It can ruin a breakfast, a bad avocado. The bagel and eggs may be all ready to go but cut open that shiny green fruit and find brown mush inside, and it’ll put a downer on anyone’s morning. “Couldn’t they have known it was a bad one?” you cry.
Well, yes. Maybe, they could. That is at least what Orbem is promising, a German start-up claiming to revolutionise the use of MRI scanning in food production.
By using AI to speed up analysis of scans, Orbem claims it can industrialise the process of scanning the inside of food and massively reduce food waste across the industry.
In January, it closed a €55.5 million Series B financing round to help it expand into the US and move from poultry into fruit and vegetables.
Orbem claims to be the first company to fully automate and industrialise MRI for commercial applications. Its CEO, Pedro Gómez, explained to AgNavigator how it plans to revolutionise the industry.
From hospital scanner to food production line
AgNavigator (AN): What is Orbem?
Pedro Gómez: We are a deep-tech company founded in Munich, specialising in AI and imaging technology. A typical scan takes between 20 and 25 minutes. We have reduced that to less than one second and integrate automation equipment so it can be used industrially to look inside and understand biology in real time.
AN: And that’s mainly food?
Gómez: Mostly food: eggs, watermelons, mangoes, avocados and other fresh produce. For every type of food we work with, we can look inside and, in real time, determine whether it is good or bad. Is the quality good? Is it ripe enough? Is it mature? Is it broken? Does it have a hollow heart, for instance, as can happen with watermelons?
We launched in poultry in January 2023 and since then we’ve scanned more than 200 million eggs. That’s more MRI images than the global healthcare industry has produced. That data is really valuable because it enables us to continue refining our AI technology. If you want to build AI, you need data. The big language companies train on the internet, but there is no internet of MRI images. We have the largest dataset of MRI images in the world, which enables us to deliver this intelligence.
Why start with avocados rather than patients?
AN: Why have you started with food? If you’ve got the ability to combine AI and MRI, why not apply it to human healthcare?
Gómez: A couple of reasons. One, you can have a ton of impact in food. The amount of food waste we have in today’s world and what that means for sustainability and greenhouse gas emissions is mind-boggling.
The second is technological maturity. When we started, we came from the clinical world. I was working with humans, neuroscience, brain tumours, multiple sclerosis – anything around the brain – and this technology was evolving. People see MRI and think humans. But the technological maturity [for what Orbem is doing] was not quite there.
In food, the requirements are exciting because they allow us to validate the technology and show that you can run MRI at really high throughputs. It enables us to collect this massive amount of data that we wouldn’t be able to collect in a constrained clinical environment.
Reducing waste before it reaches the supermarket
AN: How does this reduce food waste? Once a fruit is bad, it’s bad. So you may not have the transportation impact, but isn’t it just wasted at an earlier stage?
Gómez: It depends on what you do with the fruit. If you identify a bad fruit at the retail level, then it goes in the trash.
But at the production level, you can still do interesting things with different grades of fruit. It’s maybe not great for consumption, but it’s still suitable to go into juice or another by-product.
But also, perfectly good fruit gets sent to processing simply because it looks wrong on the outside. If you can show what is inside, you can market that fruit for what it really is. I think there is real potential there to help producers recover margin they had written off.
Building a biological intelligence platform
AN: You’ve previously spoken about “understanding biology in a way that was previously impossible”. Is that just identifying quality or does it go beyond that?
Gómez: It goes beyond that. We can detect the sex of an egg, for instance, and that’s just understanding biology. It’s identifying a biological class.
Now, when we think of fresh produce, biology informs quality. So, the biochemical composition of an avocado manifests itself in how you see the flesh.
We’re understanding the biology, and if you connect the biology with other kinds of data – maybe field data, maybe production data – then the intelligence you create is much more exciting.
AN: So, you can start to make wider judgments on why certain avocados are low quality?
Gómez: Exactly. I’d divide it into the intelligence of the individual sample, which is about quality, versus the intelligence of the population, which relates much more to the biology of the wider population. Why is this happening? Why does my entire patch look like this? Why does this other batch have a different characteristic? What explains the differences? But it all starts with being able to see and understand what’s happening on the inside.
AN: What can you do with that knowledge and information?
Gómez: Make better decisions. It’s all about enabling our customers to make better decisions. We deliver the equipment – MRI doesn’t exist outside the hospital environment – so we need to bring it to places where it was not previously possible, like a hatchery or a production facility.
We deliver the equipment, and then our commercial agreements tend to be volume-based arrangements.
And through those commercial agreements, we provide the intelligence our customers need, which they can use to inform their production processes.
Scaling up MRI for agriculture
AN: What scale are you at now?
Gómez: We have 200 people working full-time in the organisation. We have customers in eight different countries. Our revenue growth has been over 100% for each of the past couple of years since we launched in 2023.
This year we also launched with JimboFresh in Spain. They’re one of the largest watermelon producers in Europe.
AN: Do you have competitors?
Gómez: We’re, of course, competing with all the other sensors and measurement devices out there. You have hyperspectral sensors, which are really well established in the industry. You have a lot of optical systems.
The biggest difference is that we can see every single detail inside. But this is also not novel. MRI has been around since the 1980s and scanning biology with MRI is not a new idea.
What is new is that we can do it at a cost, speed and scale that makes it industrially feasible. That’s only become possible with the advent of AI technology in combination with the automation technology we bring.
The long-term vision: food first, healthcare next
AN: What’s your ambition for this?
Gómez: We see a world where our technology becomes the standard for food production and we’re scanning hundreds of millions of kilograms of food per year.
As we do that, we continue building the largest and most diverse imaging dataset in the world. In the eyes of an MRI machine, damage in a watermelon or avocado manifests itself in exactly the same way as damage to human tissue. So, as we do this, we continue building foundational intelligence that allows us to bring MRI to human health in a way that was previously impossible.
That’s what success looks like for us in the long term: becoming the intelligence company for food and health that brings AI-powered MRI to these two markets.
