AgriCreditERP Suite
Software that runs the farm, and turns its own records into credit.
It records what a crop costs while it is still growing, and turns that into a credit file a bank can read.
Bipin Sapkota, Aayush Shah
Two parties,one season
One grows the crop. The other sells it. Everything a bank would want to know happens between these two.
Together, every season leaves a full record: what was spent, what was grown, what it sold for. Today, that record lives only in a notebook.
The problem:No record, No score, No loan
Everything worth knowing already happens on the farm: what was planted, what it cost, what was delivered. The venture sees all of it, and writes none of it down. An ERP would hold that record. A credit engine would turn it into a score. Neither exists today, so no record, no score, no loan.
He grows well and delivers on time, season after season.
Nobody writes it down, so no bank has ever heard of him. He cannot borrow in his own name.
It pays for every season out of its own pocket. When that cash runs out, it stops taking farmers.
A bank could fund the rest. But a bank wants proof, and there is none to show it.
By rule 12% of its lending must go to agriculture, rising to 15% by 2028. The money is sitting there.
But it cannot tell a good farmer from a bad one. So it lends blind, or it does not lend at all.
What we willbe building
It runs the farm: plots, contracts, inputs issued, field work, harvest, sale and traceability. It earns its keep on its own, before a rupee of credit is lent.
It reads that record, scores the farmer on six checks out of 900, and sets a credit limit. The credit goes out as goods, not cash, and comes back from the buyer’s payment at harvest.
How the creditengine works
Lease, contract, inputs issued, field work, weighbridge ticket, grade sheet
What the crop has cost so far, plus what it still needs. That is the base.
Six checks, 900 points. Every check shows its evidence.
Goods from an approved supplier, never cash. Recovered from the buyer's payment.
It records the season as it happens: plots, contracts, inputs, field work, harvest and sale.
The credit engine reads that record. No ERP, no score.
Band A 630+ → 70%
Band B 540–629 → 60%
Band C 405–539 → 45%
Below 405 → referred
Capped at 80% of the buyer order.
A field officer visits the plot and confirms the category, or moves it up or down one.
The score proposes. The officer decides.
The venture sets a budget against the category and issues it as an LC, never cash.
The farmer draws inputs against it at an approved supplier.
Before the season starts, the venture and the farmer sign a contract: what will be grown, on which plot, and what the venture will pay for it.
Every plot insured,every claim checked in the field
Bad weather hits a whole district on the same day, and no score can protect against that. So every plot is insured before planting starts, and a field officer visits in person before any claim is paid.
Every plot is insured before planting starts, the crop and the field both.
The policy, how much it covers, and how much government paid all sit on that plot’s record.
If a crop is damaged, a field officer goes to the plot, notes how far the crop had grown and how much was lost, and takes photographs.
Every step, from the claim to the payment, goes on the same record with his name on it.
A farmer who loses 8 tonnes to a flood is not treated as a farmer who failed to deliver 8 tonnes.
An insured plot also gets five points more credit than an uninsured one, up to a limit of 75%. That is why farmers insure.
What wealready have
We do not need to buy anything in. Everything this build needs is already inside the Group.
Software engineers already on the Group’s payroll, with spare capacity.
Built at cost. No vendor, no licence fee, no foreign currency, and the software stays ours.
A product manager to run the build, and the Group’s own marketing team to sell it.
Field operations and training for the pilot come from people already employed here.
The Group’s own agri venture gives us a real season to build against, free.
We prove the software there first. It is a test site, not a sale.
A competitor has to find a venture willing to try software nobody has run yet. We have one inside the Group, so we reach our first customer with a full season already proven.
Whobuys it
Three buyers, and each one already has the problem this software solves.
What it costs, and when it pays back
This starts only after the build is finished.
The same subscription sells to a bank and to government, which pay for the data rather than the farm tools. None has been approached yet.
to build it, then pilot it free on the Group’s own venture to prove it on a real season. We sell only after that.
repay the whole build inside the first year of selling. Every subscription after that renews each year.
buy the same software: sugar, dairy, poultry, tea, herbs, co-operatives and banks. One build, many ventures.
Why us, and what we are asking for
We sit inside a group that already runs an agri venture, so we can prove the software on a real season before we sell it to anyone.
We build it rather than rent it, so the software stays ours to sell.
AgriERP, FarmERP, AGRIVI and Cropin are strong on the everyday features. We say so plainly.
None of them scores credit from the business’s own cost records. They score from satellite data the farmer cannot see. Ours is published, with the evidence beside it.
We sell software. We never lend, never hold anyone’s money, and carry no credit risk.
NRB’s digital lending guideline lets a technology provider help a bank read a borrower’s record, with the lending risk staying where it already sits, with the bank.
NPR 30 lakh, four people for six months, and one named sponsor at business-head level. In return: a working ERP and credit engine, proven on a full season at the Group’s own venture before we sell it outside.
Three stages, each ending in a test that must pass before the next begins: define and build (months 1 to 6), a free pilot season on the Group’s own venture, then the first outside sale.
Questions and where to read further
Every figure on these slides is written up and sourced. The proposal and the full scope are behind the QR code.
