AgTech: data for agriculture

During an event at the Montréal campus, we brought together agtech experts to discuss the use of data in agriculture. Around the table were Colline Blanc, Data Scientist, Justin Hogue and John Lan, both R&D Scientists at ChrysaLabs. The discussion was hosted by Hugo Dupuis, Program Director at Zone Agtech.

The agriculture of tomorrow begins with a better understanding of soil. Thanks to AI, it is now possible to collect and interpret a mass of information that was once difficult to access.

“We wanted to use artificial intelligence to predict soil fertility from a spectral signature, without the need to analyze each sample in a lab,” explains Colline, who works on AI model development at ChrysaLabs.

To achieve this, the team used neural networks and machine learning algorithms to find links between soil spectral measurements and actual properties, such as nitrogen or potassium levels.

From soil science to AI: when rigor comes before innovation

At ChrysaLabs, artificial intelligence is not an end in itself. It is a tool serving a deep understanding of soils.

“The goal is not to train just any model with just any data. You first need to understand what you’re trying to predict,” emphasizes John, R&D Scientist.

Justin, also an R&D Scientist, adds:

“The model is just a tool. The core is soil science. You need a good dataset—representative and well-labeled. Only then can AI do its job.”

But modeling complex phenomena in agriculture is no easy task. Colline, Data Scientist, highlights the difficulty:

“Variability is huge. Soil is never identical from one field to another, sometimes even just a few meters apart.”

To meet these challenges, the team adopted a hybrid approach, blending scientific rigor with technological creativity.

“We explored many options, from simple linear models to deep neural networks. Sometimes, the simplest model is also the most effective,” shares Justin.

Data cleaning, cross-validation, ensemble models—each step is designed to guarantee robustness and precision.

Smart models, grounded in agricultural reality

The ChrysaLabs team works hand in hand with agronomists, farmers, and field partners. The goal: to create useful tools, grounded in real agricultural practice.

“We don’t want it to be a black box. An agronomist must be able to interpret the results and put them in the field context,” explains Colline.

This is also why collaboration with users is ongoing. Justin insists: 

“Our work needs to make sense for those on the ground. It keeps us connected to agricultural reality.”

With ChrysaLabs, AI becomes a lever for better understanding, anticipating, and supporting decision-making — always in dialogue with human experience.

A technological and human adventure

At ChrysaLabs, research and development are lived as an exploration. And sometimes, that means… making mistakes.

“We went through a lot of trial and error. Sometimes we realized certain approaches were not suited at all to the agricultural context,” admits John.

But every detour is also a learning opportunity.

“It’s a very iterative process. We test, we adjust, we start over. And that’s what makes the work so exciting,” adds Justin.

Agility is at the heart of their method, with a strong ability to adapt to the complexity of the field, the seasons, and crop types. 

“You have to accept uncertainty and work with it,” summarizes Colline.

Close collaboration between scientific profiles

Another key strength of the project lies in interdisciplinary collaboration. Soil science, data science, and engineering profiles work hand in hand.

“We each had our own perspective. As a data scientist, I needed to understand the sensors, the data types, the physics behind it. The agronomy researchers helped me connect those dots,” explains Colline.

And this interdisciplinary approach is reflected in the results: a faster, more precise soil diagnostic tool, enabling farmers to adjust their practices in real time.

🎥 In this video (in French), our guests explain what drives their passion for working in agriculture

What’s next?

“AI is not a magic solution, but it’s a powerful lever to transform agriculture,” concludes John.

The ambition of ChrysaLabs and Zone Agtech is clear: to make agriculture more efficient, more sustainable, and more connected to field realities.

And above all, to ensure that technology serves those who feed the world.

🎧 Want to go further? 👉 Watch the replay of the discussion here (in French)

Interested in learning how to work with data and make it your career? Explore our Data Analytics and Data Science & AI bootcamps.