An agricultural ERP that turns the records a venture already keeps into a credit file a bank can lend against.
Needs NPR 45,000 of inputs at the right week.
No land title, no bank statement, no credit record.
Advances seed and fertiliser to its growers already.
Carries that risk on paper registers, with no way to price it.
Must hold 10% of loans in agriculture. Has no discretion.
No applicant arrives with a file that can be underwritten.
Everyone wants the same transaction. Nobody can evidence it.
The same entries, made with a date, a plot and a crop attached, become a credit file: production planned against production delivered, quality, timing, cost on the standing crop.
Those entries are already made — in registers and spreadsheets, so they never add up to anything a lender can read.
We are not collecting new data. We are making existing data legible.
Four layers. Each one is useful without the ones above it. Each further layer is a decision, not an obligation.
Land, leases, contracts, planning, procurement, inventory, cold store, field work, yield capture, grading, sales
Cover bound before sowing; claims raised from the field record
Plan against delivery, quality mix, guarantee — a score, a band, a limit
The credit file to a lender, with consent. A loan on the farmer's phone
Six checks, read from records the venture already keeps. No bureau file.
Lending percentage applies to the assessed base for the farmer's band.
A maize contract commits 20 tonnes. He delivers 20: 15 at contract grade, 5 off-grade, inside the window.
Quantity passes in full. Timing passes. The 5 tonnes off-grade count at the weight set for maize — not zero, not full.
Off-grade maize still sells at a lower price. Off-grade tomato may be unsaleable. The weight is a parameter the credit committee sets.
Has a bank account with history and financial records of his own. Assessed on his documents, plus the venture's record of him.
Nothing on paper anywhere. Assessed entirely on the venture's record of him — which is why that record has to be created in the first place.
Identity, referees, GPS-verified plot. Lease and contract.
Lease and contract. No credit.
Contract signed, venture guarantee.
Starter route: flat 50%, capped NPR 35,000, as goods.
One delivery record — quantity, grade, timing.
A band and a proper limit.
Three seasons of history.
A credit history a bank lends against on its own.
Category B is the majority — and the reason the project is worth doing.
Four revenue lines from three payers, so no single one carries the case.
First-year build and pilot. Four people, devices, one field season.
Contribution per enrolled plot, per year, after direct cost.
Payback. The pilot alone is modelled to average 1,500 plots live.
We are not building a lending app. We are building the record that makes a farmer lendable — and handing it to a bank that is already obliged to lend.