Predicted iron
5.2 mg/100g
calibrated 4.8–5.6 · 95%
The vision
Agriculture & Food, Reinvented.
We are building a future where food is grown, and rewarded, for its nourishment and sustainability. 3D farming is the engine: we measure nutrient density from how a crop is grown, verify it from the food’s own biology, and reward the farming that raises it — one connected system that turns nutritional quality into something provable, and valuable, from farm to plate.
How? 3D farming — three dimensions, in order —
1 Farm to food
→
2 Food to farm
→
3 Innovation engine
See how it works
3D Farming = New Food System
3D farming is Vitagri's name for one connected system worked in three dimensions, in sequence. First farm to food — predict. Then food to farm — verify. Then a digitalised innovation engine that accelerates sustainable change across the whole farming and food industry.
The engine · interactive
Two models that check each other — when combined, accelerate innovation
Most certification trusts the paperwork. This design does not have to — the food’s biology is the second witness. Choose a direction and watch the engine run.
Input · the farm
How it’s grown
The decisions a farmer makes.
● Start hereTillagereduced · no-till
Nitrogen ratekg N / ha
Soil organic matter% SOM
Cover cropsspecies · duration
Predicted iron
5.2 mg
Forward prediction
Logged & calibrated
The model · stacked ensemble
Output · the food
Predicted nutrition
What the model expects this food to be.
● Start hereIron5.2 mg/100g · predicted
Zinc4.1 mg/100g · predicted
Polyphenols+14% vs typical · predicted
Calibrated certaintyrange 4.8–5.6 · 95%
The model
Stacked ensemble
GLMM · XGBoost · RF → ridge
Cross-check
Forward = Reverse
→ audit-grade ✓
Predict
What the engine does
One engine, two jobs
The closed loop proves what happened. The same engine, run as an optimiser, discovers what should happen next.
Job 1 · the closed loop
Verify & trace
Audit-grade verification, provenance and anti-greenwash. The bi-directional check proves a nutrient-density claim from the food’s own biology — traceability that cannot be talked around.
Job 2 · the flywheel
Innovate & accelerate
Run as an optimiser, the same engine searches for the farming techniques that will raise nutrient density next — and prioritises the trials most likely to work. Vitagri becomes an R&D platform, not only a certifier.
Step 3 · the digitalised innovation engine
It’s time to digitally connect from farm to plate — and accelerate the optimising of the whole food system.
Verification proves what already happened. Read the loop one more way and it becomes a search for what should happen next — scoring a whole season of farming in milliseconds, trialling the winners, and confirming every gain in the food’s biology. This is how data science digitalises innovation in farming: each verified cycle sharpens the model and narrows the search, accelerating sustainable change across the whole farming and food industry.
Farm sideFood side
- 01 Search · every way to farm
- 02 Optimise · rank the biggest gains
- 03 Trial · winners to the field
- 04 Verify · confirmed in the biology
- 05 Learn · fold back & retrain
Search space remaining 11,400
New, biologically-verified farming techniques — cumulative
Cycles run
0
Techniques verified
0
Best nutrient gain
+0%
Discovery rate
×1.0
Watch new, verified farming techniques compound, cycle by cycle
Why this is the asset
Credibility that is structural, not promised
Certification businesses live or die on credibility. The bi-directional design is what makes Vitagri’s credibility structural rather than promised — and hard for a new entrant to copy.
01
Fraud-resistant by biology
C15:0 and C17:0 fatty acids are synthesised exclusively by bacteria in ruminant digestive systems. They cannot be supplemented into milk or meat. The reverse model’s evidence is a biological fingerprint, not a declaration.
02
Audit-grade, not self-reported
When the forward prediction and the reverse reading agree, the verification is evidence-based. That is a categorically stronger claim than a paperwork audit — and a stronger product to license.
