The numbers behind it
A food app that makes its numbers up looks exactly like one that takes them from the literature — you cannot tell the difference until someone says where they came from. Here is where.
How much you need to eat
Your height, weight, age and sex give your resting energy expenditure through the Mifflin–St Jeor equation — the one the Academy of Nutrition and Dietetics recommends, and the one that lands within ten per cent of measured expenditure for four people out of five who are not obese. That is then multiplied by your activity: 1.45 · 1.65 · 1.85, the bands set by the FAO, WHO and UNU. Online calculators often multiply by 1.2 or 1.375; those figures get passed around without a clear source, which is why the app sticks to bands that are backed by measurement.
Losing fat and building muscle
Losing fat is −20% (at most 750 kcal), getting lean −12% (at most 400 kcal), building muscle +12% within a 200–500 kcal band. That comes from what Garthe and colleagues measured in athletes: those who lost weight more slowly (0.7% of bodyweight a week) gained lean mass and got stronger — those who lost it twice as fast only held on to their muscle. Faster fat loss is not better. The app never goes below 1,200 kcal a day for women, 1,500 for men and 1,350 if you leave sex out.
Protein
Per kilo of body weight, not as a percentage of calories — if it came from energy, a deeper deficit would mean less protein, which is the opposite of what someone trying to keep their muscle needs. With everyday activity: getting lean 2.2 g/kg, losing fat 1.8, building muscle 1.8, maintaining 1.6. If you sit most of the day, it’s 0.2 less (not when building muscle); if you’re on your feet a lot, 0.2 more. Getting lean always stays at 2.2. From 65 the app eases protein down gradually, and from 70 it never goes above 1.5 g/kg. The International Society of Sports Nutrition puts people who train at 1.4–2.0 g/kg; getting lean goes higher because protein matters most in a deficit. Above a BMI of 30 the app uses an adjusted weight — fat tissue needs no protein. On a vegan diet the meat replacements (tofu, pulses, oyster mushrooms) carry about a third less protein, so when losing fat or getting lean it comes out at roughly nine tenths of the target; fibre, on the other hand, is plentiful, more than on a diet with meat.
Carbohydrate and fat
Carbohydrate tops out at 3 to 5 grams per kilo depending on your goal, which is the lower part of the ACSM band for light to moderate training. It is a ceiling, not a fixed number: carbohydrate is the energy left after protein and fat, so a low target leaves less of it and building muscle leaves more. Never under 130 g a day: that is what the brain burns, and it is one of the few figures the Institute of Medicine states as a firm recommendation rather than a range. Fat stays between 20% and 35% of energy — European and American reference values agree on that — and never drops below half a gram per kilo.
How big a portion is
A portion is a share of your daily target: breakfast 22%, lunch 30%, dinner 28%, each snack 10%; if you have a fresh, it takes 8% (at most 500 g and 250 kcal) and the other meals shrink in proportion. So the recipe decides what you eat, not how much — lunch is 30% of the day whether it’s a light salad or hearty pasta. Every portion has a ceiling so it stays a healthy plateful: a main meal at most 800 g of cooked food and 1,100 kcal (30% of the day for a big target), a snack at most 2× its usual portion and 400 kcal (600 on a training day or when building muscle), a soup at most 1,000 g without the broth. For children the limits shrink — to 70% for a schoolchild, 50% for a small child. Dry rice, pasta or oats are weighed as cooked. Whatever doesn’t fit under the ceiling isn’t poured anywhere else, and the app would rather pick a heartier dish — it never adds an extra meal. A skipped meal is spread over the others only up to these limits, and if your target won’t fit into the meals at home, the app says so on that day and tells you whether a snack would help. A recipe doesn’t repeat within a period, nor come back from the previous one — unless your restrictions would leave no other dish.
Children: the main thing is enough to eat
For a child, the app looks after only one thing: that they get enough to eat — it doesn’t count their protein or carbs. Children are still growing and need far more energy per kilo than adults, so the adult rules don’t fit them. The app works out a portion by size (never under 1,000 to 1,800 kcal a day; for teens by sex and activity too), but if they’re hungrier or less hungry that day, leave it to them. One thing doesn’t sort itself out, though: everyone eats the same meal, and the recipes are built for training, so they’re packed with protein — mostly from meat, quark and eggs. Protein then gives a child almost a quarter of their energy, about half as much again as is recommended for children. And that matters: a large European study found that children fed higher-protein formula in their first year were more than twice as likely to be obese at age six. It helps to give a child more of the sides — bread, rice or potatoes — plus vegetables and fruit, and a bit less meat, quark and eggs. That won’t even it out completely, because breakfasts and snacks carry plenty of protein too, but it gets you a lot closer.
