How to keep a food diary in Telegram: step by step
A walkthrough from the first message: starting the bot, getting your calorie target, logging food by photo, text or voice, correcting portions, tracking water and weight, and six habits that stop the diary dying in week two. Works in MAX too.
By the Fat Track team · Updated August 22, 2026
Short answer. A food diary in Telegram is just a chat with a bot: you send a photo, a line of text or a voice note, and the bot works out the calories and macros and files the entry into your statistics. To start, open the bot, go through the short setup that calculates your target, then log your first meal with a photo and correct the portion in a reply. After that it is water, weight, an evening summary and a weekly look at the charts in the app. Two or three minutes a day, which is why this kind of diary survives past week one.
Why a messenger beats another app
Food diaries get abandoned because of friction: open the app, search the database, pick a portion, confirm. In a messenger the path is shorter. Telegram is already open, and a meal goes in as one message. Nothing new to install, nothing to update, no password to remember. Everything below uses the Fat Track bot as the example; in MAX it works exactly the same way.
Step 1. Open the bot and run through the setup
Find the bot in Telegram (@fat_track_bot) or MAX and press Start. It asks a handful of questions — sex, age, height, weight, activity level and goal (lose, hold, gain) — and turns them into a daily target for calories, protein, fat and carbs. You can get the same number on the site in the calorie and macro calculator and adjust it by hand if you want to.
One tip: do not flatter your activity level. Two sessions a week on top of a desk job is "light", not "moderate".
Step 2. Log your first meal
Three ways in, whichever suits the moment:
- Photo. Shoot the plate from above or at an angle, in decent light. A few seconds later you get what was recognised, plus calories, protein, fat and carbs.
- Text. Write it the way you would say it: "porridge with milk 250 g, banana, coffee with milk". The more specific you are about grams and cooking method, the closer the estimate.
- Voice. Say the same thing out loud — handy when you are driving or walking.
How the model reads a plate and where it slips is covered in how accurate AI calorie counting really is.
Step 3. Correct it when it is off
You have an estimate — now look at it critically. No oil visible? Portion bigger or smaller than it looks? Two pancakes rather than three? Reply to the message with the correction: "portion 300 g", "cooked without oil", "double portion", "no dressing". The entry is recalculated. After a week you will know which of your own meals always need a nudge.
Step 4. Log water and weight
Water goes in with a couple of taps, and you can see the day's progress against your target instead of keeping it in your head. Weigh yourself in the morning, after the bathroom, three or four times a week: swings of a kilogram either way are normal, so read the weekly average, not single days. More on that on the water tracker page.
Step 5. Read the end-of-day summary
The bot closes the day with the totals: what you ate, what is left against your calorie and macro targets, where you went over or came up short. A short AI note comes with it — that protein fell 30 g short, say, and tomorrow could use some cottage cheese at breakfast. That is the moment a diary stops being bookkeeping and becomes feedback.
Step 6. Once a week, look at the whole picture
In the mini app inside Telegram, or in the web account, you get weekly and monthly charts, meal history, macro averages and your weight trend. Five minutes a week: are average calories landing on plan? Is weight moving? If not, adjust the target or the portions instead of quitting. The diary exports too, if a doctor or a coach wants to see it. What else is in there is on the app page.
Six habits that keep it going
- Log before you eat. The food is still on the plate, so it is easier to judge — and "later" usually never comes.
- Do not skip the bad days. "Pizza, 3 slices, beer" is a perfectly good entry. Gaps distort your statistics far more than a big day does.
- Correct in one phrase, not perfectly. Within 10–15% is accurate enough to get results.
- Turn on reminders. The bot has a meal schedule and prompts; let them do the remembering.
- Cook the same things often. Five to seven default breakfasts and dinners and half your entries take a second each.
- Judge the week, not the day. One day over target decides nothing. The weekly average decides everything.
What you must log, and what you can skip
Always log oil and sauces, drinks with calories (lattes, juice, alcohol), the "just a bite" snacks, nuts and dried fruit — those are exactly what turn a deficit on paper into no deficit at all. You can skip water, plain tea and coffee, herbs, spices and undressed salad leaves: their contribution is rounding error.
Common questions
Do I have to count every gram? No. Logging consistently with a 10–20% error beats logging perfectly for one week and stopping.
Does the bot work on a computer? Yes, in Telegram Desktop and the web version; the mini app opens in a browser too.
What if I use MAX rather than Telegram? The Fat Track bot is in MAX with the same features. Pick whichever messenger you already live in.
What does it cost? Seven days of full access to start, then one subscription that covers the bot, the app and the analytics — the terms are on the pricing page. The food database and the calculators on the site are free.
Open the bot, send a photo of your next meal and see how long it actually takes. Usually the question of whether to keep a diary answers itself after the first entry.
More from the blog
Read next
How many calories to lose weight: the formula and a worked example
The Mifflin-St Jeor equation, activity factors, the size of the deficit and a full worked example in numbers. Plus how not to set the target too low, and where to start tracking.
Daily protein intake: how much you need and how to hit it
How much protein you need while losing weight, building muscle or just living, how to spread it across the day, and which ordinary foods get you there without overeating. With a source table.
How accurate is AI calorie counting from a photo?
What happens between the snapshot of a plate and the calorie figure: how the model names the dish, estimates the portion and where it goes wrong. An honest look at accuracy, and when a photo is enough versus when you need text or a scale.