How Earth Observation, Spectroscopy, and AI are changing soil use forever and how can we turn soil health research into thriving businesses? Key takeaways from the Soil Health Now! conference 2025
Prepared by: Tom Hengl (OpenGeoHub), Koen van Seijen (RegenEarth), Naeem Lakhani (The First Thirty), and Ichsani Wheeler (OpenGeoHub)
Soil Health Now Conference (science, technology and business intelligence for measuring monitoring and increasing soil health) was held at Wageningen International Conference Center, Wageningen from 8–10 April 2025. Video-recordings of the majority of talks are available for viewing under the CC-BY license; links to various development projects on Github and similar are also provided. This article summarizes the key takeaways and provides links to those interested in further reading. Selection of the key takeaways, however, is purely subjective and unsorted; feel free to watch videos, comment on, rate and reflect — what are the most important takeaways for you? What have you learned and what has potentially been an eyeopener for you?
Soil Health Now! in a nutshell
Soil Health Now! was held at the Wageningen International Conference Center, Wageningen from 8–10 April 2025. It was a 3-day international conference with keynotes, discussion forums, oral talks, workshops and demonstrations + fieldwork, organized under the auspices of the AI4SoilHealth.eu project and aiming at bringing the European family of Horizon Europe projects focused on soil health together. Compared to the European Mission Soil Weeks events organized by the European Commission, the main goal of the Soil Health Now! was to bring together researchers and entrepreneurs under the same roof, and let them generate discussions in a more bottom-up approach; i.e. let the audience select topics and critically and honestly address their doubts, fears and reservations (because let’s face it: AI or similar tech will not automatically solve all our problems or provide salary for everyone). Soil Health Now! should be hence considered a primarily bottom-up event aiming to provoke new discussion and break existing dogmas.
We held three expert discussion panels focusing on key challenges and opportunities in the field of soil health:
- “The Soil Data Revolution: How Earth Observation, Spectroscopy, and AI are Changing Soil Use Forever”
- “Europe’s Innovation Paradox: How Can We Turn Soil Health Research into Thriving Businesses?”
- “The Next Generation of Soil Tech: Affordable, Accessible, and Farmer-Focused”
To stay as much as possible objective & not influence participants, we have published (unedited) results of polls (Slido) as PDFs via our Telegram channel (https://t.me/opengeohub) as raw discussion notes, immediately after the discussion panels were completed. Here you can read a short summary of what has been discussed and concluded. We do this just to make it easier for you to find information — the items are unsorted and of course the list could be further reduced or extended. The video-recordings of the sessions (almost 12 hrs of recordings) are available here. The most important takeaways from the conference (in our subjective opinion) are listed below (unsorted).
AI technology (automated chatbots replacing agricultural extension, soil experts) might reach all 9M farmers in EU even before 2030
AI technology, especially automated content generation AI tools, AI chatbots and automated navigation systems (self-driving tractors and similar) has quickly flooded all areas of life and work. The majority of the participants of the conference agree that the exponential growth of AI technology solutions will continue and at this pace it is very well likely that it will reach 99% of farmers in Europe (about 9M of them), even before the year 2030.
It is interesting to note that there seems to be, however, two clusters of opinions based on our results: (1) the “optimist cluster” that probably also believes that AI technology will be quick, successful and go beyond we can imagine, and (2) the “cautious cluster” that thinks some higher levels of expertise (wisdom) will not be replaced soon enough maybe not even in few decades (if ever). Also consider that data privacy and trust are top of mind when it comes to AI technology: audience voting revealed that data privacy and trust in AI outputs are the most significant concerns. Panelists emphasized the importance of transparency, control, and “humans in the loop” to ensure ethical and safe AI use.
AI technology does not necessarily make our work easier — you still have to study and practice stats, soil science, agronomy and similar
We are in 2025 and it appears that AI technology is expanding rapidly and can replace many of the existing jobs, eventually the majority of current jobs & tasks. Does this mean that soon we will not have to study and practice statistics (sampling, modeling, statistical testing etc) at all? Is life for the pedometricians and soil scientists becoming easier so we can all work less and retire sooner? It might seem counterintuitive, but even though AI technology can speed up the work, it will most likely require a higher knowledge of technology including computer science, on top of all basic foundations we get in statistics, soil science and similar. One of the most popular sayings in 2025 is: “AI technology will not take over your job, it is the people that are good in AI tech that will take over your job”, so definitively a message for everyone to consider that studying, practicing and mastering complex scientific fields (including AI programming) is here to stay.
