Kanika

1. One-liner

Watches the drying curve and tells the operator when to stop the heat and start tempering.

2. Trend signal — why now?

Haryana’s Custom Milled Rice policy for 2025-26 cut the permissible broken-rice limit in CMR deliveries from 25% to 10%. Millers objected on the obvious ground that breakage during milling is a natural process — and on the arithmetic. The state fixed ₹2.23 per quintal for the additional milling cost of reducing broken rice, ₹1.23 for extra storage and ₹3.33 for packaging; the millers’ own estimate of the real processing and handling cost is around ₹25 per quintal (The Tribune, 2025-09-20). Sourabh Gupta, president of the Karnal Rice Millers and Dealers Association, has been publicly pressing for clarity on the change.

That is an 11× gap between what the miller is compensated and what the miller says it costs — and it turns every percentage point of breakage from a quality abstraction into a direct, dated cash loss on a government contract.

Which puts a spotlight on the number a miller cannot see while it is being destroyed: head rice yield, the share of whole unbroken kernels. It typically runs 50–58% of paddy, and a mill can hit 67% total recovery and still lose money if breakage is high and head rice is low (Goldmohur Overseas).

The damage is done during drying, invisibly. Paddy arrives at 24–26% moisture and must reach about 14% to mill well. Dry it too fast or too hot and you create internal fissures that cannot be seen with the naked eye — which become breakage later, at the huller (Flourtech, 2026). IRRI’s guidance is explicit that in the final stages drying air must not exceed 43°C or fissuring follows. And the research is blunt about what single-pass drying does: “Breakage of rice on milling was very high if drying was completed in a single stage”, with the fix being “dividing the drying periods and tempering the paddy in between” — dry rapidly to 18–20%, then hold without heat while moisture redistributes from the interior to the surface.

The economics of getting this wrong point in two directions at once, which is what makes it interesting.

Quality. The spread between whole and broken rice in Indian mandis is roughly ₹426 per quintal — about ₹4,260 per tonne (rice ₹3,603/qtl against 100% broken ₹3,178/qtl, CommodityOnline, 2026). Published experimental work shows the trade-off sharply: maximum whole grain yield of 52.9% came at 40°C with 40 minutes of tempering, while maximum total yield of 74.1% came at 46.6°C. The settings that push throughput are the settings that cost head rice.

Energy, which is the bigger surprise. Parboiling and drying account for more than 90% of total energy consumption in a rice milling system (Energy use pattern in rice milling industries), and intermittent tempering with three tempering phases saves up to 64% of energy versus continuous drying. So the same discipline that protects head rice also attacks the mill’s single largest cost line.

And here is the measurable defect. A study of paddy drying practice across Vietnam, Cambodia, the Philippines and Myanmar found flatbed dryers being run at 60–75°C in the field (Best practices for paddy drying, 2018) — that is 17 to 32°C above IRRI’s 43°C quality ceiling for final-stage drying. The same work records a practical drying rate around 1.5% moisture per hour, with a 4-tonne batch taking 9–11 hours from 25% to 15%. A dryer running that hot, for that long, unattended, is a fissuring machine.

IRRI is direct about where the damage originates, and it points inside the mill: a major cause of fissuring is “the moisture adsorption of individual dry grains with moisture contents below 16%… when wet grain is mixed with dry grain (in storage, in the dryer or after drying in a batch dryer with a resulting moisture gradient)” (IRRI Rice Knowledge Bank). The batch-dryer moisture gradient is a miller-side defect, and it is exactly what tempering exists to fix.

Against that, the control surface is primitive. The dominant small-mill flatbed is, in IRRI’s own description, of “very simple design” — a perforated screen, “a simple axial flow fan” on a diesel engine or motor, and “a kerosene burner or a biomass stove” for heat. Indian millers favour LSU and fluidised-bed dryers for parboiled paddy because of high incoming moisture. In every case the operator’s instrumentation is a thermostat plus a handheld capacitance moisture meter at ₹5,000–18,500 — a device that gives a spot reading when someone walks over and takes one, and says nothing about the drying rate in between. Meanwhile there are 20,314 rice mills in India as of 1 April 2026, and 98.83% of them are single-owner operations (Rentech Digital).

