Rural intelligence · India · Pre-revenue
India's next trillion in GDP will be made in fields and factories.
Vrisha is the intelligence layer underneath both. We generate two things that compound: a live picture of where output is actually lost, and the skilled people who recover it.
A trillion dollars of output is not a target. It is a sum of decisions made on factory floors and in fields, one shift and one season at a time.
Almost none of those decisions are made with information that already exists. Almost none of the people making them have been trained to.
The thesis
Two sectors. The same missing layer.
01
Manufacturing
The floor throws off signal all day and almost none of it reaches a decision. Output is lost to downtime, yield drift and skill gaps that only become visible in aggregate, after the quarter has closed and the loss has already been paid for.
02
Agriculture
The field is measured once, at harvest, when nothing can be changed. Input timing, soil condition and quality of practice decide the season — and none of them are instrumented while the season is still running.
03
Both, for the same reason
Both are labour-intensive. Both are where the next trillion has to come from. And both are held back by the same two absences: no intelligence at the point of work, and no deliberate way to build the skill that would use it.
What we build
One layer. Two outputs that compound.
I
Intelligence
Instrument what a plant or a farm already produces — machine signal, process data, quality outcomes, operator actions, field condition — and turn it into a live picture of where output is being lost: which line, which shift, which plot, which practice, and what each one costs.
II
Talent
The gaps the model exposes become the training that closes them — specific to that asset and that operator, not a generic curriculum. Skill stops being a soft input and becomes a measured one.
Step 1
MeasureInstrument the floor or the field as it already runs.Step 2
LocateModel where output is lost, and what it is worth.Step 3
TrainClose the specific gap with the specific operator.Step 4
CompoundBetter operators produce better output — and better data.Each cycle makes both halves more valuable. The model gets a labelled record of what actually moved output; the workforce gets measurably more capable. Neither is straightforward for a competitor to copy, because both are earned on somebody's floor over seasons.
Why now
This was not buildable five years ago.
01
Sensing got cheap.
Measurement that used to require a capital project now runs on commodity hardware and telemetry the machines already emit.
02
Models run at the edge.
Inference happens on site, tolerant of patchy connectivity — which is the only way this works in a plant in a tier-three district or on a plot with no reliable network.
03
The capital is already moving.
Serious money is being directed at Indian manufacturing capacity and agricultural value chains. What is missing is the measurement layer that would tell anyone whether it is being allocated well.
Where we are
Pre-revenue, and specific about it.
No customers yet. Nothing deployed on a live line or a live plot yet. We are building the measurement layer first, because every claim that follows depends on it being real.
What we want next is one plant and one farming collective willing to let us prove a single number: output recovered that would otherwise have been lost. That is the only result worth putting on this page, and it will go here when we have it.
We are talking to early investors and to operating partners who have a floor or a field we could learn on.
Get in touch