Madlene Nussbaum: “the application of all these tools has become much easier, so we have to spend much less time in actually implementing machine learning and statistics, but at the same time we still need to know what we do and have a clear picture of what the to put the outcome into perspective and to actually be able to have a proper interpretation of the outcome and the results, and without a sufficient background knowledge we will not be able to to put it into perspective”.
Value of the data is in its use — let’s optimize data accuracy, production costs and usability to match what works for decision makers
One of the interesting questions raised from the audience during the discussion panel was “what is the (healthy) balance between scientific accuracy in data and its usage or usability?” Higher accuracy increases production costs, but many life-saving decisions can be made with much less accurate / approximate data. There seems to be a general consensus that the value of the data is in its use. As general conclusion of this discussion is that we should really try our best to optimize data accuracy, production costs and usability to match what works for decision makers.
Madlene Nussbaum: “We are driven by achieving the best possible product if we do research for these types of methods of course but then and at the same time we try to do this in a very general purpose manner so that we can accommodate as many users as possible. But if we look at the user sites that do for example policy implementation at a local government they sometimes just have a binary decision whether it is inside the class or outside. Therefore maybe we would need to adapt our data products more targeting their exact decision and also then optimizing the accuracy for their exact decision and achieving the most accurate accuracy level so they do not do the risk of taking a wrong decision as small as possible. So it really depends on the resolution of the final decision. And of course if we go for the three digits after the comma for soil texture it’s useless.”
Integration of soil science, hydrological and ecological knowledge into ML might be our best bet to help improve accuracy of soil data predictions
In her talk “Machine learning for soil health — Is the horizon the limit? Flaws, potentials and future challenges”, Madlene Nussbaum suggested that from the four development directions in soil mapping, it appears that investing in the knowledge-integrated ML should be our shorter term priority as it does not require as much funding and many tools are already available (just need to be further adjusted and optimized). A lot has been published on this topic recently, especially the Minasny et al. (2024) and Willard et al. (2022), are highly recommended reads.
Not every project that calls itself “Regenerative agriculture” is, unfortunately, genuine; not every regenerative action actually helps improve soil health: evaluate them case by case
Prof. Ken Giller in his recent podcast has warned of dangers of populist narratives that oversimplify agricultural challenges. How many projects today that call themselves “regenerative”, “organic” are genuine? Organic farming and organic agriculture producers in the EU are now quite strictly regulated, however, we should be much more careful about any producer self-declaring being “regenerative”, “ecological” or similar. On one hand business does not really care whether a project has some certificate as long as there is a positive balance in the books; on the other hand, some businesses seem to sell “regenerative” as a brand, i.e. they try to profit from the hype or social acceptance. The audience at the SHN also agrees that most of the projects that call themselves “regenerative” need to be evaluated case by case and possibly they are regenerative and we should champion them, but let the data (objectively collected and validated) show that.
Soil spectroscopy could be the key technology for rapid measurement of soil health, however it is still heavily underused in practice
While detailed lab-based soil tests are accurate, their time lag and lack of real-time decision support limit usefulness. Farmers often need immediate insights, especially under weather or crop stress conditions. Soil spectroscopy, especially MIR and VisNIR spectral scans, today even possible to be produced in-situ, have been in the focus of pedometricians since the early 2000’s. Some recent publications indicate that soil spectroscopy (especially MIR-based) could potentially be used to directly estimate soil health score. But in reality, the number of farmers in Europe actually using soil spectroscopy is probably in permilles. Our poll results indicate that the key limitations of the soil spectroscopy systems are that they are: still too complex to use, not plug-n-play systems and often come with too high costs. Some commercial suppliers of soil spectroscopy technology have made more robust systems with impressive accuracy. However, their business model remains focused on only serving predictions rather than producing / providing instruments for quick soil spectroscopy. How exactly predictions are derived is proprietary, i.e. a black-box system, which for many customers is not an issue, but troubles researchers. Are black-box soil spectroscopy systems the only business model that we need?