Provenance:

• Signal 1 (demand): Haryana’s 2025-26 CMR policy cut the permissible broken-rice limit from 25% to 10% while compensating ₹2.23/qtl against a miller-claimed actual cost of ~₹25/qtl — an 11x gap that converts every breakage point into a dated cash loss on a government contract, with a named association president publicly objecting — https://www.tribuneindia.com/news/haryana/rice-millers-oppose-haryanas-new-cmr-policy-seek-amendments — 2025-09-20
• Signal 2 (feasibility): The defect is measurable with temperature and time alone — field practice runs flatbed dryers at 60-75°C against IRRI’s 43°C final-stage ceiling, and IRRI names the batch-dryer moisture gradient as a major cause of fissuring; low-cost capacitive moisture sensing adds ±0.5-1.5% wet basis for rate detection while the miller’s existing ₹5,000-18,500 handheld remains the spot-grading authority — https://www.tandfonline.com/doi/full/10.1080/1343943X.2018.1543547 — 2018, practice unchanged
• Signal 3 (economic): Whole rice commands roughly ₹4,260/tonne over 100% broken, and separately parboiling plus drying consume over 90% of a mill’s energy where three tempering phases cut energy use by up to 64% — so the same intervention pays twice, across 20,314 Indian mills that are 98.83% single-owner — https://pmc.ncbi.nlm.nih.gov/articles/PMC4571202/ — accessed 2026-09-18 Category: Underserved niche

3. The opportunity

Yield monitoring exists — but it measures the damage after it is done, and it is priced for the top of the market. Bühler sells a cloud Yield Management System that “analyzes data straight from the scales in and out of your mill”. Satake’s REACH platform adds flow scales before and after milling so that “the real-time yield of head rice can be confirmed at a glance” (Satake, 2022-03-11). Both are excellent, both are enterprise, and both weigh the rice at the huller — which is hours or weeks after the fissures were formed in the dryer.

That is the gap: the incumbents tell a large mill how much head rice it lost. Nobody tells a small mill how to stop losing it, at the moment it is being lost.

The rest of the field sells machines or hours. Flourtech, a dryer and equipment vendor, publicly states the causal claim this product rests on — “improper drying of paddy too quickly or unevenly creates tiny internal cracks that become weak points during milling, causing grains to break” — while selling only the machinery. Nextech Agri Solutions, whose expertise is exactly “process control knowhow”, says plainly it is “not into manufacturing/selling/supply of any machinery or equipment” — it bills that knowledge as consultancy hours. Everything else located in the drying-monitoring space is academic: LoRa and microcontroller papers, no shipping product. Dryer vendors sell dryers, boiler vendors sell boilers, meter vendors sell meters, and consultants sell advice by the day. Nobody sells the judgement call during the dry.

That call is narrow and repeatable: given where this batch started, how fast it is losing moisture and how hot the air is, should the operator keep heating, ease off, or stop and temper — and for how long. Get it right and you protect head rice and burn less husk. Get it wrong and you find out weeks later, at the huller, when the breakage is already sold as broken rice at ₹426 a quintal less.

The reason this is unbuilt is not that it is hard. It is that the three parties who could build it all have a reason not to. The dryer manufacturer’s incentive is throughput per hour, which is the spec sheet number that sells machines — and tempering reduces throughput per hour while improving yield per tonne. The meter vendor sells a handheld device with no memory of what happened between readings. And the miller’s own operator is paid to clear the yard during a compressed procurement season, when paddy is arriving faster than it can be dried.

So the product is a small, rugged box on the dryer plus a phone that says, in the operator’s language, “stop the heat, temper for 40 minutes” — and then a record of whether they did, and what the batch yielded.

An honest limit on the claim, stated up front, because it is the idea’s main weakness. Fissuring accounts for 77% of head rice yield reduction from peak values — but a large share of it is created before the mill gate. Traditional field drying alone produces 39.67% fissured grains and head rice yield of 52.20%, varietal sensitivity to harvest moisture differs measurably between cultivars, and the optimal harvest moisture window for head rice (roughly 19–21% for long grain) sits above the FCI’s 17% procurement ceiling — meaning paddy arriving at an Indian mill has already passed through the window where much of the yield was won or lost.

The literature treats variety, harvest moisture, drying condition and drying duration as interactive rather than separable, and I could not find any study that partitions head rice yield variance cleanly enough to say “drying explains X%.” That number appears not to exist in citable form.

So the product cannot be sold as “your breakage is a drying problem”. It has to survive the miller’s first and entirely reasonable retort — the paddy came in cracked — which means the honest pitch is narrower: defend the portion the miller controls, prove it batch by batch in the mill’s own numbers, and let the energy saving carry the argument when the quality signal is noisy.

