The same AI judged 200 real plates: off by 53% from a photo, 3% when the plate was typed.* The gap is the oil, the butter and the dressing: calories a camera can't see. See the full test.
Ramen is the food diary where you type what you ate and it counts the hidden parts. Join the waitlist
Six spoons, six pages
Hidden calories live in spoons: a glug of oil, a pat of butter, a ladle of dressing. They're mostly fat, so each spoon is big. Here are six, per tablespoon, with a full page for each.
| Food | Calories per tablespoon | Fat per tablespoon | The page |
|---|---|---|---|
| Olive oil | 119 | 13.5 g | Calories in olive oil |
| Butter | 102 | 11.5 g | Calories in butter |
| Peanut butter | 96 | 8.2 g | Calories in peanut butter |
| Mayonnaise | 94 | 10.3 g | Calories in mayonnaise |
| Ranch dressing | 65 | 6.7 g | Calories in ranch dressing |
| Half-and-half | 20 | 1.7 g | Calories in half-and-half |
Each page goes deeper: the numbers by the teaspoon, a typical restaurant pour, the scenes where that food hides and what to write.
What the camera sees, and what it doesn't
A camera sees the top of the food, in whatever light you had. Oil is clear and soaks in. Butter melts. Dressing glazes the leaves. Mayo sits between two slices of bread. Cream turns coffee beige. None of it changes how the plate looks, and all of it changes the number.
One ordinary day, four spoons
Say you do this in one day:
- 2 tablespoons of olive oil to cook dinner: 239
- 1 tablespoon of butter on toast: 102
- 2 tablespoons of ranch on a lunch salad: 129
- 1 tablespoon of mayo in a sandwich: 94
That's about 560 calories, and not one of them shows in a photo of the meals. Add a scoop of peanut butter and the cream in your coffee and it's more. Each spoon is small, normal and forgettable. Together they make a whole meal.
Say the spoons
That's the whole method.
chicken and broccoli, pan-fried in 1 tbsp olive oil
salad, 2 tbsp ranch
turkey sandwich, 1 tbsp mayo
toast with butter and 1 tbsp peanut butter
Ramen counts each line and puts calories and protein beside it. If you leave the oil or butter out, it adds what's likely for that dish, and a note on every line says what it counted and assumed. Too much? Type only 1 tbsp dressing and it's fixed, with Undo. Ramen can remember your fix for that dish, so you aren't repeating it every day.
Restaurants hide the most
You can't see the pour in a restaurant kitchen. Sautés, steaks finished in butter, tossed salads and sauces are where spoons get big. Our restaurant pours on each page are typical estimates, so you get the ballpark. To log a dinner you can't measure, read how to track calories when eating out.
Where the spoons hide, dish by dish
The same dish can be twice the calories from a restaurant kitchen. Each page shows it bare, homemade and ordered out.
- Breakfast: scrambled eggs, omelette, avocado toast, pancakes, acai bowl
- Takeout: fried rice, stir-fry, pad thai, ramen, sushi rolls
- Classics: burger, french fries, pizza, chicken tenders, grilled cheese
- Bowls and dinners: Caesar salad, pasta, chicken curry, burrito bowl, poke bowl
What the spoons add to your plan
On a cut, the oil nobody saw eats your deficit. On a bulk, the real number matters just as much. A typed line gives you the spoons, and the spoons are where the number hides.
Pick lose, hold or build and Ramen turns those spoons into a daily plan: calories, macros and a pace. It never changes on its own, so a heavy-pour week can't quietly move the target. Your weigh-ins draw a journey chart, the weekly progress screen gives a verdict, and you get a sticker when you reach your goal.
Ramen costs $4.99 a month, or $29.99 for the year (under $2.50 a month), and the yearly plan starts with a 3-day trial. It's coming to iPhone.
*Typical values per spoon are from FoodData Central, the US government's food database (linked on each page); brands and recipes vary, and restaurant pours are our estimates. Test: 200 cafeteria plates from Google's Nutrition5k research data (CC BY 4.0), each estimated by the same general AI, once from a photo and once from a typed description, with the typical error 53% and 3%.