One orchard record for water, nutrition, spray and compliance, and one goal for every module on it: work out what the orchard needs, then do it. Water is the module that does it today - an irrigation decision engine that advises on every zone and, on the zones you have consented, opens the valve itself. Beside it run satellite imagery with a per-tree inventory and California groundwater reporting. No other module acts on its own yet, and none of them will say it does before it can. Everything on this page names the state it is in.
One pipeline from an image of your orchard to the water you apply and the records you file. Scroll to follow it.
Regular satellite passes build an NDVI record for every block, and multispectral imagery you fly yourself comes in through open formats - COG, GeoTIFF, LAZ and STAC - to sit alongside it.
A YOLO-based model reads the imagery and finds the trees, masked to your farm boundary so it never counts the neighbour.
Every detection becomes a tree with GPS coordinates, canopy size and a health status you can filter, edit and export.
The same map drives irrigation zones and schedules, while well usage you record rolls up in acre-feet for your GSA, and a harvest lot carries through to a public trace page.
Most growers end up running an imagery vendor, an irrigation controller, a compliance spreadsheet and a traceability binder that know nothing about each other. These are the same orchard, so they are one system - and the goal for every module in it is the same one water already meets: work out what the orchard needs, then do it. We are building them one at a time, and each says which state it is in - built, being built, or not on the map - including the ones you ask about on a call. Only the module that can act says it acts.
Imagery and NDVI
Satellite passes build a continuous NDVI record of every block, so a change in vigour shows up as a trend rather than as a surprise at harvest. Free public imagery is coarser than a drone pass, and we are direct about where that limit sits: it is enough to see a block turning, not enough to count a young orchard tree by tree.
Fly Your Own Aircraft
You buy the drone you want and Persea reads what it produces. Ingest runs on open formats rather than on one manufacturer's cloud, so a DJI T400, a competitor, or whatever replaces both in three years all land in the same place. Multispectral payloads only, because that is what makes a drone pass yield real NDVI - the same index the satellite series is built from, not a differently-coloured picture.
AI-Powered Per-Tree Inventory
A YOLO-based model reads the imagery and turns it into an inventory: every tree a row, with a position you can click. Detection finds about half the trees on the first pass on the orchard where we measured it, so it starts as an inventory you review and correct, not a census. Correcting it is a normal part of onboarding, and the map is yours to edit. Detection is masked to your farm boundary so it never counts the neighbour's block or the riparian strip, and each detection carries the model version that produced it, so a better model can re-score your history rather than restart it.
Water, Zone by Zone
Water gets to a block three ways, and they stay separate. You start and stop a zone from the dashboard. Your schedules run: Persea opens the valve at the time you set, on your farm's own clock, and closes it on the duration you set, while on the schedules you leave weather-aware a forecast-rain check drops the occurrence before anything opens and logs why. And separately from your schedules, the water-balance engine reaches its own decision - advisory on every zone, and on the zones you have consented it opens the valve on that decision itself. Consent is one affirmation, made under step-up authentication, covering the zones it lists for you at that moment; a zone added later stays out until you affirm again; every autonomous run is logged; and you can withdraw at any time, which also stops runs already in flight. Daily reference evapotranspiration and crop coefficients drive the demand picture and the FAO-56 deficit planner, so you can see what each block is asking for before you decide what to give it.
Groundwater and Traceability, Each on Its Own Record
Groundwater you record against a well rolls up against your allocation in acre-feet on the October-to-September water year. Zone runs are a separate record and do not feed it - a run's applied water is its run time multiplied by the flow rate you configured for that zone. A harvest lot carries its blocks through to a public trace page a buyer can open.
Pesticide Use, and the Report You File
Spray sits under the same goal as every other module here - the orchard's needs worked out, then met - and for spray, meeting them means autonomous application, which is the goal this module is being built toward. Spray records and pesticide-use reporting are being built now, alongside the groundwater and traceability that already ship. The record itself and the monthly California Pesticide Use Report pack are written; what is still missing is the screen a grower enters a spray on, so nothing here is usable in the product yet. When it lands, Persea will keep the record and produce the report ready for you to file - Persea will not submit anything to any agency on your behalf, and filing will stay yours to do. Restricted-materials permits and Notices of Intent will be out of scope, and we would rather say that now than let you find out at the deadline.
Disease and Pest Detection
Catching disease and pest pressure early is the next module on the map. It is not built, and we would rather name it than imply it: the capabilities it ends up with will be decided against real orchard cases, not sketched here first.
A working session on your orchard rather than a slide deck: your boundaries, your zones, your water. You will also get a straight answer about what is built today and what is still being built.
We walk the platform against a real block, not a demo farm
Bring a water bill or pump log and we work the water math with you
Every module gets one of those three answers, including the ones you ask about