4. Target market

• Primary customer: Owner of a single-unit Indian rice mill running an LSU, flatbed or fluidised-bed dryer — typically 2–8 tonnes per hour, doing custom milling for state procurement and/or own-account trading. Tamil Nadu, Andhra Pradesh, West Bengal, Chhattisgarh, Punjab and Haryana are where the density is (TN alone has 3,399 mills).
• Why they buy: They already know breakage is where their margin leaks, and they already burn husk for 90% of their energy. Two levers, one box, and both are measurable within a season.
• Rough TAM reasoning: 20,314 mills, 98.83% single-owner. Realistically addressable are the mechanically-dried mills with enough volume to care — call it 6,000–9,000. At ₹2,500/month per dryer and 1,600 mills, that is ₹4.8 crore ($550K) ARR; 3,500 mills gets near $1.2M.
• Why now for them: nothing regulatory — this is an underserved niche, not a deadline play. What changed is that the sensing and connectivity to do this is now cheap enough to sell at a mill-sized price, where five years ago the same capability came bundled inside a new dryer.

5. Product sketch (MVP)

• Clamp-on drying watch. Air-inlet and grain temperature plus humidity on the dryer, reporting continuously — no re-plumbing of the machine, no PLC integration for v1.
• The tempering call. When the drying rate or air temperature crosses into fissure-risk territory — the 43°C final-stage ceiling being the hard one — the operator gets a WhatsApp message in Tamil, Telugu, Bengali or Hindi: stop heat, temper this long, then resume.
• Batch card. Each batch gets a record: incoming moisture, the curve, peak air temperature, whether tempering happened and for how long, final moisture.
• Yield tie-back. The miller enters head rice percentage per batch from their own huller output; the product correlates it with the drying curve so the argument is made in the mill’s own numbers, not a brochure’s.
• Husk-burn estimate. Energy used per tonne against tempering discipline, because for many millers this is the number that moves first.
• Season report. Batches run, tempering compliance, yield trend — the artefact the owner uses to hold the night-shift operator to the process.

6. AI angle — what’s load-bearing

Thinner than most ideas in this catalogue and I would rather say so than inflate it. The core is time-series inference on a physical process: estimating drying rate from temperature and humidity traces, detecting when a batch is heading for fissure conditions, and predicting the tempering interval a given batch needs. That is real modelling work — the relationship differs by paddy variety, incoming moisture, parboiled versus raw, and dryer type, and it has to be learned per mill rather than shipped as a fixed rulebook.

The genuinely useful ML arrives in year two: correlating accumulated drying curves against the miller’s own reported head rice yields, so the system learns what this mill’s dryer does to this variety and stops being a generic threshold alarm. Until that corpus exists, this is largely a well-instrumented rules engine with good local-language delivery, and it should be built and sold as one.

7. Localization angle

India-first and the localisation is most of the work. The advice has to arrive on WhatsApp in Tamil, Telugu, Bengali and Hindi, aimed at an operator on a night shift, not a dashboard for an office. Pricing must clear at ₹2,500/month against a mill’s actual cash position during a season when procurement payments are slow. The hardware has to survive a husk-fired, dust-laden environment.

The same product ports cleanly to Vietnam, Thailand, Bangladesh and the Philippines, where the drying physics and the LSU/flatbed reality are identical — but India is the beachhead because the mill count is enormous, the mills are single-owner (one decision-maker, no procurement committee), and parboiling makes drying an even larger share of cost.

8. Business model — path to $1M–$5M ARR

• Pricing: ₹2,500/month per dryer, hardware included on loan. ₹18,000 one-off for the box if the miller prefers to own it, then ₹1,200/month for the service. A season-only plan (4 months) for mills that run procurement cycles rather than year-round.
• ACV: ~₹30,000/year per dryer; multi-dryer mills scale linearly.
• Rough math to $1M ARR: ~2,900 dryers at ₹2,500/month. Against a base of 20,314 mills this is roughly 15% of the plausibly addressable mechanically-dried segment — ambitious but not absurd for a product with a same-season payback story.
• Rough math to $5M ARR: needs the export geographies (Vietnam, Thailand, Bangladesh) plus expansion into the parboiling side of the process, which is the other half of that 90% energy figure.
• Expansion path: drying minder → parboiling soak/steam monitoring → husk furnace efficiency → a whole-mill yield and energy record that a lender or buyer will actually read.

