Demeter – Financial and Operational Farm Model

Demeter is an MFO-class system (Financial and Operational Model) for farms — the next generation of the FarmWise solution. It calculates and plans every aspect of farm operations: from fields and livestock, through fertilizers, fuel, and leases, all the way to subsidies, cash flow, and asset valuation. It is built on an Activity Based Costing (ABC) approach — a highly granular cost model in which every cost is assigned to a specific activity, operation, or product. The system does not operate on general overheads — every currency unit has its source, its cost driver, and its place in the production structure, making it possible to calculate the technical cost of production for both crops and milk with a level of precision unattainable in traditional budgeting models.

Calculation results are stored in a NoSQL database — a simple and convenient way to distribute computed knowledge across the company, for example as budgets, planning scenarios, or benchmarking datasets. Actual data comes from the Echo module, which parses and interprets cost invoices and writes the results into the same database in a structured form — so they can be directly compared against the plan.

Demeter’s architecture is based on the principle that each system should do what it does best. Excel provides crystal-clear calculation quality together with full model control and formula transparency. NoSQL stores and distributes computed data. Python handles integration and processing logic. React builds dashboards. Language models such as Claude support data interpretation and narrative generation. One of the key goals of this architecture is to provide AI with already calculated, reliable numerical data — in order to minimize the probabilistic nature of language models. The LLM does not calculate — it interprets. The numbers come from Demeter. At the same time, Demeter’s purpose is not recordkeeping — it is the calculation of the consequences of decisions over a forecast horizon. The system answers the question not “what happened?” but “what will happen, and what are the financial and operational consequences?”
At the computational layer, Demeter consists of nearly 4 million formula-filled cells spread across hundreds of worksheets, creating a complex and tightly interconnected logic — every module affects the others. A change in a single parameter, such as corn yield, propagates through dozens of modules: feed balance, herd feeding cost, milk technical cost of production, cash flow, EBIT. This scale of interdependence is difficult even for AI to fully grasp — which is why the module descriptions below intentionally point out cross-dependencies. Read them as a system of connected vessels, not as a list of independent functions.

AGRO: Global Data

Stores key reference data related to crops: minimum and maximum yields, crop type (winter/spring), harvest season, losses from improper storage, fuel demand per hectare, straw-to-grain ratio, and straw bale parameters. This data serves as the foundation for all financial and operational calculations for individual crops.

AGRO: Initial Data

Defines the farm’s starting state at the beginning of the analysis: herd age structure, insemination type, previous year harvests, opening technical costs and crop prices, feed inventories, and straw stocks from the prior season.

AGRO: Land Costs

Tracks costs related to the use of agricultural land. It allows specific plots to be assigned to lease agreements or agricultural loans. It makes it possible to accurately track expenses associated with individual plots and forecast future land usage costs.

AGRO: Prices – Fertilizers

Manages fertilizer and seed material costs on a quarterly and annual basis. For each fertilizer type (NPK, ammonium nitrate, urea, sulfate, potash, phosphorus, and others) as well as seed materials, it calculates cost per hectare for each crop and forecasts spending several years ahead.

AGRO: Prices – Pesticides

Tracks and forecasts crop protection costs for different crops on a quarterly basis: herbicides, fungicides, insecticides. The module is flexible and allows crop protection strategy to be adjusted in response to price changes and the specific needs of each crop.

AGRO: Action Plan

Plans and manages agronomic processes for each crop: operation schedule (stubble breaking, liming, fertilization, plowing, sowing, spraying, harvest), fuel and material demand per operation, external service costs, and machine parameters. It allows simulation of the impact of new machinery on operating costs as well as modeling the substitution of synthetic fertilizers with digestate from a biogas plant. Cost parameters are set over a 5-year period. Results are stored in the database as scenarios — for example, an agronomic plan with a new machine versus the current state, directly comparable.

AGRO: Land Costs Allocation

Precisely allocates financial and rental land costs to specific crops. For each plot and each crop, it allocates loan and lease costs proportionally to occupied hectares. It provides a 5-year forecast of costs per crop. A change in the sowing structure automatically updates costs on the crop side.

AGRO: Dotations

Manages subsidies and support payments for individual crops. It allows crops to be assigned to subsidy types, yearly subsidy values to be entered, production costs before and after subsidies to be calculated, and EBIT to be simulated at different subsidy levels (100%, 50%, 0%, and intermediate variants). It calculates margin per ton after financial support is taken into account. Different subsidy variants can be stored as separate scenarios and compared.

AGRO: Insurances

Manages crop insurance: premiums per crop with payment months, yield protection level (minimum yield covered by insurance), automatic calculation of potential payouts in case of yield shortfalls, and the effect of premiums on cash flow.

AGRO: Cultivations & Market Operations

The central crop planning module — it combines area, yields, and market prices into a complete profitability calculation. It assigns hectares to crops, takes percentage yield realization into account, uses current market prices, calculates technical cost of production per ton, margin per ton and total margin, and forecasts sales revenues. It manages the split between harvest sold and harvest used internally — surpluses automatically feed the farm feed inventory. Crop plans are written to the database as budgets and scenarios available for comparison and distribution across the company.

