Farmers and cooperatives
Better yields without wasting fertiliser, and a plan for unpredictable weather and rising costs. Next-day soil numbers and advice in a chat.
Book the Mobile LabPutting AI in every rai
Rai stacks what is known about one plot (satellite greenness, the soil map, your lab test, this season's rain and haze), lines it up with what Thailand's crop records already show, and tells farmers and breeders what to do next. With the evidence and the doubt beside every number.
Pilot partners use RaiOS, the platform, today. Self sign-up and the public crop world are coming next.
"Should I add nitrogen this week, given the rain forecast?"
Not this week: your rice is 103 days old and already past the stage where nitrogen pays.
Based on the plot's records and the ECMWF forecast · the view is marked as opinion in the app
The plot8.9 rai of KDML105 rice · Ubon Ratchathani

Satellite greennessNDVI about 0.84 · Sentinel-2, Oct 2026

Soil mapAcrisols, 23 % probable · modelled

Soil chemistrypH 5.9 · clay 20 % · modelled, 0–30 cm
Lab testNone yet: upload the PDF, or book the Mobile Lab
Updated daily · counted 31 Oct 2026. Counting began 31 Oct 2026; growth shows once there's a week of daily counts.
Public records Rai holds and lines up, including sources whose licence is still under check. They are not validated training examples and say nothing about accuracy.
01 · The problem
Take one rice plot in Nakhon Ratchasima, 45 days after planting. Here is the same plot read two ways.
"Nitrogen is below target. Apply nitrogen now."
The only instruction a month-old sample can support.
"Flush the plot, hold the nitrogen, measure again in five to seven days."
Nitrogen is 17 mg/kg, below target. But salinity is about 4.2 dS/m, and rice starts losing yield above roughly 3.0.
Same crop, same soil, opposite instruction. The difference isn't a better laboratory. It's a fresher number. Illustrative values.
Genebanks, field trials, weather records, soil surveys and papers each keep their own format, names and terms. Nobody can act on evidence they can't line up.
In our own 48-day test on six plots, a low-cost probe's nitrogen, phosphorus and potassium lines moved in exact step with one conductivity reading (correlation 0.9999). One number, dressed up as three.
Most crosses never produce a winner, and every failed cross uses field space and a season. Meanwhile the climate the variety must face keeps moving.
02 · What Rai does
Data alone doesn't decide anything. Rai lines up independent public records into one knowledge layer, so a farmer, a breeder or an agency officer gets an answer that can cite its evidence, down to the record, the place and the season.
1Scattered public data
From 3,386 public sources, each in its own format, names and terms.
2One knowledge layer

3Answers people act on
Not this week: your rice is 103 days old and already past the stage where nitrogen pays.
Farm chat · demo plot · Oct 2026Cross RD6 with a Boa Vista × Cenit bridge parent, then backcross four times: 73 % of 1,900 simulated runs gave at least one line with both blast-resistance genes within 3.5–7 years.
Crossing plan · Oct 2026 · RD6's own DNA isn't public: a KDML105 line stands inKDML105 reached at least 90 % of the typical yield at Ubon Ratchathani in 28 of 31 past seasons, 1995–2025.
Where to grow · Oct 2026 · crop model without diseaseCounted 31 Oct 2026, including sources whose licence is still under check.
03 · How it works
The fast loop advises a farm every week. The slow loop helps breeders choose crosses over years. Both use the same rice growth model, so every season in farmers' fields sharpens the breeding predictions.
04 · Proof
Every screen below is RaiOS as it runs today, captured on a demo account. When Rai can't answer, it says so and tells you what would unlock it.




05 · Who it's for
We work with cooperatives, seed companies, universities and governments. Pick where you start.
Better yields without wasting fertiliser, and a plan for unpredictable weather and rising costs. Next-day soil numbers and advice in a chat.
Book the Mobile LabPredict from DNA, plan crosses and find where a variety does well, with the odds and the caveats on screen.
Book a demoJoint studies with recorded methods, inputs and versions, so results can be checked and published together.
Partner with usStart with a proof on your own past trials. If it can't predict the past, it shouldn't be trusted with the future. You judge.
Run a proof06 · Opening up
The crop world is opening beyond our pilot partners. Here is what works today, and what is coming.
Browse crops, varieties, places and sources without an account.
Join the waitlist →Open a free account and start a farm or a project without waiting for an invitation.
Join the waitlist →DNA files (VCF or PLINK) and lab soil reports. Private to you by default; delete an upload and everything worked out from it goes too.
Become a pilot partner →Book a demo
Bring the crop, the province and the decision you face. We'll show you what Rai can answer today, the evidence behind each number, and where it can't answer yet.
Or write to hello@everyrai.ai
Explore Rai's crop world without an account, sign up yourself, and bring your own data, private unless you choose to share it. Join the waitlist and we'll tell you the day it opens.