9. Go-to-market wedge — first 100 customers

1. Prove it on one mill, in that mill’s numbers, then sell the district. Rice milling is geographically clustered and socially networked — millers in a taluka know each other and compare notes. A single documented season showing head rice yield up two points and husk consumption down is worth more than any brochure. Target one dense district first (Thanjavur, East Godavari or Burdwan), not the country.
2. Sell through dryer service technicians, not dryer manufacturers. The manufacturers’ incentive runs against tempering. The independent technicians who install, repair and commission LSU dryers already have the relationships and the site access, and they are the people millers phone when output looks wrong. Commission per install.
3. Rice miller associations are organised, regional, and currently in a fight. State associations negotiate custom milling terms with procurement agencies, and the Karnal and Punjab bodies are publicly contesting the 25%→10% broken limit and its ₹2.23/qtl compensation right now. An association whose members are arguing that they cannot hit 10% has an obvious interest in anything that moves breakage down a point. Open with the CMR limit, not with the dryer.
4. Free drying audit as the wedge. Put the box on for one week, produce the curve, show the peaks above 43°C and the absent tempering phases. The audit costs a visit and produces a specific number, which is a far better opener than a subscription pitch.
5. Target the parboiled belt first. Parboiled mills dry more, burn more and have more to gain, so the payback maths is strongest where the concentration is highest — Andhra Pradesh, Telangana, West Bengal, Odisha.

10. Build complexity — justification

Medium. The sensing is unremarkable: temperature and humidity probes, a microcontroller, and a phone connection — there is no need for laboratory-grade inline moisture measurement, which is what makes this affordable. Low-cost capacitive grain moisture sensing reaches ±0.5% wet basis in most cases and ±1.5% in all tested cases, which is fine for tracking a rate while the miller’s existing ₹5,000–18,500 handheld remains the authority on spot grading. Designing for that division of labour is the key engineering decision.

The hard parts are physical and behavioural: an enclosure that survives husk dust, heat and vibration for a full season, and a thresholds model that is right enough per variety and dryer type that operators trust it by week three. Two people with embedded experience and a local milling advisor reach a field-credible v1 in 4–5 months, most of it spent on real dryers rather than at a desk.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Ordinary industrial instrumentation. No approval to sell. Must not be marketed as a substitute for graded moisture measurement at procurement, which is a regulated determination.
Ethical — no harm / dark patterns	✅	Improves grain quality and cuts fuel burn. The one discipline required: do not overstate attributable yield gain when much of the fissuring arrives from the field.
Market exists (evidence above)	✅	20,314 mills, a documented and quantified quality-and-energy loss, an established dryer and meter market.
1–5 person team can build this	✅	Two plus a domain advisor, 4–5 months.
Launchable with <$50K / ₹40L	✅	Prototype and a small first run fit comfortably; the spend is field time.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Quantified on three axes now — ₹4,260/tonne quality spread, 90% of mill energy, and a CMR broken-rice limit that converts breakage into a contract penalty at an 11x cost-compensation gap. Docked because the loss is invisible and deferred: it surfaces at the huller weeks later, so it is absorbed as normal rather than felt as an emergency, and seasonal throughput pressure rewards ignoring it.
Demand evidence	15	11/15	Now anchored on a dated policy change: Haryana’s 25%→10% broken limit with ₹2.23/qtl compensation against a claimed ₹25/qtl cost, objected to on the record by a named association president. Plus IRRI mechanism evidence, real price spreads and 20,314 mills. Still docked because the sourced miller voice is about the policy, not about drying — nobody has been heard asking for a drying tool.
Build feasibility	15	10/15	Sensing is off-the-shelf and deliberately avoids laboratory-grade moisture measurement. Docked for a punishing physical environment, per-variety threshold tuning, and a v1 that cannot be validated anywhere except on live dryers during a season.
Distribution clarity	15	11/15	Geographic clustering, organised associations, a technician channel and a free-audit opener are all concrete. Docked because this is feet-on-the-ground regional selling with no digital shortcut, and the buying window is seasonal.
Revenue mechanics	15	10/15	₹2,500/month against a same-season payback is defensible for this wallet. Docked because $1M needs ~2,900 dryers of low-ACV, hard-to-reach customers, and seasonal plans invite seasonal churn.
Time to first revenue	10	6/10	Hardware plus a free-audit sales motion plus a procurement season that dictates timing. Pilots are plausible in months; real paid density is a season away.
Defensibility	10	5/10	The accumulating corpus of drying curves tied to that mill’s own head rice outcomes compounds, and per-variety, per-dryer calibration is hard-won. Docked further after finding Bühler’s Yield Management System and Satake REACH already selling head-rice-yield monitoring — at the milling stage and at enterprise scale, but with the brand, the installed base and the option to move down-market if the SMB band proves out. The hardware is copyable and a dryer OEM could bundle this tomorrow.
Total	100	69/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · operations-heavy · domain-expertise-required — embedded skills plus someone who can stand in a mill yard at 2am and be taken seriously by the operator. A regional-language founder or co-founder is not optional.

