Case 04 · AI integration · Health & fitness
Nimbus
A private city club asked us for a simple app where members log what they eat. We built it, and added the part that makes people keep using it: AI that counts the calories from a photo of the plate or a single sentence, and a training log for every set on the gym floor.
- Client
- Nimbus, a private fitness and spa club
- Market
- Club members, premium segment
- What we did
- Product strategy, UX/UI, iOS and Android app, AI integration, coach console
- Year
- 2026

3sec
from a photo of the plate to a logged meal
2ways
to log a meal: a photo or one sentence
1tap
to log a set, with last week's weight shown
2apps
for members and for their coaches
Product parameters. Calories are AI estimates from typical nutrition data, for guidance, not medical advice.
The story
It started as
The club came to us with a clear request: an app where members can log what they eat and see their daily calories. Its members are busy people with personal coaches, and the coaches wanted to see more than what they hear at the next session.
While we were mapping the product, one problem stood out. Calorie counters rarely fail on the maths. They fail on the effort: weighing food, searching a database for every ingredient, typing it all in. After a couple of weeks, most people simply stop logging.
So we proposed to take the counting out of it. A member takes a photo of the plate, or types one sentence, and an AI model connected through an API recognises the dishes and estimates the portions. The calories themselves come from typical values in public nutrition databases, so the numbers are grounded rather than invented.
We also suggested a second idea: log the training too. Members record every set on the equipment, see what they lifted last time and what comes next, and coaches see meals and progress in one console. We designed and built the member app for iOS and Android, the AI service and the coach console.

The idea we added
The brief asked for a calorie counter. We suggested taking the counting out of it.
Snap the plate
One photo and the AI lists the dishes, estimates each portion in grams and adds up calories, protein, carbs and fat.
Or just say it
“Two eggs, avocado toast and a flat white” works too. Typed or dictated, it becomes the same itemised meal.
Every set counts
Sets, reps and weight are logged in one tap on the gym floor, with last week's numbers and the next exercise in view.
Coaches see it all
Meals and workouts reach the coach as they happen, with short AI summaries of what changed this week.
For members
From the plate to the last set in
Nothing to weigh and nothing to search. The member sees the day at a glance and logs a meal or a set in seconds.

01Today
Calories left for the day, protein, carbs and fat, and every meal with how it was logged: by photo or by text.

02Snap
Point the camera at the plate. The AI marks each dish it recognises with its estimated portion and calories.

03Confirm
Check the breakdown, adjust any portion in grams, or tell the AI what the photo can't see. Then add it to the day.

04Train
Follow the coach's program set by set: weight, reps, rest and what was lifted last time, logged in one tap.



Three ways in, one record: a photo of the plate, a sentence about it, and every set on the gym floor.
For coaches
Every member's week
The coach console shows how each member eats and trains between sessions, so the next session starts from facts, not memory.

Members at a glance
Who is on target today, who trained when, and who has gone quiet this week.
Meals as they happen
Every logged meal with its photo or text, calories and macros, in the member's timeline.
Strength progress
Best sets by exercise over weeks, so progress is a chart, not a feeling.
AI summaries
A short note per member on what changed: protein short at breakfast, a new record on deadlift.
Programs and tips
Coaches build workout programs and send a tip that lands in the member's app.
How we built it
AI where it helps,
- Mobile app
iOS and Android, one codebase
The member app with the day's calories, the camera, the meal log, programs and the set tracker, synced with Apple Health and Health Connect.
- AI service
A model behind an API
A vision-capable AI model reached through an API returns structured results: each dish, its estimated grams and how sure it is. A test set of real plates keeps the estimates in check.
- Nutrition data
Numbers from databases
The model names the food and the portion; calories and macros are calculated from per-100 g values in public nutrition databases, not made up by the model.
- Training log
Works on the gym floor
Programs from coaches, sets, reps, weight and rest timers, logged offline where the signal drops and synced when the phone is back online.
- Coach console
A web app for the team
Members, meals, workouts, progress charts and AI summaries, with tips and programs sent straight to the member's phone.
- Privacy
Health data handled with care
Photos are used for the analysis and not kept, data is hosted in the EU, and every number is labelled as an estimate, not medical advice.
Have an idea that could
Bring us the brief. We'll build it, and show you where it can grow.