Satellite-planned crop-cutting samples so states settle crop insurance sooner
TL;DR Cutplot uses satellite imagery to plan where and how many crop-cutting experiments each insurance unit needs under India's crop insurance scheme, then tracks the field teams doing them. State agriculture departments pay, with insurers as a second buyer. Faster, cleaner yield data means farmers' claims stop waiting on paperwork.
Under PMFBY, a farmer's claim hangs on the yield the state measures through crop-cutting experiments. The rules ask for at least 4 experiments per village for major crops and 8 for others, and 24 per district where the district is the unit[1]. The scheme document itself admits this process lacks reliability and speed, and that this hurts claim settlement[1].
You can see the cost of that delay. In 2018, farmers' claims worth ₹2,829 crore from two seasons were still unpaid, with the ministry saying most rabi 2017-18 claims hadn't even been estimated yet[2]. That figure is old, but the bottleneck it describes still runs through the same field experiments.
What it does, in regulatory terms: decision support for sampling. Cutplot does not set yields or decide claims. The state remains the yield authority, and the scheme's Technical Advisory Committee rules on experiment design[1].
Before harvest, Cutplot maps each insurance unit from satellite imagery and groups fields by crop health. Uniform units get a smaller sample but patchy units keep the full count, with plots placed where the variation is. The scheme document already describes this method and expects it to cut the total number of experiments by about 30–40%[1]. During harvest, a phone app checks that each experiment happens at the assigned plot, with time stamps and photos, and flags results that disagree with the imagery.
I reframed the seed, which pitched satellite analytics to stop farmers' yield losses. Farm advisory from satellites is a crowded space, and no image ever stopped a drought. The narrower gap is the measurement step that decides when insured farmers get paid.
The state agriculture department pays, because it runs the experiments. The scheme splits the cost of technology for these experiments 50:50 between the centre and the state[1], which lowers the state's bill. Insurers are a second buyer - they want yield figures they trust and fewer disputes. My assumption is pricing per insurance unit per season, set against what states spend on field staff today.
The government's own crop forecasting centre, MNCFC, now supports PMFBY with technology programmes and ran YES-TECH training on crop simulation models in January 2025[3]. States are being asked to bring technology into yield estimation, and they need vendors who can actually run it at village scale.
There's no public figure for state spending on crop-cutting experiments or on technology-based sampling. The firmest signal is the scheme rule itself - a 30–40% cut in experiments is the stated target[1], and the centre co-funds the technology. Treat the unpaid-claims figure[2] as evidence of pain, not of budget.
MNCFC and the agencies it works with are the incumbents for technology yields[3]. A Bengaluru satellite-analytics firm sells farm intelligence, mainly to lenders, and reports 85 million hectares monitored[4]. Agritech platforms run agri-intelligence tools for agribusiness. I didn't find a vendor focused only on planning and auditing the crop-cutting sample for a state.
| Difficulty | Moderate (48/100) |
| Starting capital | ₹70 L–1.4 Cr |
| Time to launch | 9–15 months |
The remote-sensing work is well known. Selling to state governments is the hard bit, along with winning tenders and getting the Technical Advisory Committee to accept a smaller sample. You'll also need field partners in each district during harvest, because a plan nobody follows on the ground proves nothing.
No licence to operate, but you can't sell anything until the scheme's rules allow your method.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
AI must not set yields or decide claims. Those stay with the state and the Technical Advisory Committee, and bad imagery in cloudy seasons can mislead any model.
Want to work out the AI layer for Cutplot? Talk to Yash →
Honestly, this is a slow one. It suits a founder with remote-sensing depth who has already sold to state agriculture departments. If you need fast revenue then walk away now. Prove first that, in one district, the smaller sample gives yields within an agreed margin of the full sample.
Plan for roughly ₹70 L–1.4 Cr, mostly for remote-sensing staff, field partners and the long wait for state contracts.
No licence, but the Technical Advisory Committee must accept your sampling method and state procurement rules apply to every contract.
Yes, wherever insurers rely on field yield checks. Check how your local scheme sets yields and who buys the data.
The free Cutplot validation kit has the line-by-line capital breakdown, the effort scoring, the scheme and tender checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for states and insurers, and 5 assumptions to test. Download the kit →
Thinking about building Cutplot or something close to it? The make-or-break work is getting one state's technical committee to accept a smaller, satellite-planned sample, and I can help you plan that pitch and the money behind it. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.