This article was originally published by ChangeInc on June 17, 2026, by Maaike Kooijman
Decisions about nature conservation are often based on maps that are years out of date. Dominique Cirkel believed there had to be a better way.
“Look,” says Dominique Cirkel as we study a digital map of a nature reserve in Drenthe. She draws a rectangle in the top-right corner of the screen with her mouse. “I happen to know that soft rush, a grass-like plant, grows here. And here…” She marks another area. “…it doesn’t. Now we can see whether the model can predict where else the plant is growing.”
Cirkel is demonstrating the Spheer App, which she developed with co-founders Jakko de Jong and Mark Boer. At its core is an AI model trained on years of satellite imagery, which customers can fine-tune using their own data.
The app allows users to monitor changes in biodiversity within a specific area. They can, for example, map particular plant species, assess the condition of heathland and create vegetation structure maps. This information can then be used to support important decisions about nature management.
Your app can reveal everything from changes in biodiversity to water quality and soil degradation. How does it work?
“Our AI model has been trained on years of satellite imagery. It is a foundation model, similar to the large language models behind systems such as ChatGPT. But while these kinds of models have been used for years to process text and images, the same was not yet true for geospatial data. So we decided to develop our own model.
Sentinel-2, the satellite that provides the data, passes over an area roughly once every three to five days. That makes it possible to track changes over time. Sentinel-2 is a multispectral satellite, which means it records different wavelengths of light and radiation. This can be used to measure things such as soil moisture and to distinguish between different tree species. In the process, our model has also learned to recognise patterns such as vegetation growth and seasonal variation.
We train the main model ourselves, while customers only need to train the final layer using their own data. This might be done by ecologists who know an area well and know which species grow where. Using data from just a small part of an area, the model can then make predictions about the rest of it. What makes Spheer distinctive is that anyone can use it and gain access to a powerful AI model like this, even without a technical background.”
What do customers use that data for?
“At the moment, most of our customers are government bodies and consultancy firms, and they use the data in different ways. One example is checking whether subsidies for the conservation of nature reserves are still appropriate. Another is locating rare plant species, such as the fen orchid.
The satellite images themselves are too coarse to identify an individual plant visually, but the model can predict where it is likely to grow based on the surrounding habitat. And when people went to those locations, the plant was indeed found there.
Government bodies also use Spheer to assess whether policies are having the intended effect. The Province of North Brabant, for example, uses our data to monitor the implementation of its nature management plan. By comparing the plan with what our data shows on the ground, the province can identify places where forest is supposed to be present but currently is not. That gives them a starting point for further investigation and, if necessary, adjustments to the plan.”
You are an AI specialist. Where does your interest in nature and sustainability come from?
“As an AI expert, you could go into marketing and make more money for wealthy people, but that wouldn’t make me happy. My co-founders and I like working on something that has a positive social impact. At the same time, we also more or less found our way into this field by chance.
I used to work with Jakko [de Jong] and Mark [Boer] at a staffing and secondment agency, where we worked on a lot of projects involving satellite imagery and AI. We noticed that provinces and municipalities often make decisions using maps that are six or even twelve years old. We saw an opportunity to use AI to make that information both more up to date and more consistent.
These days, I actually spend less time building the product itself and more time building the company. I’m involved in product decisions, project management and team development, for example. I’m also often the person who meets with customers to understand what they need.”
How sustainable is your AI model?
“We pay close attention to its energy use. That is partly because sustainability is central to what we do, but also because computing power costs money.
It helps that at least 99 percent of the model is the same from one project to the next. We also only retrain the foundation model once every few months. We test new ideas on a small scale first and only roll them out more widely once we know they work.”
What opportunities do you see for AI in sustainability?
“To manage nature effectively, you first need a good understanding of what is actually there. But there simply aren’t enough people to map everything manually. And that makes sense: you can’t carry out field surveys everywhere, all the time.
AI is very useful for filling gaps in the data, both over time and across large areas. That can provide a much stronger basis for important decisions.”
What opportunities do you see for Spheer itself?
“We already have quite a few customers within the Dutch public sector, but that is not a major growth market. In the coming period, we want to work with more international and commercial organisations.
I think our biggest added value lies in biodiversity analysis. There are already many companies using satellite data to monitor forests and deforestation, so we are certainly not the only ones doing that. But when it comes to measuring and predicting biodiversity, there is much less competition.
I also expect the market for biodiversity credits, which is similar in concept to the carbon credit market, to grow rapidly over the next few years. The United Kingdom, for example, recently introduced BNG, or Biodiversity Net Gain. New development projects are now required to demonstrate a measurable positive impact on biodiversity.”
Who would you still like to work with?
“I’d love to work on a major international project one day. There are, of course, parts of the Netherlands that we still know relatively little about, but in other parts of the world those knowledge gaps are much larger. And the larger the area, the more gaps there are that Spheer can help fill.”
