Built for anyone counting protein

Know what you actually ate,
and whether you got enough protein.

Describe a meal in plain words or photograph it, and the numbers come back in one message. Every meal carries a seal saying where its figures came from — tap it for the range, the confidence and every item. Whether you are building muscle, eating better, or eating on a GLP‑1.

No fabricated numbers. Every meal says where its figures came from, and a guess is labelled a guess.

A food and protein tracker, not a medication service. Nutrician does not prescribe, recommend, or help you obtain medication. If you take one, the prescription is between you and your clinician; we help with the eating.

Download on theApp Store Start in your browser

Seven days free, then a free tier you can stay on.

Built by a professor of computer vision who publishes his sources. Guessing looks exactly like knowing ›

The Nutrician dashboard: protein, carbs and fat as dials, calories and fluids side by side below them, and the conversation folded away above the composer.

How it works

The coach cannot hand you a number
it did not source.

Most food apps show a calorie count and leave you to assume it is right. Here the model writes the sentence, and the figures are filled in afterward from what you actually logged. Anything it tries to state on its own is stripped before it reaches you — and every meal is stamped with where its numbers came from.

“two eggs and toast”› broken into items› label, menu or database› stored

from a food database the seal on the meal. Tap it for the range, the confidence and every item.

an unsourced guess› the gate

Blocked dropped mid‑sentence, before you see it.

This is architecture, not a prompt. Where a food is in no database at all, the app still gives you a figure, and labels it plainly as its own estimate with a range. What it will not do is dress a guess up as a fact.

These are the seals, in the app’s own words. You will see them on your first meal.

exact numbersexact figures — from a label you scanned or numbers you gave me, not a guess
restaurant’s numbersthe restaurant’s own published numbers
your numbersthe numbers you gave me are exact; the rest is still my guess
from a food databasematched to a trusted food database
guessed from your photomy best guess from your photo — treat it as about‑right
Weakest winsA meal takes the weakest seal inside it. Averaging would flatter us.

Adequacy, not restriction

Protein is a floor to reach,
not a ceiling to stay under.

Protein is a floor, not a ceiling

Something to reach, not stay under. Whether you are lifting hard or appetite is gone, the day’s card points you at one of your own meals, and at one thing worth trying. A light day is never called a failure.

Ate half of it? Say so

The numbers rescale by that fraction. Or photograph the plate afterward and it works out what is left, instead of counting a portion you did not eat.

Shot day, side effects, food noise

All of it in the same chat box. Anything that needs a clinician is pointed at your prescriber — plainly, and every time.

Building muscle? A different number, same floor

Set your own protein target and the app holds you to reaching it. Choose the muscle‑building goal and a calorie surplus is the plan, not a fault.

The progress panel: a measurement field, a list of wins such as clothes fitting better and more energy, and a note explaining a pattern in the data.

The rest of it

A dashboard for the day,
a conversation when you need one.

The add menu open over the dashboard, offering a food photo, a nutrition label, how you are feeling, and nearby restaurant menus, with a note on which photos are sent and which stay private.

The day at a glance

Protein, carbs and fat as dials, with calories and water below them and what to eat next. The conversation is one tap away, folded until you want it.

Knows where you are

Sitting in a chain restaurant, it works out the venue and matches your meal to that chain’s menu where we hold it. Where we don’t, you get a food‑database estimate with a range, and the seal says which.

Repeat a meal

Yesterday’s dinner is two taps at dinnertime, copied exactly, nothing guessed twice.

Apple Health

Weight reads in so your trend stays current. Meals write out with their name, and the source seal rides along in the export.

Honest streaks

Rest days are free and nothing is sold back to you.

What we will not claim

We are not the most accurate app.
Nobody can prove that one.

Portion size is the hardest part of tracking food and no app has solved it, this one included. A database entry is an average, not a measurement of what is on your plate. Even a nutrition label is a range: federal rules let the protein in a food run as low as 80 percent of the declared value, and the calories as high as 120 percent.

The claim we do make is one you can check. Every figure shows its origin, and our guesses look different from our facts.

The long version, on why AI food apps sound so certain and what we built instead: Guessing looks exactly like knowing.

Know what you actually ate,and whether you got enough protein.

The coach cannot hand you a numberit did not source.

Protein is a floor to reach,not a ceiling to stay under.