It would've been easy to build FasalDrishti as a pure satellite product. Point a scan at a field, spit out a health score, call it agritech. Plenty of products do exactly that. We think it's the wrong call, because a satellite can tell you where something's wrong. It can't tell you what.
A patch showing chlorophyll stress could be nitrogen deficiency. It could be moisture. It could be a pest problem building up, or just a cloud shadow that happened to sit over that pixel. Remote sensing is great at flagging where to look. It's not a substitute for knowing what's actually in the soil.
So the pipeline runs in two stages, on purpose. The satellite scan zones in on stress first and raises a hypothesis — possible nitrogen deficiency here, possible moisture stress there. It's labeled exactly as that. A hypothesis. It doesn't become a fertilizer recommendation until a soil lab result confirms the real N/P/K/Mg/pH numbers for that zone.
This matters more in Indian farming than almost anywhere else. A wrong fertilizer call isn't an abstract inefficiency. It's money a smallholder farmer often can't get back. Skipping the soil check to ship a flashier demo would've been optimizing for us, not for them.
Same discipline shows up in a decision that's easy to miss: which satellite feed runs by default. We use the free Planetary Computer feed, not a paid tier dressed up as a trial. Paid Sentinel Hub sits alongside it for anyone who wants higher resolution, but the core scanning a farmer relies on was never going behind a paywall.
Same logic for the AI advisor built into the app. It's a free, locally hosted model trained on Indian agronomy, answering questions with no per-query cost and no data plan required. A tool that needs a subscription to help a farmer make a better call hasn't actually solved their problem.
Where to look and what's wrong are two different questions. FasalDrishti answers them in that order, every time.