies, privacy concerns and the costs of implementation. I would add a conceptual caution that is widely recognised: a blockchain can make records difficult to alter after they are entered, but it cannot verify that the original entry was true. If a sensor is faulty or a person enters false data, the system will faithfully preserve the error. Trust therefore still depends on verification, audits and good governance. You may also have heard of digital product passports. They were introduced by the Ecodesign for Sustainable Products Regulation, Regulation (EU) 2024/1781, which has been in force since July 2024. The first product groups identified for ecodesign requirements and passports are textiles, mattresses, furniture, tyres, iron and steel, aluminium and energy-related products, so food is not in the first wave. I mention it because it shows the direction of EU policy toward product-level digital information, and food chains may be affected in the future. For farmers and cooperatives, traceability can be an opportunity or a burden. If it is designed so that producers share in the benefits, for example through premium prices for verified quality, it can strengthen their position. If it is imposed as an unfunded compliance cost, it can increase the pressure on small actors..

[Audio] The last stage of the chain is the interface with consumers, and here digital technologies are creating new routes to market and new information flows in both directions. First, the routes to market. Online marketplaces, direct-to-consumer shops, subscription boxes and community-supported agriculture platforms allow producers to reach customers with fewer intermediaries. For producers of premium, local or specialised products, this can mean higher margins and a direct relationship with customers. Cooperatives and producer groups can also run joint online shops, sharing the costs of logistics and marketing. But there are trade-offs: direct sales require skills in marketing, customer service and delivery, and they can be demanding for small farms without support. Second, information at the point of sale. A QR code on a package can link to the product's origin, production practices, certification and even the farmer's story. When this information is verifiable, it can differentiate products and justify a price premium. When it is not, it risks being seen as marketing, which can damage trust. Third, and often less visible, is the information flowing back from consumers to producers. Retailers and platforms collect large amounts of data on what people buy, when, and at what price. If this feedback reaches producers in a useful form, it can guide what they grow and how they plan. However, this data is usually held by the retailers and platforms themselves, which strengthens their position as orchestrators of the ecosystem. This links back to our concern from Module 1 about power asymmetries. There is also an interesting consumer dimension in the research. In their 2017 opinion paper in PNAS, Walter and colleagues argue that smart farming can reduce the ecological footprint of agriculture and make farming more profitable, and that it also has the potential to boost consumer acceptance. They stress, however, that this requires the avoidance of accompanying risks and proactive development of supportive policies and legal and market frameworks. In other words, consumer trust is not an automatic by-product of technology. It has to be earned through transparency that is credible. At this stage, I invite you to take your product sketch from Module 1 again. Where does information from the consumer currently enter the chain, if at all? And who would need to cooperate to make that feedback loop work?.

[Audio] Let us now step back and bring together what we have seen along the chain. At each stage we found potential benefits. In input supply, more precise use of inputs and new advisory services. On the farm, better monitoring, automation and decision support. After harvest, reduced losses and better matching of quality to markets. In logistics, visibility and condition control. In traceability, faster response to problems and verifiable claims. And at the consumer end, new channels and a closer relationship with customers. But the evidence also shows that the benefits do not flow automatically, and that adoption is shaped by barriers that are not only technical. Cost and the size of the farm matter. Skills matter, both for farmers and for advisors. Connectivity is uneven. Systems from different vendors do not always communicate. And there is often uncertainty about who will capture the benefits. This last point deserves emphasis. In a 2019 review in Sociologia Ruralis, Rotz and colleagues examined digital agricultural technologies through a political-economy lens. They observe that many farmers struggle to see the benefits of the digital data that their own farm technologies are producing. If the farmer generates the data but the value is captured elsewhere in the chain, the incentive to adopt and share is weak. That is a business-model and governance problem, not a technology problem. Another important perspective comes from Klerkx, Jakku and Labarthe, whose 2019 review of the social science literature on digital agriculture, smart farming and agriculture 4.0 shows that research has grown rapidly but remains scattered. They identify new research directions, including digital agriculture policy processes and digitally enabled transition pathways. The takeaway for us is that digital agriculture is a socio-technical system. Devices, institutions, skills, rules and business models evolve together, and success depends on getting the whole combination right. Let me summarise the module in four points. First, digital technologies create a data layer along the whole chain, from sensing to decision support. Second, each stage offers specific opportunities, but benefits depend on fit with the farm, the skills and the surrounding ecosystem. Third, traceability and transparency are powerful but depend on trust and verification, not only on technology. Fourth, the central unresolved question is how value and data are shared among actors. That fourth point leads directly to the next module, where we focus on data, platforms and interoperability: who owns the data, how can it be shared fairly, and what are data spaces? Before moving on, please take a moment for your reflection. Which stage of your chain has the largest gap between what technology could do and what is actually happening, and what is the main reason? If you are watching live, share your answer in the chat..

