I build Calk, a calorie tracker for iPhone and Android that logs food as editable meal templates instead of database searches. This page is about why it exists. The short version: I counted calories for ten years, got tired of it, and found the data underneath was broken in ways nobody was fixing. Why Calk exists
I started with MyFitnessPal, then spent years on FatSecret, then bought MacroFactor, Cal AI and SnapCalorie at once when photo recognition arrived. Every one of them assumes the diary never ends. I don't think that is the job. A focused month tells you what you actually eat; after that the scale trend is the quieter signal, and you come back for another short check only when something changes. Why food logging should have an endpoint
Most people who quit tracking don't quit because the math is hard. They quit because the same twenty meals have to be typed in again, every day, forever. The fatigue is the product's fault, not the user's. Calorie-tracking fatigue and how to start again
Pick "chicken breast" in any tracker and you get five rows with a twofold spread: raw or cooked, skin or no skin, a brand entry, a user entry, a unit guessed by someone in 2014. Choosing a row is not logging. It is guessing with extra steps. Why calorie databases disagree
The real problem is that food is not a product, it is a recipe. I love soup, and soup cannot be looked up. I took two ordinary home borscht recipes, both the kind half the country cooks. One came to 44 kcal per 100 g, the other to 125. On my 400 g bowl that is 176 versus 500 kcal, in a dish with one name. The difference is broth on bone or water, oil in the pan or not, a spoon of sour cream. Those are the things a database row cannot hold. How the cooking method changes calories
That is where meal templates came from. Save the borscht once, as its parts. Next time change only what changed today: no sour cream, smaller bowl, meat broth. Two taps, and the assumptions stay visible. What a meal template is and how Calk logs a real meal
I photographed a bowl of borscht. A few days later I photographed a different borscht in a different bowl. The app returned the same answer both times, byte for byte: 542 kcal, 31 g protein, 35 g fat, 27 g carbs. Inside the breakdown was 1.97 grams of onion. Nobody has ever cooked borscht with two grams of onion. The photo had recognised the dish name and pasted a stock recipe scaled to "one bowl". Photo calorie counting: what 18 accuracy tests found
A camera is good at telling you what the food is called. It is poor at the two things that carry most of the calories: how much is on the plate and how it was cooked. A model that guesses those will be confidently wrong, and the confidence is the dangerous part. Calk shows you the assumption and lets you correct it instead. How Calk compares with photo trackers on search, photo, barcode and recipe
Once I saw how bad the rows were, I couldn't ship someone else's. Calk's food data is built from national food tables and open sources, one ingredient at a time, with the reasoning recorded next to each value. Automated tests run on every change and catch a value that drifts from the evidence. How Calk tests its food data
Accuracy is a claim, so I publish it as one: what was measured, on which version, with which method, and where the estimate is still an estimate. How accurate is Calk
The point of the month is not a number per day. It is a picture: which meals you repeat, which food groups barely appear, what a normal week costs. Calk turns the record into a Nutrition Report and a Variety Map, so you can see the gap and pick one thing to add rather than a diet to follow. What a 30-day food audit reveals
Maintaining weight is a different job from losing it. Losing asks whether the trend is moving. Maintaining asks whether anything changed and where to look if it did. Most trackers are built for the first job and quietly fail at the second. Weight maintenance is a different job
Calk runs on your device. Food records stay there unless you export them, and there is no ad network reading your dinner. Privacy and your data
I am one person, and the app changes as I learn. If you find a wrong value, a broken source or a claim that reaches beyond its evidence, write to [email protected] and I will fix it.