Indian spices are a billion-dollar export — the soil underneath is not ready.

Soil organic carbon is the single most load-bearing metric in Raj's account of Indian farming. A healthy soil holds about five percent. South Indian fields he tests now read 0.3 to 0.4. He uses the human-body analogy without softening it: organic carbon is the heart of the soil, and once the heart slips, every other system — nutrition, immunity, pest response — runs on steroids. The fertiliser load goes up, biotic stress increases, pest attacks become easier, the input cost rises ten percent a year minimum, and the yield does not move. The number is dangerous because it has decayed slowly enough that no single season produced a crisis.

The discipline Raj draws is to repair the substrate before optimising anything above it. Krishikan's package-of-practices begins not with crop nutrition but with soil testing, then with carbon repair, then with the calibration of urea and complex fertilisers. The package looks slower in the first season; the cumulative input cost over a three-year horizon is materially lower because the field stops needing to be propped up.

Raj's reading of the post-1979 shift is precise. The country needed more food, so the optimisation moved from soil-centric to plant-centric. Nurseries proliferated, hybridisation became default, and farmers stopped doing the fifteen-days-before-the-rain preparation his own childhood farms practised — spreading farmyard manure, watching the first rain, sowing the seed. Now a five-acre farmer orders seventy-five-thousand saplings, transplants after the first rain, and treats the soil as the dirt the sapling sits in. The plant gets the attention; the field becomes a substrate.

The cost of that shift was visible only in cumulative carbon depletion, not in seasonal output. The yield held at twenty quintals an acre across decades while organic carbon fell by an order of magnitude. The Green Revolution's productivity number is therefore not what it appears to be. It is a measurement that ignored the variable that has now started to bite.

The agronomy literature, the agricultural universities, and the bank's scale-of-finance all agree that a per-week spray on chili should cost about two thousand to twenty-five hundred rupees. The actual farmer is spending four to five thousand. Why: the local dealer is also the diagnostician, the diagnostician's compensation is the margin on the bottle, and the farmer carries a folk theory that more chemistry will buy more yield. The myth and the margin reinforce each other. Sixteen sprays at four thousand is sixty-four thousand on spray alone — against a bank-sanctioned crop loan of sixty thousand. The arithmetic does not balance even before labour and seed.

Krishikan's intervention is to substitute the dealer with an agronomist and the prescription with the package-of-practices. The total input cost compresses from one-point-two lakh to roughly sixty thousand. The yield rises ten percent, which on chili at current mandi prices is roughly forty thousand of additional revenue. Net effect: fifty-thousand-rupee per-acre swing per season.

Raj cites the standard supply-chain number: thirty-eight percent of farm-to-consumer fruit and vegetable production is lost in the chain. Spices, being non-perishable, lose less in absolute volume but lose quality at every handoff. He gives one concrete example. Red chili at harvest has thirty percent moisture; the factory needs twelve. If the farmer dries the chili on bare soil — the default — the fruit absorbs the chemicals still resident in the field, and the contamination is locked into the produce that reaches BB Royal or any other branded sourcing. Krishikan's instruction is tarpaulin sheets: faster drying, no contamination, less rejection at the factory gate.

The intervention is not a software product. It is a one-paragraph standard operating procedure delivered village-by-village by a field officer with an agriculture-diploma background. The cost to Krishikan is the field-officer salary. The cost to the farmer is the price of a tarpaulin. The recovered value is measured in factory acceptance rates and in the spice's eventual residue-free certification for export.

Raj's account of the lease-farmer trap is the most institutionally specific moment in the conversation. The landlord moves to a city for the children's education, lets thirty acres in five-acre parcels, and each lessee then has to fund the crop on land that is not legally his. The bank will not lend to him because he does not own the land. So he goes to the local shop guy, who lends at thirty-six percent and writes the chili variety, the spray schedule, and the buyback clause into the same loan. The crop is grown to the shop guy's specifications because the loan covers his preferred chemistry.

Krishikan does not solve this by making the lessee a bank-eligible borrower — the document trail makes that impossible. It solves it by being the alternative buyer at the farm gate, by financing the inputs through an NBFC product released in stages, and by paying the farmer within fifteen days of harvest. The lease farmer's options widen; the shop guy's monopoly weakens. The structural debt does not vanish; the alternative becomes credible.