Only 5 CEO’s in the Netherlands basically determine what you will eat
Although producing food is highly sophisticated and many farmers work hard to produce quality food. Consumers are increasingly detached from the process. As successful food corporations become larger and larger, we finish depending on few people determining what we eat. Geert van der Veer: “The Netherlands is about 18 million people. In the middle there are five big offices who are buying and reselling the food farmers produce and what comes and what we are eating. In the Netherlands 6,400 supermarkets sell food. But all the food farmers produce will reach the people but only five (5) CEOs can make a difference. This is what the consumer finds in the supermarket. When you walk in a supermarket there are so many products that people cannot see anymore that there is a farmer involved with that product. It’s marketing. It’s big. I always say farmers produce food and we eat products.” How to change this concentration of power and a total ignorance of consumers? Few strategies come to mind. We need to involve and educate consumers to appreciate food, where it comes from, how it was made, what effect its production had on the environment and similar. We also need to teach consumers to connect and trade with small-scale farmers directly, without some corporation taking the majority of profit for themselves.
Europe is lagging behind in the entrepreneur culture in comparison to the USA, Canada & China, but only in numbers, not in spirit!
Europe is unfortunately known for its lagging behind USA, Canada and China (just to mention few economies of comparable size), especially when it comes to investments and entrepreneur culture. We have asked participants why they think this is and got a wide diversity of answers: too much bureaucracy, too many languages and cultural barriers seem to come somewhat higher on the list. Most young entrepreneurs probably do not even know when and how to start. To reduce some of these problems, the European Commission has launched the European Innovation Council Accelerator with many interesting funding opportunities, but we probably need much more.
More than 50% of the participants are planning to start a company in the next 2 years!
This is surprisingly high considering the current number of start ups coming out of the soil health science world. Based on our poll, more than 50% of the participants are planning to start a company in the next 2 years.
Many state that the key barrier to building a company is: the missing market. Panelists shared their journey in starting different companies and the huge risks they had to take. Starting a company isn’t for the faint of heart.
Investor’s interest in investing in soil and agriculture is growing rapidly, but investors need help in defining and tracking KPI’s
The large language/cultural and general understanding of each other’s fields of experience remain barriers to having significant money flow to restore soil at scale. Large Corporations are investing heavily in regenerative supply chains. But they need help in defining regenerative agriculture and how to work with hundreds and in many cases thousands of farmers on their regenerative journey.
‘Boring’ accounting but very impactful: moving from a cost to an investment
Dr Tobias Bandel showed how they are working with the large accountancy firms to allow companies to change the way they account for investments in regenerative agriculture supply chains. Currently this is considered a cost not an investment, but moving it to the ‘investment side’ of the balance sheet will unlock billions into regeneration. Accurate cheap soil health measuring technology will be key to enable this.
Is soil the most undervalued natural asset on the planet?
Soil health is one of the planet’s most undervalued assets, largely because its profound influence on human health and economic productivity remains underappreciated.
While soil health’s relationship to planetary health is clear, its value is neglected even though 95% of global food production depends on soil and >50% of global GDP is reliant on natural capital. Worse still, the relationship to human health is obfuscated by a lack of data and understanding between what directly links soil and human health. Meanwhile chronic diseases, driven by metabolic dysfunction, now claim 41 million lives each year and costs of diet related diseases are projected to reach $47 trillion by 2030.
Connecting the dots means investing in soil health is a direct investment in climate resilience, food systems, and public health representing a massive opportunity. Further, reframing soil as a cornerstone of human vitality and societal wealth is critical to building systems that prioritize long-term planetary and human health.
Data accessibility is a bottleneck for many farmers
Despite extensive data from long-term experiments and public sources, much remains inaccessible. Panelists emphasized the urgency of democratizing soil data while safeguarding privacy and ownership. Organizations such as SoilHive and Zenodo.org could play an important role in making soil data more accessible and better structured. Beyond that, it appears that only a smaller portion of participants use commercial satellite products. It appears that most of the bigger EO data & service providers such as Umbra.space, Planet, Pixxel.space, Satellogic, seem to be focused on bigger industry, government agencies and similar. How to reach all farmers with effective and affordable products?
Next Gen Soil Tech proximal & drone-based soil sensing offer new potential
Panelists highlighted the importance of proximal soil sensors, especially for mapping peat soils and managing CO2 emissions. Drone-borne sensors were described as promising for precision agriculture. Some panelists argued that soil tech is still too expensive for most farmers due to financial constraints and debt. Others noted that large farms with capital will invest if tools clearly show return of investment. Innovations discussed included real-time sensors, plant communication, and combining AI with traditional knowledge. These hybrid approaches offer promise for tailored and ethical solutions.
