01
Tech-first, by default
Sensors, vision, and a single data layer go in before the first row is planted. The farm is instrumented as a system from day one — not retrofitted later.
India · Khopoli · Sahyadri foothills
Coming soon
Farming, re-engineered.
A 20-acre, tech-first, AI-first farm in the Sahyadri foothills near Khopoli. Protected cultivation, orchards, a small post-harvest line — all running on a stack we build and operate ourselves. Early days; the land is being shaped now.
Where we are
Site identified, surveys underway. We’re talking to growers, builders, and a few labs about walking the first seasons with us.
Scroll
01 · Vision
Why we’re doing this
The next decade of Indian agriculture gets written in soil and software. Today most of it stops at a dashboard. We’d rather build the farm and the stack as one thing — tech-first, AI-first, from the soil up.
So we’re trying it. 20 acres, a real plot, a small team that knows both the field and the stack. Sensors, vision models, and a few in-house agents handling the boring decisions so people can spend time on the rest.
The point isn’t to disrupt anything. It’s to build a farm that works — through monsoons, labour cycles, and the next twenty years — and to write down what the tech actually changes while we do.
02 · How we work
Four things we’ve decided early.
Not a manifesto. Just the rules of thumb we keep coming back to when we have to make a call.
02
AI-first, where it actually helps
In-house agents for irrigation, scouting, and harvest planning take the routine calls and escalate the rest. We don't want AI on a slide; we want it making the boring decisions while people focus on the hard ones.
03
Run it like a product team
Agronomists, drone pilots, and engineers sit at the same table. Prototype, measure, change one thing at a time, write it down. Most decisions on the farm are reversible — we treat them that way.
04
Buy what's good. Build what isn't.
Best-of-breed sensors, robotics, and post-harvest gear from partners; the data layer, the models, and the agents we build ourselves. The interesting work is in how the pieces fit together.
03 · What we do
Four blocks,
one farm.
The land splits into four working blocks. Each does its own job; water, power, and data move between them.
I
Protected cultivation
A small block under climate-managed cover for things that don't really tolerate the open Konkan summer — strawberries, leafy greens, herbs. Fertigation and sensing built in from day one.
II
Orchards and agroforestry
Mango, dragon fruit, and avocado planted in mixed blocks suited to lateritic soil. Long-cycle crops that fit the slope and the rainfall, planted with shade and ground-cover species, not in monoculture rows.
III
Specialty crops and livestock
A meaningful share of the land stays in rotation with traditional crops and indigenous breeds. Partly to keep the soil working, partly because we like the variety.
IV
A small place to stay
A handful of guest keys looking out over the Sahyadris. The point isn't a hotel; it's somewhere people can come, watch the farm work, and not be sold an experience.
04 · The stack
A tech-first, AI-first farm stack.
Sensors and vision feed a data layer we own. On top of that sit a handful of in-house agents — for irrigation, scouting, harvest planning — that take the routine calls and surface the rest to a human. Best-of-breed hardware where it makes sense; we build the connective tissue and the models.
- 01Data layerOne owned schema. Every block writes to it.
- 02In-house agentsIrrigation, scouting, harvest planning.
- 03Computer visionPhenology, defects, yield, pests.
- 04SensorsSoil, canopy, micro-climate.
- 05DronesNDVI, scouting, occasional spray.
- 06Tractors & roboticsInter-row work, light automation.
- 07Greenhouse controlsClimate, fertigation, CO₂.
- 08Energy & waterSolar, biogas, drip, recycling.
05 · Field notes
What we’re working on.
A short, honest log. What’s underway, what’s stuck, who we’re talking to.
On siteUnderway
Surveying the Khopoli plot
Walking the land, mapping the slopes, and figuring out where water actually goes during a Konkan monsoon. Less glamorous than it sounds, more important than it sounds.
EngineeringComing soon
First sensor lattice for the orchards
Soil moisture, leaf wetness, and micro-climate nodes feeding into one place. Nothing exotic — just enough resolution that we stop arguing about what the orchard 'feels like' today.
PartnershipsIn conversation
Talking to vision teams about phenotyping
Most off-the-shelf models don't know what an Alphonso looks like in week 14. We're chatting with a few labs about co-training on what we grow. If that's your area, get in touch.
OpinionWorking note
Why we're not building a vertical farm
Short version: the unit economics don't pencil out for India yet, and the things we want to grow don't need it. Longer version coming — happy to argue about it.
Part of the Fynd ecosystem
One company. Many frontiers.
Fynd Farm is one of four sister projects under the Fynd umbrella — each focused on a different frontier, all sharing the same engineering culture and long-horizon view.
- 01FyndCommerce platform.Where this all started. Retail and commerce infrastructure used by a lot of Indian brands.
- 02Fynd FrontierApplied research.Where our AI and agent work happens. Some of it ends up in products, some of it doesn't.
- 03Fynd FoundationPeople and learning.Our community arm — programmes and a vibe-coding course for people who want to build.
- 04Fynd FarmYou are hereA working farm.This site. 20 acres near Khopoli where we're trying to run a real farm with the same engineering instincts.
External links open in a new tab.
06 · Come and build with us
The early seasons are
the interesting ones.
We’re looking for a small set of design partners — growers, robotics and sensor teams, a few research groups, and the occasional architect. If any of that sounds like you, talk to us before we’ve cast everything in concrete.