Somewhere around batch thirty, one of us said out loud what everyone had been thinking privately for two weeks: maybe dosa just doesn't want to become a chip. It was 11 p.m. in a rented kitchen in Dindigul, and that day's batch had come out of the air fryer looking correct and tasting like cardboard that vaguely remembered being a dosa once. We ate it anyway, because throwing away a full batch felt worse than eating a bad one, and went home more tired than the actual cooking should have made us.
The problem, we eventually figured out, wasn't the frying. It was everything before the frying.
A chip magnifies every flaw a plate hides
Real dosa batter is forgiving on a tawa. Pour it, spread it, a slightly-off ferment still turns into something edible, maybe not great, but edible. Turn that same batter into a thin chip and every flaw gets magnified instead of hidden. Under-fermented batter came out flat and starchy, more like a rice cracker than anything resembling dosa. Over-fermented batter turned aggressively sour in a way that worked on a plate with sambar to balance it, but tasted almost unpleasant on its own with nothing to cut it. We had to find a fermentation window narrower than what any single household recipe actually uses, because a household dosa gets eaten with three other things on the plate, and a chip gets eaten completely alone.
Then there was thickness, which sounds like the boring part of this story and turned out to be the hardest. Too thick, and the air fryer left the centre gummy while the edges turned nearly burnt. Too thin, and the whole chip shattered into powder before it even reached the seasoning stage. We went through more failed thicknesses than we're comfortable admitting in one paragraph, testing on the same secondhand air fryer that eventually earned a name in our group chat, because we'd used it for so many attempts it felt rude not to.
Guessing by taste only gets you so far
Somewhere around the fourth month, we brought in someone who actually studied food technology for a living, because three engineers and an MBA guessing our way through fermentation science had clearly hit its ceiling. That's Vandhana, who now runs quality control and R&D here, and the first thing she did was make us throw out our informal "tastes about right" testing method and replace it with something closer to an actual protocol, timed fermentation windows, consistent batch sizes, proper documentation instead of someone's memory of what batch nineteen tasted like. Embarrassingly, our results got dramatically more consistent within about two weeks of doing this. We'd been doing real food science by vibes for four months. It mostly worked, until it didn't, and then it took a trained person to explain why.
The seasoning took its own separate education. We assumed, going in, that dosa chutneys would translate easily onto a dry chip. They didn't. A wet chutney and a dry seasoning are fundamentally different things, texture, moisture, how flavor releases in your mouth, and the peanut chutney flavor alone took several rounds to stop tasting like "peanut" in a generic sense and start tasting specifically like the roadside chutney it was supposed to evoke. Coconut was worse. Coconut's flavor fades fast once it's dried and seasoned onto something crunchy, so getting a coconut chip that still tasted like coconut by the time it reached someone's mouth took more iteration than any other flavor in the lineup.
None of this was glamorous. Most days looked like small adjustments, five degrees less on the air fryer, twenty minutes more on the ferment, a slightly different grind on the rice, followed by a lot of family members and friends eating the results and being asked to describe exactly what was wrong, because "not bad" was never a useful answer to anyone in that kitchen.
What we ended up with, after enough of these small corrections stacked on top of each other, was a process, not a shortcut. Soak, grind, ferment for hours, air-fry, season with real spices and herbs, test again. Every flavor on our shelf failed several times before it passed. We're telling you this not because failure makes a good origin story, though it does, but because it's the actual reason our ingredient list stays short and our fermentation time stays long. We tried the fast version. It tasted like the fast version.