Giving up one more nutrition tracking app.

Open the app store and search for “nutrition.” You will see four hundred apps making the same promise in slightly different fonts. Log every bite, get every answer. Most people who download one of them have already downloaded three this year. The fifth is not the one that finally works.

Nutrition apps do not fail because users are lazy. They fail because they ask a question almost nobody actually wants to answer every day: what exactly did you eat, in grams, with the brand name correct, weighed before cooking, including the splash of olive oil? The cost of the question is higher than the value of the answer, and within two weeks the icon is on the second screen of the phone.

what the app was supposed to do

It was supposed to make you visible to yourself. That part is real. Almost everyone underestimates total intake and overestimates protein. Almost everyone has one or two quiet patterns — the late-night handful, the liquid calories on Friday, the missing vegetable on Wednesday — that are invisible without a record. A record beats a feeling. That is the whole reason food logging exists.

What the app was not supposed to do is turn eating into bookkeeping. The moment the question becomes “how do I get this entry to balance” instead of “what did I notice this week,” the tool has stopped serving you. It is also the moment users quit. This is not a willpower problem. It is a design mismatch.

the one-document replacement

The replacement is unromantic. One document — a note, a doc, a single page — with three columns or three lines per day. What you ate, roughly. How you felt, two hours later. One sentence about the day. That is it. No barcode scanner. No macro pie chart. No streak.

• A meal line — “eggs and toast, coffee, big salad with chicken, pasta, two squares of chocolate.” Words, not grams. The point is the shape of the day, not the audit.
• A signal line — energy at 11am, hunger at 4pm, sleep that night. One word each is enough.
• A weekly read — once a week, paste the seven lines into a general-purpose AI tool and ask it the question the app could not: “What pattern do you see, and what is one small thing I could change next week?”

why the AI replaces the dashboard, not the food

Five years ago this approach would have been useless, because nobody was going to read seven days of scrappy notes back to themselves. A general-purpose reasoning model reads them in three seconds and asks better questions than the app ever did. “Your energy dips on the days you skip breakfast — is that a pattern you want to test?” That is a different conversation than “you are 12g short of your protein goal.”

The model does not need perfect data. It needs honest data. Honest data is what you actually write down when the friction is low enough that you do not start lying to a screen. A note app has lower friction than a barcode scanner. That is most of the upgrade.

1. Delete the food-logging app on your phone today. Not the data export — the app. The icon being there is the trap.
2. Open one note. Title it with this month. Each day, three lines: meals (words), signals (one word each), one-sentence summary.
3. On Sunday, paste the week into a free general-purpose AI tool. Ask: “What pattern do you see? What one small change would you test next week?”
4. Pick the smallest version of that change. Run it for a week. Log it the same way.
5. If after four weeks you have learned nothing, you were already eating well. That is also a useful answer.

“The shape of the week beats the precision of the entry. A model can work with a shape. A spreadsheet of guessed grams is just a tidier version of not knowing.”

what this is not

It is not for someone in active treatment for an eating disorder, or for someone whose clinician has prescribed exact macro tracking for a medical reason. Both of those are clinical contexts and the rules are different. This is for the very large group of people who have downloaded MyFitnessPal three times, used it for eleven days each, and felt slightly worse about themselves on day twelve.

It is also not a clever way to avoid noticing. The opposite. The point is to notice more — patterns, signals, the lived shape of a week — by lowering the cost of the noticing until it stops being a chore. The model does the dashboard work. You do the eating, and the brief honest record.

the quiet leverage, again

Every app you delete is a small reclamation of attention. The nutrition app in particular has been pretending for years that the bottleneck was data quality. The bottleneck was always interpretation. A free model and a one-page ledger handle interpretation better than a paid subscription handles entry.

If your phone has more than one nutrition app on it right now, that is the signal. Keep the noticing. Keep the curiosity. Drop the bookkeeping. The thing you were paying for is something you can write down in a note and ask a free model to read back to you on a Sunday.