Dr. Maryna Kuzmenko

Regenerative agriculture & agritech at the ‪@Innovateuk‬ & ‪@IUK_Connect‬ Annual Showcase.

Field notes.





🌱





For quite a long time, regenerative agriculture was often presented almost like a philosophy, and agritech was often presented almost like a product catalogue. But here, the two were meeting in a much more grounded place: practical adoption.

ADOPT grants are exactly about this.





🌱





Soil and growing media were a strong part of the conversation. Biochar now is not a cool word to know, but something linked to soil improvement, carbon, compliance, and practical farm use.



Peat-free growing media also stood out as a very real topic, especially because re-united with Alicja SilviBio :)





🌱





Crops also told an interesting story.

Nut farming appeared as a regenerative and long-term resilience strategy, especially through the example of UK-grown hazelnuts and walnuts, agroforestry, and the creation of wildlife habitats alongside food production.



I discussed buckwheat with the team of Waitrose farm (one of their own-baked breads includes seeds of this useful alternative crop + we have some nice buckwheat news at ‪@petiolepro‬ Petiole Pro 🤩).





There are also plenty of interest to neglected and underutilised crops that bring not just novelty, but resilience for farms.





🌱





Biodiversity.

The need has appeared in recent years, and now we can speak about a clear niche of AI-powered biodiversity monitoring, using cameras and sensors to track presence, sounds, and other signs of life generated by biodiversity populations. Plus, this direction is very active on the international scene as well!





🌱





Water was another major theme. Filters for water streams, water quality, water availability, and, critically importantly, improvements in nutrient and nitrogen cycling all came up - because eutrophication is the elephant in the room, making everyone uncomfortable.





🌱





Carbon credits are not money from air, but a real revenue from soil, plants, data, and verified practice change.





🌱





Sometimes agritech is hungry and goes to food :).

Apart from eating the walnuts and oats and drinking the beetroot juice, I was impressed by industry-level work by Scottish researchers on substituting palm oil for bakery applications. Equally interesting was cocoa substitution based on fermentation of cocoa shells, which looked and tasted like a serious answer to headaches in sustainability and supply chain pressure.





🌱





Music as a part of regenerative agriculture?

Yes, absolutely. Speak to amazing family-owned Glastonbury Nut Farm - they can tell you real stories about agritech, regenag plus the well known annual festival, which runs next to their field





🌱





Being too optimistic about life is not always a good thing.

Because reality can hit hard.

But after this event, I felt definitely better than before it :).





There is the promise in the agrifood industry, based on the long-term decisions already made by the UK farmers.



Thank you for a dose of inspiration.



#innovateuk #agrifood #drmarynakuzmenko

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

6 days ago | [YT] | 10

Dr. Maryna Kuzmenko

What happened yesterday at the ‪@Innovateuk‬ & ‪@IUK_Connect‬ Agrifood Annual Showcase (RegenAg + Agritech) I will write about on Monday.



For today — just a few moments so you can feel the vibes of our nice conversations.

Indeed, it was not about pitching and presenting, but about checking how the adoption of tech and regenag is actually going on real farms.

I'm so glad that ‪@petiolepro‬ was an active participant!



I was lucky to ask my three questions during the panels, and I also got my one million other questions answered in the demonstration area and networking sessions 🤩. Some of the hottest discussions happened in the queue for lunch, at the phone charging area, and with some ladies in our traditional queue to the well-known facility :).

The range of questions included, but was not limited to: biochar, biodiversity net gain, regenerative agrivoltaics, quality control of nuts-in-shell using hyperspectral imagery, stomata count (unexpectedly!), soil monitoring, water reservoirs, carbon capture... to name just a few.



My half of a day was spent with SilviBio because, as a lettuce-addicted person, I can't survive without real plants - check the video, their leafy greens + peat-free growing media were there :).



Anyway, I'm really glad that:
1. regenerative agriculture and agtech adoption are speeding up in the UK on a practical level,
2. ADOPT grants are in full swing, and
3. AI is here to help with more accurate computer vision & deep learning models for quality control, plant phenotyping and everything in-betweens and beyond.


Huge thank you to Innovate UK & Innovate UK Business Connect Agrifood Team, for the productive day in Birmingham 👍🏼🚀🌱



#drmarynakuzmenko #innovateuk #agrifood

1 week ago | [YT] | 1

Dr. Maryna Kuzmenko

Today, we will speak about AI in plant science.
I have two important points.



First, and in AI it is always first — we are starting with humans.
Because, surprisingly, machine learning models are nothing in comparison with real brains, giants in the field, and expert intuition.





