AI for vertical farm and greenhouse decisions, shown working
Your growing history, harvest forecasts, lighting costs and customer demand, weighed before the crop plan changes. Nothing in this layer controls equipment.
How it fits your operation
This layer turns your records into decisions: the conditions behind your stronger cycles, harvest forecasts that show their method and confidence, and lighting costs priced against your tariff. It informs the next crop plan. It never sends a command to equipment.
What changes on site
The conditions behind your best yields
Where cycles have finished with a recorded yield, the platform summarises the air temperature, humidity, pH and EC that the higher-yielding records ran at, as conditions worth trying in the next recipe rather than proof of cause.
Price the light schedule
Target DLI and canopy PPFD set the hours the crop needs. The platform tries each price period on the tariff as a start time and finds the lowest-cost window that still meets the target.
Assessment that reaches the model
Health scores, defects and stage quality recorded against the batch feed the daily features and trends that the prediction service reads.
Forecasts that say how they were made
Harvest-date forecasts use a trained model when data quality supports it and fall back to a cold-start estimate when it does not. Growth-stage estimates are currently rule-based. Both return confidence; harvest results identify cold-start estimates, while growth-stage results identify their method.
Orders beside capacity
Customer orders and available stock sit beside zone occupancy, so the team can see what is sold, what is growing and where capacity remains.
Cost before commitment
Capex, opex, energy and wastage modelling put a commercial figure beside an agronomic option before anyone signs it off.
Recipes that learn from every harvest
The trained harvest-date forecasting model reads daily records of air and water temperature, humidity, CO₂, VPD, light (PPFD, DLI and photoperiod), pH and EC, together with how closely each zone held its recipe targets, nutrient and irrigation records and growth observations. The same records train the recipe targets: from your highest-yielding completed cycles of each crop, the platform learns target air temperature, humidity, pH and EC and shows them against every batch in progress. Every recipe still runs inside the same fixed safety limits.
What is included
Each module has its own page. They share the same batch, zone and timeline as every other module on the platform.
Forecasting & prediction
Harvest forecasts and growth-stage estimates, each showing its method and confidence.
Learn more
Energy & tariff optimisation
Time-of-use tariff data against your DLI target, so light is bought when it is cheap.
Learn moreDemand-led planning
Customer orders, stock and suppliers in the same system as the crop plan.
Learn moreWhat the models decide, and what they do not. Yield comparisons summarise historical records; they do not establish cause and effect or guarantee a better recipe. Harvest forecasting can use a trained model, while growth-stage estimates use rules and observations. Tariff comparisons calculate costs from the supplied rates and lighting assumptions. A person reviews these outputs before changing a recipe or schedule.
Explore the other layers
Add the layers in this order or choose only what your site needs. Decision tools can be used without enabling automated control.
Core and monitoring
See what is happening across the site and trace each batch back to its plan, conditions and work.
Explore
Autonomous control actions
Run recipe setpoints automatically, within fixed safety limits that recipes and models cannot change.
Explore
The whole platform
Every module, and how the layers fit together on one record.
OverviewTest the forecasts against your own data
A 30-minute walkthrough against your zones, crops and sensors. No slides unless you ask for them.
- Walkthrough led by someone who has built the system
- We map your zones, crops and sensors before the call
- Straight answers on what fits and what does not