There's no dramatic moment when you quit tracking. You log breakfast, you mean to come back at lunch, and then you don't. A day turns into a week, and a few months later you delete the app while clearing out storage, which is how most of these end.
If you've done that, you probably filed it under willpower. I'd push back on that. The tracking part is easy — eat food, write down food. What broke was the gap between deciding to log dinner and having dinner logged.
The actual work
Take a normal Tuesday. Salmon, some roasted potatoes, whatever vegetable was still good in the drawer.
In a search-based tracker that isn't one thing to log, it's three, and each one starts with a search box. You type "salmon" and get back a wall of results, most of them submitted by other users, several of them wrong, a few of them the same fish carrying different numbers. You pick one and hope. Then you set a portion, using a dropdown that offers grams, ounces, "fillet" and "serving" — none of which is the unit you'd have used if someone had just asked you out loud. Then you do the whole thing again for the potatoes, where "1 medium" is apparently a measurement.
None of that is difficult. It's just work, and it's work that comes back on Thursday, because you cook out of the same short rotation everyone cooks out of, and the tracker never learns a thing from any of it.
That's the part I kept circling while I was building Graham. The friction in calorie tracking was never the arithmetic. Adding up numbers isn't what beat you. The app made a simple thing feel like homework, and homework is a thing you fall behind on and then avoid because you're behind.
It gets worse the more normal your food is
The best case for a search-based tracker is a packaged food with a barcode. For me that's a small slice of what actually gets eaten in a week.
The worst case is a bowl of something you made from a recipe that lives in your head. There's no entry for that. So you either build one ingredient by ingredient, which takes long enough that the food gets cold, or you find something in the database that seems close and accept that the number is a guess wearing a decimal point.
I usually took the third option, which is to skip it. And skipping one meal is where the whole thing quietly comes apart, because a day with a hole in it isn't a day you can learn anything from, so there's no reason to open the app tomorrow either.
I think that's the real cost of making logging expensive. It doesn't just make you log less. It makes the log worth less, which makes you log less again.
What I wanted instead
I wanted to say the sentence I'd say to a person.
That's most of what Graham is. You tell him "salmon and roasted potatoes, the fillet was maybe six ounces" and it's logged. When he can't tell what you meant, he asks you one question — which yogurt, how big the burrito was — instead of quietly picking something and moving on. The listening is done by an AI. The numbers aren't: your targets, your totals and your trends are computed in code, so the same input gives you the same answer every time.
He also remembers. Teach him your breakfast smoothie once, with what's actually in it, and the next morning "breakfast smoothie" is two words and a logged meal. Change the recipe and you tell him once. The meals you eat over and over are the ones a tracker taxes hardest, so they're the ones worth making free.
And when you genuinely don't know — you ate at a friend's house, someone else cooked, the portion was whatever was on the plate — he'll make an honest call and mark it with a tilde so the log stays truthful about what was measured and what was estimated. A rough log beats an abandoned day. I'd rather have your imperfect Tuesday than your perfect nothing.
The part I'm not claiming
I don't think this makes tracking effortless, and I'd be suspicious of anyone who tells you their app does. You still have to notice what you ate and say so. That's the irreducible part.
What I think can go away is the data entry — the searching, the scrolling, the picking between forty versions of the same fillet, the mental math about whether "1 serving" means what you think it means. That's the specific thing that made you quit, and it's the specific thing Graham is built to remove.
If that's how your last attempt ended, it might be worth a look.
— Ryan