Cedric Isubol is a web developer in Bacolod City in the Philippines. Before any of this was a product, his doctor told him that if he did not change how he lived he would be on medication for the rest of his life. He was obese, his blood pressure was high, and the sentence stuck.

So he did what most people do and downloaded the popular calorie trackers. Every one of them wanted a barcode or an exact match in a food database, and none of them had heard of chicken adobo or sinangag. Logging a single meal took ten minutes. After three days he quit, and he quit every time he tried again. The apps were not built for how he actually ate.

The idea was small and obvious once he had it. What if you could describe a meal the way you would text a friend and the software worked out the rest. Two cups of rice, grilled tilapia, and a small bowl of sinigang. No search, no scan, no database. He opened his laptop that night and started building.

He called it Caldef AI, short for calorie deficit. The stack is deliberately cheap: React Native with Expo on the front, Supabase for backend and auth, RevenueCat for subscriptions, and GPT-4o-mini doing the food parsing. Infrastructure costs him about twenty to thirty dollars a month. Around the logging he added the parts he needed himself, a deficit planner that computes a daily calorie target from the Mifflin-St Jeor formula, a weight chart that puts the real curve next to the expected one, and food suggestions drawn from the user's own country rather than American meal plans.

He ran the app on himself while he built it. In the Indie Hackers write up that anchors this story he says two months of building and using it left him thirteen kilos lighter, with normal blood pressure and off the medication watch list. Only then did he put it on Google Play.

That write up is also where the dates come from. Cedric published it on 30 April 2026 and opens by saying he launched on Google Play two weeks earlier, which puts the launch in the middle of April 2026. The two week count below is measured from that launch, not from the day he started coding.

Two weeks in he had more than a thousand downloads, fourteen active subscribers and thirty five dollars of monthly recurring revenue. Zero spent on ads. One marketing channel.

That channel was Facebook groups. He did not build an audience first. He went to health, fitness and local community groups where the people already were, and posted three things on repeat: his own before and after story, plain explanations of calorie deficits and TDEE, and native language posts showing how the app handles food that western databases have never indexed. The personal story is what carried. His own reading of it is that people do not want another corporate wellness app, they want a tool built by someone who was in the same position they are.

Price is the other half of the argument. MyFitnessPal runs around twenty dollars a month. Caldef is 179 pesos, roughly three dollars, a number chosen so the app is actually buyable in the market he was selling into. Fourteen subscribers at that price is thirty five dollars. Not a living, but recurring money from strangers.

He was blunt about what was not working. Conversion from download to paid sat at 1.4 percent, and he suspected he was losing people during onboarding with a paywall that came too early. The app was Android only because he could not yet afford a Mac to build the iOS version. And resting the whole thing on Facebook groups is fragile: one moderator ban or one algorithm change and growth stops.

He appears to have taken that last point seriously. Through May and June 2026 the Caldef site filled with a blog, more than ninety posts bylined by him, in English, Tagalog, Arabic, German, Spanish, Polish, Greek and a dozen other languages, most of them built on the same pattern of local food and calorie deficits. It is a search play running next to the group posting, aimed at the same audience the app was built for.

One caution on the numbers. Two of those blog posts retell the origin story with figures that do not match: thirteen kilos lost in one and eighteen in the other, one month of building in one and six in the other. They read as search variants rather than careful accounts, so every figure in this story comes from the Indie Hackers post and nowhere else. As of late August 2026 the Play listing showed five thousand or more downloads, and his Indie Hackers product card showed sixty dollars, about where you would expect a small paid app to sit four months after launch.