Key assumptions to validate (3–5)

1. Assumption: The miller-controllable share of fissuring is large enough to matter after field damage is accounted for. How to test: Instrument three mills for one month. Compare head rice yield across batches with similar incoming moisture but different drying curves. If well-dried and badly-dried batches from comparable paddy differ by less than about a point of head rice, the quality half of the pitch is dead and the product has to stand on energy savings alone.
2. Assumption: Operators will act on the alert during peak season rather than ignore it to clear the yard. How to test: Run the alerts silently for two weeks, then live for two weeks, and measure compliance. This is a behavioural product and this is the behavioural risk.
3. Assumption: Millers will pay ₹2,500/month. How to test: Free audit on twenty mills, then ask for the subscription. Count conversions, not enthusiasm.
4. Assumption: The energy saving is realisable in a husk-fired context, not just in controlled studies. How to test: Measure husk consumption per tonne before and after tempering discipline at two mills.

Risk flags

1. Attribution is genuinely hard. Head rice yield moves with variety, harvest timing, field drying and storage as well as mill drying. Proving the product’s contribution in a noisy real-world signal is the central commercial difficulty, and it is also what the miller will demand before renewing.
2. Seasonal everything. Procurement seasons compress buying, deployment, validation and churn into a few months a year. Cash flow and learning cycles both suffer.
3. The incentive runs against the operator. During peak arrivals, the yard manager’s job is to clear paddy. A device telling them to pause the heat is asking them to go slower at the worst moment, which is exactly when it matters most.
4. Weak demand evidence. No miller has been heard asking for this. The loss is documented; the appetite is not.
5. OEM bundling and down-market incumbents. If a dryer manufacturer decides yield-per-tonne is a better sales story than tonnes-per-hour, this becomes a feature of new machines. More pointedly, Bühler and Satake already sell head-rice-yield monitoring to large mills; if the SMB band proves lucrative they have the brand and the channel to come down. The installed base of 20,000 existing dryers is the defence, and it is real but not permanent.
6. Policy reversal. The CMR hook depends on the 10% limit surviving miller lobbying. Associations are actively seeking amendments, and if the limit is relaxed or the compensation raised, the sharpest part of the pitch softens back to quality-and-energy. Track the state CMR policies each season.

14. Structured verdict

Score:                  69/100
Verdict:                VALIDATE
Confidence:             Medium
Best-fit builder:       Embedded/hardware founder with an Indian regional-language
                        co-founder or advisor from the milling trade; comfortable with
                        seasonal, feet-on-the-ground selling in milling districts
Time to revenue:        4–6 months to paid pilots, keyed to a procurement season
Capital to launch:      ₹18–30 lakh ($20–35K)
Top 3 assumptions to validate first:
  1. Miller-controllable fissuring is worth at least ~1 point of head rice yield —
     instrument three mills, compare batches of comparable incoming paddy
  2. Operators comply with tempering alerts during peak arrivals — silent-then-live test
  3. Millers convert from a free drying audit to ₹2,500/month — twenty audits, count sales
Kill criteria:
  - Abandon if comparable-paddy batches show under ~1 point of head rice difference
    between good and bad drying curves — the quality pitch is then unprovable in the field
  - Abandon if operator compliance with tempering alerts is below 50% in peak season
  - Abandon if fewer than 4 of 20 audited mills convert to a paid subscription
  - Abandon if a major dryer OEM ships integrated tempering control as standard

15. Next step — 1-week validation sprint

• Day 1–2: Get inside three working mills in one district during drying. Log what the dryer’s controls actually are, what the operator does at each stage, whether tempering happens at all, and what the air temperature peaks at. No hardware needed — a borrowed thermometer and a notebook answer most of it.
• Day 3: Pull those mills’ own head rice yield records for the last month and look for whether batch-level variation exists at all. If the mill does not measure head rice per batch, that is itself a finding — the product would have to create the measurement before it can sell the improvement.
• Day 4: Price test. Show three millers the ₹4,260/tonne spread and the 64% energy figure, describe the box, and ask for ₹2,500/month. Listen to the objection, which will be the real one.
• Day 5: Decide.

Falsifiable outcome: proceed only if (a) at least two of three mills show measurable batch-level head rice variation that tracks drying practice, (b) operators confirm tempering is skipped under load, and (c) at least one miller says yes to the price without being talked into it. If mills do not measure head rice per batch at all, the product’s first job changes from advice to measurement — a different and slower business that should be scored again from scratch.