Running the estate,
not just reporting on it.
Execution at scale is the moat. Anyone can run a good demo plot; the hard part is repeating it across hundreds of scattered micro-plots, every cycle, with rotating crews.
Three gaps that keep
estate decisions late.
None of the three is a shortage of data. The data usually exists somewhere — a spreadsheet, a message thread, a notebook in the field. What is missing is the path from that data to a task somebody owns.
The data is older than the decision
Satellite passes weekly, the census once a season, the cost book once a month. By the time the three are reconciled into a report, the window to act on it has closed.
Averages hide the block that is failing
An estate-level score says nothing about which plots are dragging it down. Without a per-block view, effort gets spread evenly across ground that needs very different things.
Instructions leave no trail
Work assigned by phone call and message cannot be audited. Nobody can show what was done, by whom, on which block — so the same mistake gets paid for twice.
Composite census score · thresholds: ≥80 Healthy · 60–79 Watch · <60 Critical
- 4,968 plants need attention (25%)
- 991 dead plants or empty points
Last observed 15 Jul 2026 · satellite + sensors · 4–7 day freshness
Census Dec 2025 · satellite imagery 20 Jul · 98% coverage · NDVI, moisture and biomass simulated
Estate A in the working app: blocks coloured by census score, the blocks needing attention listed above the map, and the estate panel on the right carrying plant health, alerts and sensing freshness.
From raw data to a dispatched task.
Data in
Satellite, IoT, farmer input, weather and transaction history aggregated into one live layer — from financials through operations to markets.
Core engine
Models rank what needs attention — pest pressure, yield deviation, water stress — and show the reasoning behind each call. A human approves before a recommendation becomes a work order.
Precision ops
Input placement, irrigation and nutrition schedules optimised per block, with guided scouting routes for field teams.
Outcomes
Higher yield and quality with less waste, visible to the landowner and the buyer in the same real-time dashboard.
Planned
In progress
Verify
Done
Planned → In progress → Verify → Done. Every card carries an owner, a block and an SLA clock.
Twelve modules that close
the estate work cycle.
Every module shares the same objects — block, crop, worker, cost —
so adding a new module never means building a new integration.
Command Center
One screen for the estate: KPIs, open incidents, and what moved since yesterday.
Estate Map
Block polygons coloured by census score, with NDVI, moisture and infrastructure layers over them.
Estate & Crops
The register of blocks, plots and commodities that every other module reads from.
Plans & Work
Season plans broken into work orders, each with an owner, a block and an SLA clock.
Tasks
The crew queue — what to do today, in what order, with the versioned SOP attached to it.
Agronomy & Monitoring
Crop health and pest surveillance, soil and fertilisation from test results, and the cultivation library.
Water
Irrigation scheduling and water infrastructure, planned against soil moisture and the forecast.
Resources
Workforce and attendance, material inventory, harvest handling and equipment.
Alert Center
Thresholds that fire when a block drifts, ranked by what the drift is going to cost.
Performance & Risk
Cost and yield per block, SLA compliance, and the risks that need a decision this week.
Field Mode
Capture that runs fully offline; idempotent sync means a resent form never lands twice.
Census Mode
Plant-by-plant census feeding the health score, with a digital passport per tree.
Switched on per estate, in any combination. Whichever set runs, the block, crop, worker and cost records underneath are the same ones.
See everything
KPIs, NDVI and moisture layers and open incidents in one place — so a problem moves from hidden in the field to actionable now.
Prioritise by impact
A portfolio map with layers and legends lets you rank blocks by what they will cost you, not by guesswork.
Hold the line
Clear task ownership and SLA accountability across every cluster — and crews keep working when the signal drops. Field mode runs fully offline, and idempotent sync means a resent form never lands twice.
| Lot ID | Commodity | Ha | Stage | NDVI | Status |
|---|---|---|---|---|---|
| JB-PCK-014 | Chili | 14.0 | Grow · D42 | 0.78 | On track |
| JB-GRT-031 | Potato | 28.5 | Grow · D66 | 0.71 | Scouting |
| JT-WNS-007 | Shallot | 19.2 | Harvest | 0.64 | Grade-A 68% |
| JB-BDG-022 | Cabbage | 11.8 | Grow · D29 | 0.52 | Moisture low |
| JB-PCK-018 | Long bean | 6.4 | Plan | — | Land prep |
The rows above are Elevarm's own open-field horticulture. The platform runs the same way on tree crops — avocado, durian, oil palm — plus corn and sugarcane, where the health score is a composite of the census: 80 and above healthy, 60–79 watch, below 60 critical.
Impact that is measured,
not promised.
The platform does not grade its own homework. Every figure we quote traces back to a baseline recorded before we started and to a record written while the work was happening.
Baseline before change
Before anything is switched on, the block's own last cycle is recorded: yield, grade mix, input cost per hectare. Every later claim is measured against that, not against a regional average.
One ledger per block
Work orders, input issues and harvest all post against the block that paid for them, so cost and yield per hectare fall out of the operating record instead of a spreadsheet assembled afterwards.
Checked outside the system
Grade mix is graded at handover and residue results come from an accredited lab. VarmOS reports those numbers; it does not award them to itself.
The same agronomy, pointed at the smallholder.
VarmOS is the operator's tool — it dispatches work to crews we are accountable for. The Mitra App is the other side of it: the same agronomy delivered to partner smallholders in our clusters, who are not on our payroll and do not take work orders.
The farmer's own view of the cycle
What to do this week, which input at which rate, and when the harvest is due — the delivery channel for the same protocols our field teams run on, in the hands of the grower rather than the operator.
An agronomist that answers at midnight
A generative assistant inside the app. It takes cultivation questions in plain Indonesian and reads a photograph of a sick plant to narrow down what is wrong — so a diagnosis does not have to wait for the next field visit.
Advice has always been rationed by driving distance
One agronomist covers only as many blocks as they can reach in a day, which is why good practice holds where somebody is watching and drifts where nobody is. An app removes that limit — and costs us nothing we were selling.
Start with one estate.
See the difference yourself.
We use your own block structure, commodities, and cost chart — not demo data. One estate, one season, measured against its own baseline.