03
A compounding data network
Every certified farm produces validated, labelled data that retrains the model. More data → better accuracy → more credible certification → more farms. The dataset, not the algorithm, is the durable advantage.
Gearbox
For the technical reader
The narrative above is the whole story at one level of zoom. Under the bonnet, the forward model is a stacked ensemble — a GLMM, XGBoost and a Random Forest combined by a ridge meta-learner — with isotonic-regression calibration and a Brier-score threshold so its confidence scores are honest rather than optimistic; the reverse model reads biological markers such as the odd-chain fatty acids C15:0 and C17:0 that ruminant gut bacteria make and no one can supplement in, which is what makes verification fraud-resistant; and run as an optimiser the same ensemble becomes a fast simulator that searches for the next nutrient-density gain. The full engineering treatment — every model, the calibration maths and the optimisation layer — is taught in Module 9 of the Vitagri Academy. Learn the technical detail in the Academy →
Questions
Frequently asked
3D farming is Vitagri’s name for one connected agricultural system worked in three dimensions. First, farm to food: a forward model predicts a food’s nutritional outcomes from how a crop is grown. Second, food to farm: a reverse model verifies those outcomes from the food’s own biology, with no paperwork required. Third, a digitalised innovation engine: the same predictive model is run as an optimiser to discover and accelerate the farming techniques that raise nutrient density. The three dimensions form one closed, self-improving loop — which is why we call it 3D farming.
A two-way machine-learning system. The forward model predicts a food’s nutrient density from how a crop is farmed (tillage, nitrogen rate, soil organic matter, cover-crop use). The reverse model takes biological markers from a food sample and infers what farming system produced it. Together they form a closed verification loop: one side predicts which farms should produce nutrient-dense food, the other checks whether the food actually carries the biological signatures of that farming.
Most food and farming certification is self-reported and audited on paperwork. The reverse model removes the reliance on self-reporting — a food’s biology cannot easily be faked. When the forward prediction and the reverse reading of the biology agree, the certification is evidence-based rather than declaration-based.
The forward model predicts a food’s nutritional outcomes from how it is grown; the reverse model verifies them from the food’s own biology. Together they give assurance schemes an evidence base that does not rely on self-reporting — and because every check is digital, audits that today depend on paperwork and farm visits can run continuously in software. Vitagri is building these digital verification tools on top of the model.
The biological markers the reverse model relies on — odd-chain fatty acids, microbial diversity, polyphenol profiles — arise from fundamental biochemical pathways in living organisms. They cannot be artificially created, supplemented, or faked. C15:0 and C17:0 in particular are synthesised exclusively by ruminant gut bacteria and cannot be added to milk or meat.
Every certified farm generates validated, labelled data that retrains the model. More data means more patterns, better accuracy and more credible certifications — which attracts more farms, a compounding virtuous cycle. A new entrant cannot replicate the accumulating verified dataset without the same network of verified farms.
Both. Beyond verifying and tracing farming for compliance, the same forward model can be run as a simulator — it scores proposed farming-decision combinations in milliseconds. The system searches for the farming predicted to raise nutrient density, prioritises the highest-expected-gain field trials, validates the outcomes with the reverse model, and feeds the results back to retrain the model. Verification becomes a continuous-improvement flywheel that accelerates the discovery of new nutrient-density-raising techniques over time.
The full technical treatment — the stacked GLMM, XGBoost and Random Forest ensemble, the ridge meta-learner, isotonic-regression calibration and the Brier-score threshold — is taught in Module 9 of the Vitagri Academy, and the underlying evidence base is set out in the Growing Health white paper.
In collaboration with the Bionutrient Institute
“We, at the Bionutrient Institute, are delighted to be working with Vitagri to accelerate the measurement of how farming can increase nutritional density.”
Dan Kittredge · Bionutrient Institute
Join the network
The engine for a nutritional density future
The model gets stronger with every verified farm. If you are an investor, a research institution, or a food business that wants nutritional quality to be provable — we should talk.