Sensors tell you what to do. This is the record of what you did.
Every spray, irrigation, harvest and repair, logged against the block it happened on — spoken into the chat from the field rather than typed into a form at nine at night. It becomes your spray register, your cost per hectare, and the warning that stops a harvest landing inside a withholding period.
A system that only knows its own advice is half blind.
TerraCast could tell you a block needed water and could record that you accepted the suggestion. What it could not tell you was whether anyone actually irrigated, with what, for how long, or what it cost. Advice accepted is not work done — and the moment you need to prove what went on a block, to a buyer or an auditor, the difference is the whole point.
So the record came second, deliberately. It is not a farm management system bolted on the side; it is the other half of the same loop. What you did feeds back into what the advisor tells you next.
One timeline per block, whatever the job was.
Products, rates and the interval they carry
Each product on the job is a line item with its rate and quantity. The pre-harvest interval and re-entry period are copied onto the record at the moment of work, so editing your catalogue next season cannot rewrite what was true last season.
Hours, litres, and millimetres applied
Enter it however you have it and the millimetres are derived from the block's area. A run that actually happened through a connected control module writes its own record — no one has to remember to log it.
Yield per block, in the units you weigh in
Tonnes, bins, crates. Yield per hectare falls out of it, and so does cost per tonne once the season's inputs are against the same block. Mixed units are never silently added together.
What you saw, in your words
A walk of the rows, a note about botrytis on the western end. Kept verbatim rather than reduced to a category, because the wording is usually the useful part when you read it back a season later.
Hours on the job, and the service they push toward
Machines are registered once for the business and based at a farm. Hours logged against a job accumulate against the service interval and raise a reminder as one comes due. The hourly cost is what makes cost-per-hectare mean anything, since machine time is most of it.
What is in the shed, without a stocktake
An append-only ledger: purchases in, jobs draw down. On-hand is summed from the movements rather than stored as a number that can drift out of step with its own history. Per shed, so two farms are never one pile.
Say it once, in the field.
This is the part that decides whether a records system is used or abandoned. Forms do not get filled in from the cab of a tractor, and a log written up at nine at night is a log of what someone remembers.
So you tell the advisor: "sprayed block 4 with Kumulus this morning, 300 litres." It comes back as a card showing exactly what it is about to write — block, product, rate, date. Nothing is saved until you confirm it. If it got the block wrong, you say so before it lands, not after.
The warning arrives before the mistake, not after the audit.
Pre-harvest intervals, computed from what was actually sprayed
Not from what was planned. If a block is inside a withholding period, it says so on the morning briefing — the same list you already read before deciding where to drive. A harvest booked inside one is flagged as a breach rather than discovered by a buyer.
Who can walk into which block, today
Re-entry periods run from the recorded application, so the block a picking crew is headed for is either clear or it is not. It is a question with an answer, rather than a phone call to whoever did the spraying.
A spray register you can hand over
One row per product per application, exportable as CSV. Because the record is append-only — a correction writes a new entry that supersedes the old one, nothing is edited in place — the register can show its own history rather than just its current state.
Cost per hectare, from the work rather than the invoice.
Product prices are snapshotted onto each job at the time it happened, and machine hours are costed at that machine's hourly rate. Add them against a block's area and you get cost per hectare that came from the jobs themselves — split by spraying, fertilising, irrigation, harvest — and cost per tonne once the fruit is in.
Nothing is inferred where it is not known. A job with no priced inputs reports no cost rather than a zero, because a zero reads like "this was free" and quietly drags an average down.
Your own history becomes something the advisor can recall.
Records are indexed into a corpus of their own, kept separate from the agronomy knowledge base so a note you wrote about your own block is never confused with published guidance. Ask "what did we spray on Block 4 last spring, and did it help?" and the answer comes from your farm, with the dates and products it is built from.
The longer you use it, the more it has to reason from — which is the honest version of "the system gets smarter". Not a model that improves on its own, but an advisor with more of your farm's actual history to answer from.