Kisan-credit-card disbursements give a farmer roughly three lakh rupees on the day the documents clear, against a six-month crop cycle that does not consume the money on day one. A home loan releases against verifiable milestones — first floor, second floor, roof. A crop loan does not, because there is no bank-side infrastructure to walk the field and verify the sowing-flowering-fruiting stages. So the farmer receives the full sanction up front, real life intervenes, and a wedding or a hospital bill absorbs the working capital. The bank's instrument is solid; its release shape is wrong.

Krishikan's NBFC partner releases against the same cycle Krishikan already maps through its field officers. When the farmer buys a specific input, the NBFC pays the manufacturer directly. The farmer never holds the cash, which means the cash never becomes the family's emergency line of credit. The bank's underwriting risk is materially lower; the farmer's leakage to non-farming uses is materially lower; the field outcome is materially better.

Raj walks the chain from the inside — he was a fifty-acre okra farmer before Krishikan. At eleven each morning, a village aggregator picked up his produce, then accumulated a thousand kilograms across nearby villages, then handed off to the mandal-level aggregator, then to the taluka headquarters, then either to Bangalore's Kalasipalya market or another regional mandi. Each hand took two to three percent. The price gap from farm to retail was almost a factor of two — sixteen rupees at the farm becoming thirty at the city. Half of that lift is real cost of transport and bagging; the other half is what calcified intermediaries continue to charge.

The aggregation made sense in an era of poor discovery and difficult transport. Both inputs have collapsed in cost. Smartphones, YouTube, the regional grade of road infrastructure, the JioFiber networks have rendered the discovery problem trivial. The layers that remain are pricing for problems that no longer exist. Krishikan's structural move is to be the only buyer between farm gate and factory, at a price two to seven percent above farm gate, with a fifteen-day payment.

A spice factory needs a truckload — fifteen metric tonnes — consistently. Fifteen metric tonnes do not come from one farmer; they come from five or six farmers, in peak season, with quality variation that runs anywhere from two percent to twenty. The standard mandi workaround is the four-hundred-bag truck with the top two hundred bags matching specification and the bottom two hundred padded with degraded fill. The factory accepts because the alternative is to hold up the production line.

Krishikan's value to the factory is consistency, not price. Because the company has walked alongside the farmer through the six-month cycle, it knows the quality before harvest. The bag-padding trick disappears because there is no third-party aggregator in the middle who needs to fill volume by mixing grades. The factory pays a stable rate and stops compensating for the volatility itself. Across a year, the savings on rejected batches and on processing time more than pay for whatever premium Krishikan charges over a typical mandi supplier.

Raj's commercial argument is grounded twice. Once in spend: a Indian household spending a thousand rupees on food puts three hundred to three-fifty on spices — chili, turmeric, black pepper, cardamom, ginger. The masala segment alone is sized at fifty-thousand crore and projected to a lakh crore by 2025. Total spices is roughly twenty billion dollars. Export is less than five percent of that — the head room is enormous and visible.

Once in chemistry: chili contains capsaicin, the molecule that binds to the TRPV1 receptor, the discovery of which won the 2021 Nobel Prize for David Julius and Ardem Patapoutian. The molecule is an anti-carcinogenic agent and a body-temperature regulator. Raj's point is structural. Every spice carries a medicinal property; growing the plant with sixteen sprays of red-class chemistry turns a therapeutic ingredient into a counterproductive one. The argument for residue-free practice is not an organic marketing line. It is a chemistry argument about whether the value of the spice is preserved or destroyed.

Raj sizes the addressable problem precisely. India grows red chili on about two million acres. By his estimate, less than five percent of those acres are farmed with anything resembling the package-of-practices Krishikan codifies. The remaining ninety-five percent run on volume-game logic — chase the season, spray to the dealer's recommendation, sell at the mandi, repeat. The volume game has held because Indian spice consumption is rising fast enough to absorb the quality variance.

