Leanly
A calm calorie tracker I designed and built with AI: quick to log and honest about where you stand.
Browse the prototype- Role
- Sole designer & developer, AI-assisted
- Industry
- Health & Fitness
- Scope
- Product Definition, UX Research, UI Design, Branding, AI-Directed Build
- Tools
- Claude Code, Figma, Illustrator
- Year
- 2026
Overview
I wanted to get fitter and track what I ate properly, so I built an app for it.
A few years ago that wouldn’t have been realistic. With AI, a designer can take an idea all the way to a working product. I built exactly the app I wanted: calm, quick to log and honest about where I stand.
The real work was directing: knowing what to ask for, what to send away for research, which references to share and when to send work back.
I was the sole designer and developer, with AI doing the building.
The Challenge
I knew the app I wanted. I didn’t know how to get an AI to build exactly that.
An AI will confidently build whatever you describe, including the wrong thing. The challenge was describing what I wanted precisely, and spotting quickly when the result missed.
Research & Discovery
For the look, I skipped verbal descriptions and handed over concrete design systems and visual references.
For everything else, AI did the digging: the calorie maths and how nine other trackers handle it. I read it all before deciding anything, then kept some findings and discarded others.
Strategy
- 01
Start loose, then converge
I opened with a rough idea and refined it in conversation: what it’s for, who it’s for and which features earn their place. Unexpected questions surfaced early, and the vague version became specific before any design began.
- 02
References, not adjectives
“Clean and minimal” gets you someone else’s idea of clean. An actual reference, with the important part named, gets you yours.
- 03
Research before drawing
Nothing was designed until the science and the competitor patterns were on the table and I had reviewed them.
- 04
Approve a spec, then review every pass
Every screen and rule went into a document I signed off first. I then checked each round and sent it back when it drifted.
The Solution
- 01
One question per screen
Home answers only “what’s left today”, with a week of context beside it, so a heavy Tuesday reads as recoverable, not a failure.
- 02
Numbers that stay calm
Figures in monospace, like a lab label. Over budget turns apricot, never red, and there are no streaks to break. This was the decision I held hardest: as the user, I needed an app I’d still open on a bad week.
- 03
One interaction worth getting right
Describe a meal in plain language and the ingredients tally up like a receipt printing, landing on the total.
- 04
One system, everywhere
Forest-green ink on warm snow white, depth from tinted panels instead of shadows, and green kept to touches. Drawn from the systems I chose and applied to every screen.
The Result
Nineteen screens across five tabs: onboarding and a personal plan, a daily home view, three ways to add a meal, a food library and a progress tab.
Setup takes six screens and about a minute. Questions are grouped, and those that only refine the estimate are marked optional. The daily loop is short: open the app, see what’s left, add what you ate by search, photo or a typed sentence.
Nothing to dismiss, nothing to earn back, and what you came to check sits at the top of the first screen.
What I Took From It
How I asked decided what I got back.
Most of my time went into defining exactly what I wanted before asking. Describing every screen, empty state and edge case clearly enough for something else to build gets those decisions made early, instead of late.
AI covered a lot of ground quickly: the research, the calculations, the code. It couldn’t decide how the app should feel, or catch what was technically right but still wrong. That part I kept doing myself.