I recently spoke about this with Dr Lidia Babenko — a plant scientist with deep practical experience, including work in controlled environment agriculture and vertical farming.





And, long story short (because we spent a good hour in conversation about how micro-changes in experiment design can lead either to success or to the inexperienced failure of a project 😉):





No matter whether you work in industry or academia, the reality is simple — AI brings a boost only when a plant expert looks at the output and says: this will work and this will not.





Because in plant science, especially in plant tissue culture, you are not just following protocols. You are dealing with a living system that:

-> reacts before you see it

-> fails before you measure it

-> decides outcomes based on factors you did not even record



So a perfectly written MS-based medium, a clean auxin/cytokinin ratio, and even a well-structured experimental plan can still bring no regeneration.



Why? 🤔

Because plants are not equations.

They are context-dependent systems.

And this is where intuition appears.

It is something like compressed experience acting instantly.



With Lidia we spoke about real examples of lab work as well as the hype around AI-generated recipes.



Some of them are correct on paper, but wrong for the selected species. Others are simply AI imagination 😂



Because the explant will not respond or the design ignores something critical.

But with 30+ years of experience, Lidia said:

“You cannot always explain it in one sentence. But you know why something will work — and it actually works”



Lidia is interested in working on new projects and would be a valuable member of your team, with the wealth of her knowledge and experience 🤝



And I thought that was the main point about AI in plant science and the role of experts in it.





But three days ago, something happened.

This is my point number two for today.



The preprint on PlantScience.ai was published.

It's about a plant-specific virtual system built to solve a key problem.



Plant literature is growing fast.

It's highly fragmented & is full of specialised terms that general LLMs often misunderstand or cite poorly.



The system built a structured plant knowledge graph from open literature, to answer questions.





It was tested on:

- 1,770 questions across

- 19 plant-science categories, with evaluation by

- 54 domain experts from the John Innes Centre & The Sainsbury Laboratory consortium,

- plus an AI judge 🤩





PlantScience.ai outperformed general models on many queries, reaching 82.3% vs Gemini Pro, 87.6% vs Opus 4.5 & 73.5% vs GPT-5 in expert comparisons.

This means more news is coming soon.



What's your experience of using AI for #plantscience? 🌱

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

1 week ago (edited) | [YT] | 3

Dr. Maryna Kuzmenko

Do you know what was in Canary Wharf in London about 100 years ago?
It was a berth built to handle fruits and vegs from the Canary Islands.
So, roughly speaking, last week we restored a little historical truth and brought berries back to Canary Wharf 😉


I was excited to attend two days of Fruitnet Berry Congress 2026 in London. Day 1 of the event took place at The Pelligon, which is a kinda big glass greenhouse. As a result, some really tasty opportunities & collabs were grown right there 🤩


Day 1


1. Technologies in berry breeding


This applies to all soft fruits too, I suggest, but please correct me if not. AI-driven predictive breeding, advanced phenotyping, genomic selection, and even gene editing were discussed as practical tools for solving real commercial problems. Berry breeders are trying to find a balance: on the one hand, transportability requires harder fruits, but this should not negatively affect Brix/acidity & texture.
Plus, it was interesting to hear about seedless blackberries. Earlier this year at Fruit Logistica, seedless fruit also took a huge chunk of attention.
An obvious trend driven by consumer wishes.


2. AI in production & monitoring

AI, particularly visual AI based on computer vision and deep learning, is present in greenhouses and fields more strongly than before. The berry sector was described as the “tip of the spear” for ag robotics, because soft fruit combines high labour pressure with a strong culture of innovation. But! Some of the lower-hanging fruit in automation may not be robotic picking itself (still a bit slow and expensive), but logistics-support robots, disease-treatment robots & AI-based crop forecasting.


3. Postharvest quality control

Postharvest quality control is becoming more predictive, not only reactive. The industry is moving from “inspecting quality” to “predicting performance”. The discussion around berry quality is moving beyond checking fruit after problems appear. More companies speak about integrated systems from pre-harvest to retail with real-time monitoring, condition tracking & better visibility across the chain.
The best inspiration of the day: keep calm and berry on 👍
Thanks, Mike, for this quote.


Day 2

What an amazing tour to DP World London Gateway.
It was especially interesting to see the logistics side behind what we later see on supermarket shelves. London Gateway is now part of a Morocco–UK–Europe fresh produce route to link Agadir / Casablanca with the UK.

Plus, my personal discovery: Gemini.