References. European Commission (2002). Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January 2002 laying down the general principles and requirements of food law, establishing the European Food Safety Authority and laying down procedures in matters of food safety. https://eur-lex.europa.eu/eli/reg/2002/178/oj European Commission (2011). Commission Implementing Regulation (EU) No 931/2011 of 19 September 2011 on the traceability requirements set by Regulation (EC) No 178/2002 for food of animal origin. https://eur-lex.europa.eu/eli/reg_impl/2011/931/oj/eng European Parliament and Council (2024). Regulation (EU) 2024/1781 establishing a framework for the setting of ecodesign requirements for sustainable products. http://data.europa.eu/eli/reg/2024/1781/oj ESA. Sentinel-2 User Handbook (ESA Standard Document). https://sentinels.copernicus.eu/documents/247904/685211/Sentinel-2_User_Handbook FAO (2019). The State of Food and Agriculture 2019: Moving forward on food loss and waste reduction. Rome: FAO. https://www.fao.org/newsroom/detail/A-major-step-forward-in-reducing-food-loss-and-waste-is-critical-to-achieve-the-SDGs/en Kamilaris, A., Fonts, A. & Prenafeta-Boldú, F.X. (2019). The rise of blockchain technology in agriculture and food supply chains. Trends in Food Science & Technology, 91, 640-652. https://doi.org/10.1016/j.tifs.2019.07.034 Klerkx, L., Jakku, E. & Labarthe, P. (2019). A review of social science on digital agriculture, smart farming and agriculture 4.0: New contributions and a future research agenda. NJAS: Wageningen Journal of Life Sciences, 90-91, 100315. https://doi.org/10.1016/j.njas.2019.100315 Liakos, K.G., Busato, P., Moshou, D., Pearson, S. & Bochtis, D. (2018). Machine learning in agriculture: a review. Sensors, 18(8), 2674. https://doi.org/10.3390/s18082674 Lowenberg-DeBoer, J. & Erickson, B. (2019). Setting the record straight on precision agriculture adoption. Agronomy Journal, 111(4), 1552-1569. https://doi.org/10.2134/agronj2018.12.0779 Rotz, S., Duncan, E., Small, M., Botschner, J., Dara, R., Mosby, I., Reed, M. & Fraser, E.D.G. (2019). The politics of digital agricultural technologies: a preliminary review. Sociologia Ruralis, 59(2), 203-229. https://doi.org/10.1111/soru.12233 TALLHEDA (2024). Brainstorming visit at the Agricultural University of Athens. https://www.tallheda.eu/post/tallheda-brainstorming-visit-at-the-agricultural-university-of-athens-a-resounding-success Verdouw, C.N., Wolfert, J., Beulens, A.J.M. & Rialland, A. (2016). Virtualization of food supply chains with the Internet of Things. Journal of Food Engineering, 176, 128-136. https://doi.org/10.1016/j.jfoodeng.2015.11.009 Verdouw, C., Tekinerdogan, B., Beulens, A. & Wolfert, S. (2021). Digital twins in smart farming. Agricultural Systems, 189, 103046. https://doi.org/10.1016/j.agsy.2020.103046 Walter, A., Finger, R., Huber, R. & Buchmann, N. (2017). Smart farming is key to developing sustainable agriculture. PNAS, 114(24), 6148-6150. https://doi.org/10.1073/pnas.1707462114 Wolfert, S., Ge, L., Verdouw, C. & Bogaardt, M.-J. (2017). Big Data in Smart Farming: a review. Agricultural Systems, 153, 69-80. https://doi.org/10.1016/j.agsy.2017.01.023.

[Audio] Thank you!. thank you!. TALLHEDA has received funding from the European Union's Horizon Europe research and innovation programme under Grant Agreement No. 101136578. Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them..