Solutions by species
Four records, whatever is eating the feed. Hedge is built on those primitives rather than on pigs and chickens specifically — which is why the same ledger carries a farrowing house, a layer shed and a banana block without three separate systems arguing about the same bag of feed.
Species
Status
What Hedge does about it
Farm Creek built this on its own two sites before it was sold to anyone, so what follows is what a Rwandan farm gets in the first cycle — whichever species is eating the feed.
The tally sheet at the shed door does not change. Hedge reads it as it was written — Excel, a PDF, a photograph of handwriting — and nobody in the shed retypes anything.
The five-step pathNineteen per cent of rural Rwanda is online. An entry that needs a connection to exist is an entry nobody makes, so capture holds on the phone and settles when signal returns.
Built for the fieldHead counts that cannot go negative. A cumulative total that cannot be filed as a day's figure. An edit that has to leave a reason. The guards sit in the database, not the form.
Control and auditA management report and a full statement pack under your own letterhead, in PDF, Excel or CSV — and every figure on them drills back to the rows that made it.
What the modules doPigs
Rwanda keeps 1,006,890 pigs. Ninety to ninety-five per cent of the farms holding them are smallholdings of one or two sows; there are fewer than a hundred commercial pig farms in the country. Thirty-eight per cent of farms keep any record at all, and the slaughter slabs keep none.
Plate 01 · Pigs
Feed is first by a distance, and feed is the one constraint a record can act on immediately — because a feed issue booked against a named pen is the difference between knowing your cost per kilo of gain and guessing it.
381 Notifications to WOAH, with 7,463 deaths recorded between 2005 and 2019
Outbreak reporting is only as good as the farm record behind it. A mortality event that reconciles against head count — rather than drifting from it — is the difference between a notifiable event and an unexplained shortfall at year end.
One cysticercosis cyst deducts RWF 5,000 at the slab — twenty to forty per cent of a medium pig's value. Nothing at the slab is written down, so the loss never reaches the farm's books and the same herd repeats it next quarter.
Frontiers in Veterinary Science, pig value-chain mappingThree of the five constraints above are answered by the same thing: a booking against a named pen, on the day it happened. Not a note in a book that agrees with nothing else.
A feed issue is a stock movement and a journal entry at once, charged to the pen. Cost per kilo of gain is then a figure the system already holds, not a month-end guess.
Mortality moves head count, so a death cannot quietly drift from the herd it came out of. What you notify and what your books say are the same number.
Herd, ages, movements and treatments are already the record PSTA 5 will ask for by 2028/29 — kept because it is useful, not assembled the month the rule lands.
Layers
1,272,783 layers, a third of a national flock of 3.86 million. Purchased birds outnumber home-hatched ones eight to one, which means nearly every layer in Rwanda came off a commercial hatchery with a breed manual behind it. Rwanda produced 112,856,811 eggs in the last survey year and sold 68% of them — the market is not the constraint either.
Plate 02 · Layers
The manual for the birds you bought already says what they should return. Closing that gap needs three numbers kept honestly, every day.
Eggs collected, birds lost, feed issued and water, against the house that produced them. A day's lay rate is a day's rate — the system refuses a cumulative total filed as a daily figure, because that is how a flock quietly reads at 181%.
Your lay rate and cumulative eggs per hen housed, set beside the manual for the genetics you actually bought — not a generic average.
Feed issued to the shed is a stock movement and a journal entry at once, so cost per egg is a figure the system already holds rather than a month-end reconstruction.
Broilers
240,711 birds, 6.2% of the national flock, selling into a domestic chicken-meat market worth RWF 153 billion — thirteen times the egg market. Rwandan houses clear at 42 to 45 days and 2.4 to 2.5 kg. The Cobb 500 supplement puts 2,521 g on day 35.
Plate 03 · Broilers
Nobody hides those ten days. They simply are not measured, because measuring them needs a daily weight and a daily feed issue booked against the same house, on the same date, in the same system — and on most farms the weights live in a notebook and the feed lives on an invoice. Hedge posts both to the same batch, so the cumulative feed conversion of a house is a number you can read on day 30 rather than a number you infer after the birds have gone.
A batch is a clock. Every figure below is stamped with the day it belongs to, which is the only reason the tenth extra day can be seen at all.
The day the house was filled, every weight sample since, and the day the birds left — one batch, one timeline, so age is never reconstructed from memory.
Cumulative feed and cumulative weight against the same batch, so the FCR is a number you read on day 30 rather than one you infer after the house is empty.
A house standing full is a house not filled. Clearance date and cost per bird sit in the ledger together, so the decision to hold or clear is arithmetic.
Crops and plantation
Banana covers 267,676 hectares of Rwanda and the crop grew 4.1% last season, past 1.3 million tonnes. Almost none of it is recorded by block. A plantation writes down what it sold, because that is what came back as money — so the bunch given away, the bunch that split on the path and the bunch eaten at home vanish, and the yield of the block is wrong by exactly the amount nobody was paid for.
Plate 04 · Banana
A block is a herd that does not move. It has a population, it takes feed, it yields and it loses — so it runs on the same four records as the sheds, and its costs land in the same ledger.
Mats are counted per block and inputs are applied to a block, drawing stock down as they are recorded. A fertiliser run is a cost with a place attached, so the block that is losing money can be named.
Sold, collected by the buyer, given away, damaged, taken for home use — five dispositions, so the harvest that earned nothing is still on the record and the damage rate is a figure rather than an impression.
Mats, bunches per mat per year and average bunch weight are held against each block, so the forecast is computed from the plantation you actually have instead of last year's tonnage plus optimism.
Other species
The primitives carry across: cattle, fish, rabbit and bees are all a herd or a batch, a feed issue, a growth record and a loss. But no Hedge customer runs them today and we have no Rwandan benchmark data for them worth publishing, so this is a description of what onboarding would involve — not a product tour.
Plate 05 · Not yet
Plate 06 · Fish
Cattle
The animal, the weight log and the feed issue already exist. What is missing is a lactation curve, a dry-off calendar and milk recorded per animal per milking — the one place we would want a customer in the room before building.
Fish
Stocking, sampling weights, feed and mortality map cleanly onto what is already there. Dissolved oxygen, temperature and water exchange do not, and a tilapia operation without them is not being managed.
Rabbit
Does, kindling, litters and weaning are structurally the same as a farrowing house. This is the one we would expect to run soonest, and the one where we would ask you to check our stage transitions before trusting them.
Bees
Colonies split, abscond and merge, so head count is the wrong primitive entirely. Honest answer: this would be new modelling, not a configuration change.
If you run one of these, come and be the first. We will tell you exactly what exists, what we would have to build, and how long it would take — and if the answer is that you should use something else this year, we will say that instead of selling you a roadmap.