It’s not only an AI model, which I love to use 🤗
It's also a long-term operational collaboration between two logistics giants, Maersk and Hapag-Lloyd.
So there are at least two impactful Geminis in our world directly linked to berries.
Just imagine: if Gemini (AI) empowers Gemini (logistics), what a great result that would be for society 🫐🤝🍓

Final point: Strawberry Taste with Blueberry Colour.
Have you seen it?
I hope you’ll enjoy it as much as I enjoyed #BerryCongress :)

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

1 week ago | [YT] | 1

Dr. Maryna Kuzmenko

Have you seen the chocolate fountain?
I mean berries 🤩.
I mean discussions about robotic berry pickers 😉.
However, it was actually a discussion about new berry varieties and permissible indulgence 😂

As you can see, Fruitnet Berry Congress 2026 was a really nice event, with many topics, speakers, and attendees.

So I was truly delighted to be there.

Plus, it’s an actual fact — we were allowed to eat the exhibition :).
I tasted the berries, and they were genuinely nice.

Keep calm and berry on!
Watch and enjoy :)

2 weeks ago | [YT] | 1

Dr. Maryna Kuzmenko

Welcome back to our Train Reading Club 🤗



However, members of the Plane Reading Community or the Bus Reading Group are also very warmly welcomed :)



Last time, we looked at one of the main book for computer vision in agriculture.

This time, we have something a little less agri, but much more about life, ideas, courage, and inspiration.



The book is called Zero to One, and it is such an amazing title.

It has more than one meaning and more than one value.



1. On the surface, it is a book about startups, innovation, business, and the logic of creating something new. But underneath that, it is also about a mindset. About the difference between copying and creating. About the huge gap between improving something that already exists and bringing into the world something genuinely new.



2. Going from one to two is progress too, of course. But going from zero to one is a very special kind of leap. It means that before, there was nothing.

No product, no idea in practice, no visible shape, no system.

And then suddenly, because someone believed enough, worked enough, and dared enough, there is something. Hooray! 🥳



3. That is probably why the title stays in the mind so strongly.

It is short, clean, almost mathematical — but also deeply philosophical.

And the author does not require any introductions. His name is Peter Thiel.



4. For me, this is one of those books that is not limited to founders or investors only. Yes, it speaks a lot about startups.



Yes, it is often placed on the shelf of entrepreneurship literature. But in reality, it speaks to anyone who is trying to build something meaningful from scratch: a company, a project, a scientific direction, a new career, a new chapter of life, or even a stronger version of oneself.

The latter is much more important than everything else.

And that’s why I love this book.



5. Many of us spend years moving from one to two, from two to three, from three to four and further on. Yes, it’s not possible to create all the time. We are also improving, adjusting, polishing, optimizing. All this matters.

But sometimes life asks a different question: where in your world is the missing zero to one moment?

Where are you still repeating, and where are you ready to create?



I also like that this book makes you think not only about success, but about originality.

In a world full of AI, imitation, fast trends, and endless noise, there is something refreshing in the invitation to think independently. To not automatically follow the crowd. To ask whether your work is truly adding something new, or simply rearranging what is already there.



If you are still not a member of our Train Reading Club, I sincerely invite you to join me and other weird guys who carry books, not only gadgets, on their journeys 😂



Ask for your VIP pass to the Club in the comments 😁, and enjoy your journey if, like me, you are also somewhere on the road — closer or further away 🙏🏼🤝📚



#readingclub #zerotoone #drmarynakuzmenko

Skip the Carousel

1 2 3 4

2 weeks ago | [YT] | 8

Dr. Maryna Kuzmenko

Yesterday I was truly honoured to participate in the Investing in Women Code Annual Signatory Summit, hosted by NatWest.



And it left me thinking not only about investment into agritech, but about something even more direct: investing in women in ag.



A very relevant topic, by the way, because today in San Francisco the World Agri-Tech Innovation Summit opens its doors & there is no need to read tea leaves (although I definitely love to measure and “read” leaves & plants 😉): investment in agritech, and women in agritech, will be discussed there as well.



But regarding our event in London - first of all: what is the Investing in Women Code?
It was my big discovery yesterday 🤩



The Code was launched in 2019 and has now grown from 12 original signatories to 300+. These are all kinds of financial institutions, ecosystem builders, lenders, VCs, and other key players who bring money into female founders’ ideas.

Considering the scale, it is already a serious signal.


The issue is no longer niche, symbolic, or a “nice to have” to include in some ESG report or publish on social media because of IWD.
It is a structural and economic direction.



One of the strongest messages from the Summit was pretty simple: if women got the same opportunities as male entrepreneurs, women-led businesses could generate an additional £250 billion for the UK economy.



And yet female-led businesses still receive only 2% of equity investment.

