The farm was the most exciting place of my childhood
I come from a farming family. Going to the farm during the holidays, spending time out in the fields โ that was the most exciting part of the whole year. The smell of the soil, the sound of the tractor, the colour of a ripening crop โ all of it still lives inside me.
But in the middle of all that excitement, one thing always nagged at me. Whether I sat with family or spent time in the company of other farmers โ they all shared one thing in common: worry. Everyone was wrestling with some question. And the most painful part was that no one had a solid answer to any of them.
The same questions came back, every season, on every farm:
These were not small questions. A whole season's income hung behind each one. A wrong guess, a decision made too late, a missed diagnosis โ and months of hard work were washed away. This was not one family's story; every single farmer I met was in the same place.
As I grew up, something even sharper dawned on me โ the answers to these questions existed. In the agricultural universities, at the weather department, in the mandi data, in research papers โ it was all written down. That knowledge simply never reached the farmer, in his language, at the moment he needed it.
"My family and the farmers in my village were never short of questions โ they were only ever short of answers. khetigpt exists to answer those very questions."
So we didn't just build 'an AI'
The easy way to close the distance would have been a generic chatbot that gives every question a textbook answer. But a textbook answer is the same old problem: correct, but not about your farm. So we built khetigpt through intense research โ a custom solution in which our own algorithm layer sits on top of highly efficient LLM models.
That middle layer is our real edge. It does three things for every farmer โ it discovers, it engages, and it delivers competitive analysis.
What does this mean? When a farmer says "the leaves are turning yellow", our layer doesn't send that straight to an LLM. First it joins his district, crop, sowing date and last week's weather to shortlist likely causes โ then has the most efficient model write the answer, and finally passes every recommendation through a safety check so dosage and cost are always right.
๐จโ๐พ Where the farmer isTouchpoints
๐ฃ๏ธ Language & intentLanguage & Intent
๐ง khetigpt Farm-Intelligence LayerCustom Algorithm
โ๏ธ Efficient LLM foundationEfficient LLMs
๐ฐ๏ธ Ground-truth dataGround Truth
And choosing efficient models isn't just a technical detail โ it's exactly why we can keep it free for farmers, forever. When each question costs little, there's no need to price information out of a harvest.
Every rule carries that childhood in it
So three principles of khetigpt will never change. It will never charge a farmer โ because knowledge should not be priced out of a harvest. It works by voice โ because not every hand can type, but every farmer can speak. And it lives on WhatsApp โ because learning a new app and finding room for it on a phone should never be the farmer's burden.
We didn't want to build a middleman who stands in between and takes a cut. We wanted to build a companion โ one that walks with you all season, remembers your farm, and counts your progress as its own success.