Kodagu is India's largest coffee-producing district. For most families this is not a hobby, and the information that decides a season still moves by phone call and WhatsApp group. Kodavāme Market is the attempt to make it legible without pretending to predict it.
Price feeds are not a moat. Coffee Board pages, Agmarknet, ICE futures — any agri-startup can scrape those, and several have. Clarity is real value, and it is copyable in a weekend.
The local layer is the moat. What the curing works in Gonikoppal actually paid last Tuesday. The going picking rate in Virajpet this week. Whether it rained 40mm or 90mm on the Ammathi side last night. That information already moves through Kodagu, and it moves because people know each other. A small, high-trust district with real identities is the one asset that makes crowdsourced local data viable at all.
So the feed is the acquisition surface, and the network is the product.
A large share of Kodagu’s growers are not Kodava. Crowdsourced data dies below a participation threshold, and gating this on Kodava identity would put us under it — as well as making a price board that serves one community in a district with real communal fault lines. Market sits under the Kodavāme umbrella and is open to every grower in Kodagu.
The instinct is to forecast prices. Coffee is a globally traded, near-efficient commodity: the Indian grower’s realized price is essentially international futures minus a differential, moved by Brazilian frost and Vietnamese drought far more than by anything local. A confident number that is wrong a few times destroys trust — and can push somebody to sell at exactly the wrong moment.
Information and decision support, never financial advice. No sell signals. Scenarios with bands, not point forecasts. Sources and timestamps on every figure.
Not by feature ambition. Each layer only becomes possible once the one above it has earned enough trust.
Coffee Board daily prices, ICE futures, live INR, Kochi pepper — aggregated and legible instead of scattered across government pages. Plus a daily plain-language digest of what actually moved the market, seasonality charts, a scheme and subsidy finder, and offline calculators for cherry-versus-parchment and outturn.
Rainfall. Nearly every estate already has a gauge and someone who writes the number down each morning — a century-old habit we only have to receive. One integer, ten seconds, works offline, and it builds a hyperlocal map the IMD grid is far too coarse to give.
What a real buyer paid a real neighbour this week, and what pickers are being paid per box this month. Anonymous, aggregated, median not mean, suppressed below a minimum count, never naming the buyer. Get this wrong and the feature is dead inside a month.
Pooled-lot signalling, labour booking, estate-brand storefronts for the micro-roasteries and homestays the diaspora is already the customer for.
It is talking to fifteen growers across scale and taluk, plus two curers and an exporter, about what they check before they sell and what they argue about on the phone. Half of the plan will re-rank after those conversations. Then a WhatsApp daily card, then rainfall before the monsoon, then the calculators.