You read that correctly: not 20, not 200, but just two percent.
I had to mentally reread that number too 🤔



Plus, there was another point — very interesting and thought-provoking.

A very quick question to you:
can you name one female CEO of a multibillion company who could be a role model for the next generation of female entrepreneurs?
80% of young people can’t name even a female entrepreneur (Santander UK, 2020).
That point really touched me, because it shows that the challenge is not only access to capital, but also access to visible role models.



If girls and young women rarely hear the names of female founders, CEOs & builders, then the pathway can start to feel abstract before it even begins.

And one more thing: agritech specifically.


As someone working and building in AI for agriculture, I can tell many stories about doing something in a male-dominated environment. This topic is very personal & often sits in a difficult place between “too technical, too niche, too earl,” and “impossible to explain quickly.”


Being a woman founder? It's on top of that.
I suggest you can feel the pressure & so many amazing ladies feel it too.

That's why it's a question of ecosystem-market fit. That is why this Summit mattered, and I hope we will have more events like this.


Thanks to Innovate UK as a signatory of the Code & to all organisations that speak more directly about investment into women.



Feeling the vibe of yesterday’s conference hall, I know: we will definitely continue 🤩🚺💰

#investingininwomencode #femalefounders #innovateuk

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

2 weeks ago | [YT] | 7

Dr. Maryna Kuzmenko

AI and numbers. I took our own Petiole stats to illustrate how it works.
12,684 leaves - this is one of the largest “per experiment” research batches automatically processed so far.
For operational production in the last 3 months, there are 48K+ detected objects: irregularly shaped biological residues on the surface of membranes during a quality control procedure. Very niche, but needed for compliance. No more details and obviously no photos — NDA.


1. So if you have something to count and ask, “Can AI count this?” — in 99.99% of cases, the answer is yes.
But the real value of AI counting is in volume and accuracy.


Thirty leaves are not a big deal — spend 10 minutes with our app and get your numbers. Job done. By the way we also have some enthusiastic users who took hundreds of images and process them by tapping a screen of Petiole app — that is also possible, and much faster than ImageJ for the same routine workflows. However, it’s not automation really.


Automation with AI-powered counting truly blooms when it is dozens, hundreds, or thousands of photos per day or week. Then you definitley feel the difference. It is not the same feeling as for one or ten leaves :).


2. Regarding objects that can be counted — the sky is the limit.
I officially declare: we cannot count and measure the sky.
But almost everything else is pretty possible.
Let’s consider the most popular categories:


Leaves — leaf area measurement if a calibrating plate is present; if not, greenness and actual leaf count.


Grains — count, size if calibration is in the photo, and presence of abnormal objects.

Buds — bud count and possible size measurement, but this requires a special data collection procedure (not really painful, but a bit extra-equipped :)).

Fruits — similar to grains, plus colour classification automatically; and yes, it is possible to fine-tune what “ripe fruit” means for you - you say how your ripe strawberry looks and get the automatic classification.

Stomata — done already - check photo no. 5, but required some customization, since we worked mainly with wheat leaf stomata.

Germinated plants — in a tray or in a gutter, no problem at all. Plus, classification of seedlings is possible based on size, either relative size or empirical metrics such as cm².


3. Count and grade. Count and size. Count and colour.
As you see from the previous point — it’s possible. But in all cases, count is always the foundation. It is like in real life: first we do inventory to see what we have, and then everything else.


Again, it depends on the use case, and not all objects can be further assessed. There are real limitations for some objects, and it is not about us — it is about physics.

Colour — colour assessment becomes impossible or unreliable when objects are too small, too densely packed, heavily overlapping, partially hidden, out of focus, or photographed under unstable lighting, reflections, shadows, glare, or mixed light sources. Massive batches photographed in different conditions can easily distort colour consistency, even if counting still works well.

Size — if sizing is relative within your batch — for example, you have 1,000 seedlings and want to compare them with each other — that is a perfect task, no extras needed. If you need the real size, then add a calibrating plate to the photo and we can do the rest.

Grade — the scheme is simple: you give the specification, we tell AI what to do based on that specification, whether it is a global standard, a national USDA scheme, an EU regulation & marketing standards, or simply your own SOP.


As you see, numbers are important and possible to get.


Even for those who have big numbers.
Numbers give us the freedom of business creativity, decision-making, and seeing the wider picture.


What shall we count next? You name it.
In the meantime, I will count something in the comments right now :).

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

2 weeks ago | [YT] | 6

Dr. Maryna Kuzmenko

After seeing the audiences of some of my recent subscribers, I became extremely motivated to continue my YouTube journey 🤩🚀🌍

3 weeks ago | [YT] | 3