The arbitrage between current practice and competent practice is a yield uplift of about ten percent, an input cost compression of roughly fifty percent, and a quality grade that makes the chili exportable to US FDA and European specifications. Even at five percent penetration, Krishikan's addressable acreage is one-hundred-thousand acres of chili alone — before turmeric, black pepper, ginger, cardamom. The market is not a niche; it is the entire chili belt of Andhra and Karnataka, sitting under the current low-quality ceiling.

Raj calls the 2016-17 soil-health-card rollout the biggest revolution of the last ten years in Indian agriculture. The intent was right: every land parcel gets a card recording NPK levels, pH, organic carbon and electrical conductivity, refreshed roughly every six months as the soil chemistry shifts. The execution is uneven. Soil-testing labs are fragmented across districts. The benchmark sits with a single ICAR-affiliated lab in Hyderabad. KVKs run partial panels. Most farmers receive a card once and never re-test against the cycle.

What Raj credits the programme for is the institutional principle. The card said, explicitly, that soil chemistry is a measurable variable that should drive input choice. Krishikan integrates with the soil-health-card data wherever it exists and supplements it with portable testing kits from technology partners, building a six-parameter readout that the package-of-practices then uses. The state did the policy half; the operational half is what private players have to do.

Every Indian agrochemical bottle carries a coloured triangle indicating toxicity class: green, blue, yellow, red — with red being the most dangerous to soil and to the human handling it. Most countries have banned the red class. India still permits its sale. Raj's policy position is direct: pulling red and yellow off retail shelves is far easier than training every farmer to refuse them, because the retailer cannot sell what the manufacturer cannot ship. Once the red and yellow molecules are off the supply, the only formulations left are green and blue, which are user-friendly for the soil and the operator.

The position cuts against a default agritech instinct — build a recommendation engine that suggests safer chemistry. That instinct is correct as a secondary measure. The primary measure is regulatory. The Indian central government is already moving in this direction under the residue-free practice umbrella, but the move is slow because every banned molecule has a manufacturer lobby behind it.

Raj's split between farmer-facing and team-facing technology is operationally precise. The farmer uses WhatsApp. The agronomist's question, the photograph of the diseased leaf, the location pin — all go through the group. Backend, Krishikan's central agronomy team processes about eighty to eighty-five percent of those queries proactively, recognising the pattern from prior data, with the remaining ten to fifteen percent handled reactively through a one-on-one field-officer visit.

The internal tool is bespoke, indigenous, built from day one to fit Krishikan's specific workflow rather than purchased off the shelf. Field officers carry it on their phones; the central team reads it as a dashboard; farmers never see it. The point of the architecture is that the farmer's surface is what the farmer already trusts, and the technology only intermediates the manager's decisions. Most agritech failure modes invert this and try to teach the farmer a new interface.

Mandi sales leave a farmer waiting one to three months for full payment; sometimes the wait is six months. Inflation alone makes the deferred rupee worth less, but the more important damage is psychological. The farmer treats the post-harvest period as service fee for the commission agent's work, accepts the staggered payments, and never resists. The structural insight Raj articulates is that on-time payment is what changes the relationship, not price.

Krishikan transfers directly to the farmer's bank account within fifteen days. No paper invoices, no commission slips, no two-or-three-handoff chain. The mechanism that makes this work at scale is the zero-balance Jan Dhan account most farmers now hold, plus the universal Aadhaar-bank linkage. The financial-inclusion infrastructure built in the last decade is what makes Krishikan's payment promise operationally possible. The promise itself is the wedge.

Asked at the end of the conversation what he reads, Raj names Peter Thiel's Zero to One. He uses it not for the contrarian-monopoly thesis but for one discipline: before you spend a year building, be sure there is a customer for the problem you are solving. Many agritech founders enter the space because money is available, not because they have stayed with the problem long enough to know its texture. Raj's first startup was on the input side; his second was on perishables and the soil; Krishikan is his third, deliberately on non-perishables because the dollar income is higher and the space is less crowded.

The corollary Raj draws is about scope. A founder can change a hope, a tactic, a business-model detail. The core business should not change. He cites the negative example of Google never moving into search-adjacent territories it could have afforded. Stick to relevance; invest in adjacencies if you wish, but do not pretend that having money is the same thing as having the right to do the next thing.