Policy, Scale, and Simplicity Must Align: simplicity is desired, but not at the cost of local context. Tools must account for farm size, climate, and cultural practices. Regional implementation of broad EU policies must reflect local realities.
Farmers as Co-Creators, Not Data Laborers
Tools need to be developed with, not just for, farmers. Trust, ownership, and integration into daily routines are key for adoption. Farmers were described as the best sensors on the field. Farmers should be treated as co-creators, not only as data laborers. As Fabio Volkmann (Co-Lead Climate Farmers) puts it “It is not about scaling up solutions to farmers, it is about scaling up farmers’ solutions”.
The Role of Open Source and Ecosystems is Expanding
Panelists highlighted the importance of an open ecosystem — open-source models, API-first infrastructure, and a plug-and-play architecture — for flexibility, speed, and scalability in AI adoption. One example of an open data open source community is the Open Soil Spectral Library (OSSL). It took almost three years to establish, but now comes with an active community and many potential applications. Other similar examples of open source solutions for soil data are the Open Soil index and the Soil Health Data cube for Europe.
Funding Must Go Beyond Governments
Panelists stressed that stakeholders beyond government (e.g., health insurers, water utilities, food companies) should invest in soil health tech, given their shared interest in resilient food and health systems. Governments and the European Commission have recognized the importance of soil health monitoring, but corporations need to do their part too.
Professor Lubos Pastor and two Wharton researchers recently estimated that the social cost of the carbon emissions of US companies will amount to $87 trillion through 2050. Callahan and Mankin (2025) calculated in detail how trillions in economic losses are attributable to the extreme heat caused by emissions from individual companies. This could be a birth of the new (realistic) economy where ecosystem services and costs of ecosystem degradation are included in the price of products, or taken from massive profits some companies make. The bottom line is: we (people, entrepreneurs, policy makers, share-holders) are all in this together, and how economic analysis is run needs to change to be more realistic, in needs to match total costs of running any business on the planet.
We know so much about land use in Europe, but so little about soils
Below is an example of three maps describing the state of the pan-EU environment. Note that when it comes to monitoring soils as an asset we actually know much less than we know about location of businesses, traffic infrastructures, price of real estate or similar. We argue here that everyone in Europe has a right to know what is happening with European landscapes, soils and biodiversity, without delays and at the finest possible spatial resolution. The data on the state of each of the three should be of the highest possible quality, open and analysis-ready. The European Soil Data Center (ESDAC) and EU Soil Observatory here play a crucial role, but other organizations and countries also need to contribute.
Letters to a young farmer we should all read
Scientists and entrepreneurs tend to obsess with their work so that we often forget why we do our work and where we come from in the first place. Anouk Schoors from the Nest Family office reminded us to read beyond research publications, from people on the ground, trying to make a change day by day. One of such activists is Karen Washington, who has participated in many events and books on farming and conservative agriculture. One of these books is the “Letters to a young farmer”, which was also cited by Anouk at the end of her talk. If you feel that you do not see forest from the trees anymore, this could be a good read for you.
Summary points
Soil Health Now! has proven to be a much needed event and we very much appreciate your positive feedback we received via Linkedin, Mastodon, Bluesky and similar from all participants. There were almost 200 participants registered, but most importantly we (organizers) managed to get entrepreneurs, researchers (including leading Horizon EU Soil mission projects) and soil data collectors together inside one ceiling. Communication between fundamentally different groups of people is not always trivial. Finding the balance between accuracy, usability, costs of data and technology will remain at the forefront of the field. European Commission is investing almost 1B EUR into soil health research, so expect a lot of new breakthrough data sets, technology solutions and innovation that should eventually be able to reach all farmers in the EU.
We believe that there is a general consensus among the participants about the following 8 pillars of reaching healthier soils / restoring lands:
- Soil is one of the fundamental assets of human civilization. It is an intergenerational asset that we can’t afford to lose, even if we stop producing food from soil, it is still the Earth’s biggest pool of biodiversity, also the key ecosystem responsible for clean water, stable climate and healthy biomes, so soils need to be protected.
- Soils need to be monitored and people need to be warned about any degradation processes on time. People have the right to know what is the soil health index of their land, how much have the previous owners degraded soils and which were drivers of degradation? The data about the status of soils should be open, transparent and easy to access and use.
- Soil health is still largely a concept; we now need robust indices and cost-effective in-situ tools for measuring soil health. We need to engineer soil health assessment (especially to engineer derivation of robust soil health indices) so businesses can further develop useful products.
