SAY IT
t r a c k i e
LOGGED

Tell it what you ate. In your own words.

Your meal

Eggs, boiled
“two boiled eggs”
2 large · 100 g 155kcal
ESTIMATED
Whey protein
“a scoop of whey”
1 scoop · 30 g 120kcal
ESTIMATED
Toned milk
“around 300 ml toned milk”
300 ml · 309 g 174kcal
MEASURED
Banana, raw
“one small banana”
1 small · 101 g 90kcal
MEASURED
Meal total kcal
47 g protein 08:30, breakfast

One sentence. A whole meal log. No database search, no twenty lookalike entries, no weighing every bite.

The problem

Food tracking fails because logging is tedious.

Form apps TAPS: 3…
Search

People give up within weeks.

Just-ask-the-AI apps ASKED: 1×
Same meal, asked again

Confident. Precise-looking. Made up.

AI interprets. Code calculates.

The model reads the sentence
TwoCOUNT boiled eggsFOOD · AS EATEN and oneCOUNT smallSIZE bananaFOOD

It returns words and their meaning. It never returns a number of calories.

Plain code does the maths
Banana, raw · USDA FoodData Central
89 kcal per 100 g ×101 g eaten (1 small) ÷100
= 90kcal
nutrients per 100 g × grams eaten = total

Same meal. Same answer. Every time.

Questions

Asks only what matters.

“Two boiled eggs”
Small, medium or large?
Tap an answer. It moves the total by up to 37 kcal.
“One small banana”
You already said. No question.
“100 g paneer”
“300 ml milk”
Used as-is.

A meal with several foods gets at most the few questions worth your time.

Honesty

Honest when it doesn’t know.

If trackie can't find a food it can trust, it says so, instead of quietly swapping in a lookalike.

gulab jamun
“two gulab jamun after dinner”
SEARCHING USDA…
POWDER · 1 KG
gulab jamun MIX (dry)
Closest name match · 402 kcal / 100 g
NOT THE SAME FOOD
Trust labels

Every number says where it came from.

MEASURED

You stated the weight, volume or size.

“300 ml toned milk”
ESTIMATED

A typical portion was used because you didn't say.

“a scoop of whey” → 30 g
ASSUMED

You skipped a question, so an average was used.

eggs · size skipped → medium
Meal total
Energy kcal
Protein g
Carbs g
Fat g
Fibre 2.6+ g Fibre: missing for 1 item (whey)
Missing is never counted as zero.
Real food data

Powered by USDA FoodData Central, kept locally.

foods, raw to as-eaten
dalrotiricepaneerapplebananaeggtoned milkdalrotiricepaneerapplebananaeggtoned milkdalrotiricepaneerapplebananaeggtoned milkdalrotiricepaneerapplebananaeggtoned milk
curdpohaidlichicken currylemongulab jamunwheyroticurdpohaidlichicken currylemongulab jamunwheyroticurdpohaidlichicken currylemongulab jamunwheyroticurdpohaidlichicken currylemongulab jamunwheyroti
Built to be measured

Every change is scored.

SAMPLE VALUES, NOT RESULTS
Accuracy 92% Sample figure for layout
Right questions asked 88% Sample figure for layout
Honesty about unknowns 97% Sample figure for layout
Cost per meal <1¢ a fraction of a cent
Accuracy by run Each change runs the full test set.
past run latest target
92%
TARGET 90
RUN
Status
EARLY & HEADLESS: the engine comes first.
NOW

Sentence → trusted meal log

NEXT

Break dishes into ingredients when a dish isn't in the data.

NEXT

Remember your usual foods and portions so logging gets faster.

Just say it.

No spam. One email when it's ready.