Cooking for more than one
Breakfast, lunch and dinner are always cooked together and only the portion size differs — so at one table, one person can be building muscle while another is losing fat, from the same pot. In the shared pot, the meat, tofu or fish is split by how much protein each person needs, and the side makes up the calories — whoever is losing fat takes more meat and less of the side. A snack of your own goes only to someone who eats at least four meals at home that day: someone building muscle gets a protein shake or a carb snack, someone losing fat and even a bigger share of the meat isn’t enough gets a lean snack that part of their portion moves into. At most one person a day gets an extra. If everyone is short of protein, a higher-protein shared snack is chosen instead and nobody gets an extra meal. The extra is always cut down to what’s actually missing, and if the table alone is enough, it isn’t given at all. When several people have the app as Teammates, each rates meals for themselves: a thumbs down from an adult switches the meal off, one from a child only makes it less likely to come up, and a thumbs up brings it back sooner only if nobody has given it a thumbs down or “It was OK”. On the recipe card the app rounds each share of the pot to the nearest ten grams and ten kcal — nobody splits a pot to the gram anyway. Two people with the same share see the same number; each person’s exact calories are on the meal card under Today.
What counts as expensive
Every ingredient has an estimated price from an ordinary supermarket, and the quantities in the recipe add up to the cost of a portion. A dish counts as cheap if it’s in the cheapest 35% of its category — breakfasts are compared with breakfasts, dinners with dinners, because a snack will always come out cheaper than a lunch. With “On a budget” on, the app builds your plan only from those; if there aren’t enough of them for the plan, it tops up with the cheapest of the rest, never the expensive ones. Measured across six types of household: food comes out about a quarter cheaper, and calories, protein and carbs stay on target. The app shows no prices in pounds or euros — they vary from shop to shop and country to country, whereas the order from cheapest up is the same everywhere.
What the app cannot do
The app knows nothing about your health, nor how much body fat you carry — and that is precisely what would sharpen the numbers most. Leave out your height or age and you get an average, which can be more than 200 kcal a day out. What you eat away from home the app can’t see — it simply cooks less, depending on how many of your meals you have elsewhere. And above all: Cheflex is a cooking tool, not a health service.
Show your working
Every app says its numbers add up. This one shows them. The same engine that runs on your phone planned weeks for people living solo, for couples and for families with kids — for someone losing fat and someone building muscle. For every day it added up what each person actually got on their plate and held it against their target. No cherry-picked days, no filter.
- 100% of days on target for energy
- 98% of days with enough carbs
- 96% of days with enough protein
- 100% of days with fat in range
14 households · 6 random weeks each · 588 planned days · 1,713 recipes in the collection · measured 7 October 2026
Extra protein? Fine. Too little? Not fine.
For protein and carbs, the target is a floor, not a ceiling. That’s why this counts the days with at least 90% of the target — getting more isn’t a mistake. Where protein falls short, it’s the same story every time: one meal, one pot, and three people at the table with different goals. The app solves it with the portion: whoever is losing fat takes a bigger share of the meat or tofu from the pot and a smaller share of the side, and their calories stay the same. If even that isn’t enough, a personal extra comes in — but only for someone who eats at least four meals a day at home, and for at most one person a day. Cooking two different meals just for protein is something the app deliberately doesn’t do.
One pot, everyone their own share
A family of four: one parent is losing fat, the other is building muscle, plus two kids. Everyone eats the same thing, just in different portions. The bar shows which meal the day’s energy came from, and on the right is how big a share of the energy each person took from the pot at lunch. The meat or tofu and the side are divided up separately — whoever is losing fat takes more of the first and less of the second, so they leave the pot with the same calories but more protein.
- Breakfast
- Snacks
- Lunch
- Dinner
- Personal extra
Where these numbers come from
- Mifflin M. D. et al. (1990) — resting energy expenditure equation
- FAO/WHO/UNU (2004) — physical activity multipliers
- Jäger R. et al., ISSN position stand (2017) — protein and exercise
- Helms E. R., Aragon A. A., Fitschen P. J. (2014) — contest preparation
- Garthe I. et al. (2011) — rate of fat loss and lean mass
- Iraki J. et al. (2019) — off-season muscle gain
- ACSM, the Academy of Nutrition and Dietetics (AND) and Dietitians of Canada (2016) — nutrition and athletic performance
- EFSA (2010) and the Institute of Medicine — dietary reference values
- ESPGHAN (2017) and the American Academy of Pediatrics (2023) — child nutrition