AGRO: Straw Production

Manages straw production as a separate product and resource. It calculates the straw-to-grain ratio, total straw weight, and maximum number of bales, dynamic herd demand for straw (bedding and feed), production and transport cost per bale and per ton, straw production profitability, and a 5-year forecast. The straw balance — own production versus herd demand — results from the interaction between AGRO and HERD data.

AGRO: Crops Technical Costs

Complete technical cost of production calculation per crop — it aggregates data from all AGRO modules. It calculates direct costs (fuel, fertilizers, crop protection products, seeds, external services), indirect costs (financial land costs from allocation), technical cost of production per ton, margin per ton and total margin, EBIT under different subsidy variants, and the value of by-products (straw, beet pulp). This is the output document for each crop — stored in the database as part of the budget. Actual data from Echo (invoices for fertilizers, crop protection, fuel assigned to crops) is automatically confronted with this plan, giving ongoing performance versus target per crop.

HERD: Global Data

The foundation of reference data for the herd. For each animal class (calf, heifer in different age brackets, dairy cow in different lactation cycles, dry cow) it defines: age and weight, daily weight gain, lactation cycle, daily manure and slurry output, bedding straw consumption, heat probability, market value of the animal, and financial classification (Investment for heifers, Earning for dairy cows).

HERD: Straw Bedding

Plans bedding straw consumption for each animal class. It defines daily requirement per animal, calculates quarterly demand taking herd population changes into account, and computes straw consumption cost per class per quarter based on the dynamic cost of in-house straw production. Data from this module directly affects the straw balance in AGRO Straw Production.

HERD: Buildings & Straw Beddings

Manages housing space and bedding schedules. It allows bedding dates per building to be set, percentage straw demand indicators to be defined, and building dimensions and floor area to be managed. It automatically calculates cow density (animals/m²) based on herd size and building area.

HERD: Cow Barn Allocation

Allocates animal groups to specific buildings per quarter. It assigns animal classes (milking cows, heifers, dry cows) to building IDs, tracks changes in herd size, calculates density indicators per building, and monitors free and overcrowded space.

HERD: Feed Formula

Creates feed rations for the herd and manages their costs. It allows up to 10 different ration formulas to be created with validity dates, ingredients (corn silage, alfalfa, beet pulp, grain, soybean meal, CCM, GPS, and others) and their quantities to be defined, ingredient cost from own production and purchased sources to be calculated, and daily, monthly, and multi-year feed demand and feed cost to be computed. Rations are stored in the database with version history. Actual data from Echo (invoices for purchased feed components) is automatically compared against planned costs.

HERD: Inventory Transactions

The central register of all herd movements: purchases, sales, transfers between classes, deaths, and losses. The system automatically assigns animals to production cycles based on age and transaction date and identifies the moment of herd exit. It enables reporting on current herd composition and transaction history. Data from this module feeds all production forecasts — milk, calvings, and feeding costs.

HERD: Feed Demand & Consumption

Plans herd nutritional demand per quarter in a horizon of up to 5 years. It distinguishes between home-grown and purchased feeds, forecasts availability of key feed components, and supports inventory management and purchase planning while taking herd production cycles into account. The feed balance results from the interaction between Feed Formula (how much per animal) and Inventory Transactions (how many animals).

HERD: Organic Fertilizer Output

Plans and monitors the production of natural organic fertilizers from the herd. It calculates monthly and daily production of manure (kg) and slurry (l) per animal class as well as total production for the whole herd. The available quantity of organic fertilizer automatically reduces planned demand for synthetic fertilizers in crop technical cost calculations.

HERD: Veterinary Services Prices

Manages the price list and cost forecast for veterinary services: vaccinations, preventive procedures, and therapies per disease category. It calculates quarterly veterinary OpEx and allows comparison of prevention costs versus treatment costs. Actual data from Echo (veterinary invoices) is compared against this plan by category.

HERD: Veterinary Vaccinations and Services

Manages vaccination and treatment schedules by animal age. It defines vaccinations per disease (brucellosis, leptospirosis, BVD, IBR, anthrax, and others) with month-of-life assignment and frequency. It automatically assigns animals to vaccination cycles based on age — without manual management per individual animal.

HERD: Illness Treatment – Probability

A statistical model of illness probability by animal class and age bracket. It covers mastitis, lameness, ketosis, BVD, leptospirosis, bacterial infections, and others. It differentiates risk by life stage (heifer, dairy cow, dry cow) and forecasts the number of cases based on herd structure. When the herd grows in a scenario, the model automatically recalculates the expected number of illnesses.
The essence of the module is to define the statistical norm for a given herd, species, and housing conditions — determining how many cases of a given disease are statistically expected in a given period. That norm becomes the point of reference: if the actual number of cases (from actual data supplied by Echo) deviates materially from the norm — upward or downward — the system generates a signal. An upward deviation may mean that something concerning is starting to happen in the herd. A downward deviation may suggest that input data is incomplete or incorrect. In both cases, the signal requires attention — Demeter does not interpret the cause, but it indicates that the norm has been exceeded.