- Open Earth Observation is the key foundation for soil monitoring. European programmes such as Copernicus with Sentinel satellites and data are a basis of monitoring the state of soils, and so is LUCAS soil and the EO Soil observatory. European programmes generating open data do not only generate data, but they also help connect researchers and farmers across borders.
- AI technology, drones, soil spectroscopy and similar can play an important role in making soil health monitoring more robust, especially if they help decrease costs of monitoring. AI technology might soon reach all 9 million farmers in EU, so we need to be ready for exponential growth of applications and users.
- AI technology should be used with the highest care. AI technology brings solutions, but also challenges. We generally agree that AI technology should not be used (in the next 10+ years) as a replacement for our expertise, but primarily as a tool for reducing costs, speeding up restoration and automating repetitive tasks.
- There is a need for a new, more complete economy where ecosystem services of soil and costs of ecosystem degradation are included and accounted for by governments / tax offices. Businesses should prioritize high tech solutions that both bring high revenues and ensure strict sustainability of natural resources.
- Farmers should be invited as Co-Creators into the soil health innovation and land restoration, not as just users. “It is not about scaling up solutions to farmers, it is about scaling up farmers’ solutions”.
The AI4SoilHealth project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement №101086179.
If interested in Soil Health Now! events, please sign-up to our mailing list and start preparing for other upcoming events in 2025/2026 e.g. the European Mission Soil Week 2025, 5–6 November 2025 in Aarhus.
Cited references:
- Callahan, C. W., & Mankin, J. S. (2025). Carbon majors and the scientific case for climate liability. Nature, 640(8060), 893–901. https://doi.org/10.1038/s41586-025-08751-3
- Leendertse, J., Schrijvers, M., & Stam, E. (2022). Measure twice, cut once: Entrepreneurial ecosystem metrics. Research Policy, 51(9), 104336. https://doi.org/10.1016/j.respol.2021.104336
- Miglio, E., Albanito, F., Sala, S., & Smith, P. (2024, April). Bridging the Soil Health Data Gap: informing prioritization of soil data collection efforts. In EGU General Assembly Conference Abstracts (p. 13121). https://dx.doi.org/10.5194/egusphere-egu24-13121
- Minasny, B., Bandai, T., Ghezzehei, T. A., Huang, Y. C., Ma, Y., McBratney, A. B., … & Widyastuti, M. (2024). Soil Science-Informed Machine Learning. Geoderma, 452, 117094. https://doi.org/10.1016/j.geoderma.2024.117094
- Panagos, P., Borrelli, P., Jones, A., & Robinson, D. A. (2024). A 1 billion euro mission: A Soil Deal for Europe. European Journal of Soil Science, 75(1), e13466. https://doi.org/10.1111/ejss.13466
- Ros, G. H., Verweij, S. E., Janssen, S. J., De Haan, J., & Fujita, Y. (2022). An open soil health assessment framework facilitating sustainable soil management. Environmental Science & Technology, 56(23), 17375–17384. https://doi.org/10.1021/acs.est.2c04516
- Rutgers, M., van Leeuwen, J. P., Vrebos, D., van Wijnen, H. J., Schouten, T., & de Goede, R. G. (2019). Mapping soil biodiversity in Europe and the Netherlands. Soil Systems, 3(2), 39. https://doi.org/10.3390/soilsystems3020039
- Safanelli, J. L., Hengl, T., Parente, L. L., Minarik, R., Bloom, D. E., Todd-Brown, K., … & Sanderman, J. (2025). Open Soil Spectral Library (OSSL): Building reproducible soil calibration models through open development and community engagement. PLoS One, 20(1), e0296545. https://doi.org/10.1371/journal.pone.0296545
- Sanderman, J., Savage, K., & Dangal, S. R. (2020). Mid‐infrared spectroscopy for prediction of soil health indicators in the United States. Soil Science Society of America Journal, 84(1), 251–261. https://doi.org/10.1002/saj2.20009
- Veum, K. S., Sudduth, K. A., Kremer, R. J., & Kitchen, N. R. (2017). Sensor data fusion for soil health assessment. Geoderma, 305, 53–61. https://doi.org/10.1016/j.geoderma.2017.05.031
- Willard, J., Jia, X., Xu, S., Steinbach, M., & Kumar, V. (2022). Integrating scientific knowledge with machine learning for engineering and environmental systems. ACM Computing Surveys, 55(4), 1–37. https://doi.org/10.1145/3514228