HERD: Illness Treatment Planner

Manages the herd disease treatment schedule: start/stop dates per disease, monthly treatment frequency, average number of cases per quarter, percentage of the population covered by treatment, forecast treatment cost per disease, and veterinary budget allocation. The treatment plan results from the interaction between the probability model and the veterinary price list.
The Planner translates the statistical norms from the Probability module into a concrete cost and operational plan. If actual data from Echo (veterinary invoices by disease) falls within the norm, everything is progressing according to plan. If treatment costs for a given disease materially exceed the plan, it signals either that the statistical norm has been calibrated incorrectly for that herd or that something genuinely concerning is beginning to happen in the herd and requires intervention. A deviation below the norm also requires verification — it may indicate incomplete invoice data. In this way, the Planner functions not only as a budgeting tool, but also as an early warning system for herd health.

HERD: Insemination Prices

This module goes far beyond a simple price list — it is an analysis of insemination as an activity with full financial and production consequences. In the Activity Based approach, each insemination method is a separate activity with its own cost profile, effectiveness, and impact on the herd.
Different insemination methods — conventional semen versus sexed semen — differ not only in price, but above all in their consequences: sexed semen determines offspring sex, which directly shapes herd structure in subsequent years (more heifers means faster expansion of the dairy herd). Procedures have different success rates and different probabilities of repetition — affecting both costs and calving schedules. The module includes cost per insemination cycle, procedure cost, medicines, analyses, monitoring, and additional costs, and translates the whole picture into quarterly insemination OpEx.
Because each failed insemination means another cycle, additional cost, and a delayed calving date — and therefore a delayed start of lactation — the module is tightly linked to Dairy Cycle and Milk Production. Actual data from Echo (invoices for semen and procedures) is automatically compared with the plan, allowing the true cost and effectiveness of each method to be assessed.

HERD: Dairy Cycle

Plans herd reproductive and lactation cycles. It defines the heat cycle schedule, insemination cycles (1–7) with technology selection (traditional / sexed), re-insemination rate, automatically calculates calving dates and cost per technology, defines average milk yield per lactation cycle (cycles 1–7), and forecasts herd losses per age wave (9 waves) per quarter. Results are stored in the database as scenarios — two variants with different insemination technologies form a direct comparison of breeding strategies and their impact on herd structure and milk production over the next 2–3 years.

HERD: Calvings

Tracks and forecasts calvings and natural herd losses. It calculates the number of calvings from conventional and sexed insemination per quarter, proportions of heifers and bulls, herd renewal rate, average monthly calvings, and natural losses per reproductive cycle per quarter. It is a direct output of Dairy Cycle and feeds Inventory Transactions as planned events.

HERD: Milk Production

Forecasts daily, monthly, and quarterly milk production. This is one of the modules where the Activity Based approach is most visible — average daily production is not a simple average for the whole herd, but the result of a calculation that takes into account the age of each cow and the production cycle it is currently in. Production is matched to herd capability and milking technology. The module also incorporates the reproductive process — both the current and future insemination policy — which means that changes in reproductive strategy automatically translate into milk production forecasts in subsequent quarters. It also visualizes current feed quality and its impact on milk yield.
The key feature of the module is its ability to calculate precisely when the herd will start producing milk and what that production will look like over time — depending on whether the herd is expanding or shrinking. Because each cow is tracked by age and production cycle, the model knows exactly how many animals will be in lactation, how many will be dry, and how many are heifers not yet producing — in every subsequent quarter. Herd expansion through heifer purchases or more intensive insemination does not translate into immediate production growth — Demeter shows exactly when and by how much production will increase, taking the biological cycle of each animal into account.
It allows virtual populations to be defined in order to simulate technology scenarios (new milking technology, disease impact on production). It automatically calculates milk revenues. Forecasts are stored in the database per scenario and confronted with actual data.

HERD: Technical Cost – Milk Production

Calculates the technical cost of milk production. It combines milk revenues with full operating costs: internal and purchased feed, straw, energy, fuel, veterinary care, insemination, and insurance. It calculates margin, technical cost per liter of milk, and profitability per production cycle. This is the output document of the dairy operation — it aggregates data from all HERD modules and is stored in the database as a budget. Actual data from Echo (cost invoices assigned to milk production) confronted with the plan provides ongoing performance versus target by cost category with the ability to drill down to a specific invoice.

HERD: Technical Cost – Milk Production Benchmark

An advanced benchmarking analysis of milk technical cost of production over time. It tracks costs and revenues in detail by month and year, identifies periods of highest and lowest profitability, and compares results with previous years and other scenarios. It is the only module that looks backward — storing historical data from system inception and allowing milk technical cost trends to be analyzed and causes of change to be identified without manual data compilation.