Geode Fit
GeodeFit is a fitness app built for people who actually track. Your training sits on one side, your nutrition on the other, both living in a single dark interface that stays out of the way. No clutter. No noise. Just your numbers, your progress, and an honest read on what's working and what isn't.
Overview
GeodeFit is a personal fitness app that keeps training and nutrition under one roof. One account, one dark interface, both sides of the work in the same place. It sets calorie and protein targets around your own body and goals, builds a diet around foods you actually eat, and turns daily logging into a streak and a trend you can read at a glance. I built it solo, from the first wireframe to production, around a single question: why do people quit fitness apps, and what would keep them coming back?
The Challenge
Chasing a fitness goal usually means juggling a drawer of half-used apps. One for workouts, one for calories, one for weigh-ins, none of them talking to each other. So the single goal gets split across tools that each own a fragment, and the plan never feels whole.
The numbers make it worse. Popular calorie trackers run on Western food databases, so anyone eating paneer, dal, or rajma spends more time searching than eating, and logging quietly dies. And when the targets are generic and the macros land as a wall of raw numbers with no context, people stop trusting them.
Put those together and you get the pattern every fitness app knows and few solve. People sign up with real intent, log for about a week, feel overwhelmed, and ghost it.
The brief was simple to say and hard to do: one focused app that personalizes the plan, fits how people actually eat, and makes progress obvious enough to pull someone back tomorrow.
Research & Insight
Fitness apps have a brutal, well-known failure mode. People subscribe with real intent, then ghost the app within weeks. I started GeodeFit from that problem, not from a feature list, drawing on my own training experience and a close read of the apps people already abandon.
The reframe I began with: the leak isn't usability, it's upstream and psychological. An app like this doesn't fail because it's hard to tap. It fails when it stops keeping the promise that got someone to pay, and never builds a habit before the signup-day motivation fades.
A few patterns kept surfacing.
Generic targets erode trust. When the calorie number ignores your height, weight, age, and goal, it feels arbitrary, and people stop believing it.
Food libraries decide adherence. If someone can't find what's on their plate, they stop logging. For an audience eating Indian meals, that gap is a dealbreaker.
Training and diet aren't two problems. They're one goal split across two habits, and treating them separately is why progress stalls.
Motivation lives in visible momentum. Streaks, weekly adherence, and a clear trend do more for consistency than any feature list.
The throughline: friction and vagueness are what kill fitness apps, not a lack of features.
To pressure-test that, I treat retention as the real question. "Return" means a completed workout, not a hollow login, so I instrument the signup, first-workout, return funnel and look for the cliff, separating people who never finished a first workout from those who bounced at setup, since those are two different problems. Then I take it to the people who went quiet with JTBD interviews and a short diary study, and contrast them with the users who stuck.
I stay honest about the limits. It's my own product with a small, friendly user base, so I'm the most biased observer in the room and the sample isn't representative. I flag the riskiest assumptions as assumptions and would validate them with a wider recruit before betting the roadmap.
Where it points: the fixes live upstream and in the first week, onboarding that builds a habit, a trigger that earns the return, and personalization that makes coming back feel worth it. Then test whether return actually moves. Behaviour over opinion.
Product Strategy
Every bet here answers something the research surfaced.
Two habits, one product. Training and diet aren't two problems, they're one goal split across two habits, so workouts and nutrition share the same account and the same daily rhythm, and the goal stays whole.
Personalization from minute one. Generic targets erode trust, so a short onboarding captures the essentials and GeodeFit computes calorie and protein targets that are actually the user's own.
Diet that fits the plate. Food libraries decide adherence, so preferences split into vegetarian, non-vegetarian, and eggetarian, with a food set built around what people in this market genuinely cook and eat.
Goal as the compass. Lose weight, stay fit, or build muscle. That single choice shapes intensity and calories downstream, which is what keeps the personalization believable.
Progress as the hook. Motivation lives in visible momentum, so streaks, weekly workout adherence, and seven-day charts keep the win in view.
Web first, native next. Since friction and speed decide whether a product ever gets tested, shipping on the web meant fast iteration and a real, usable product early, with a native App Store and Play Store release planned through Capacitor.
Experience Architecture
The structure exists to do two things the research demanded: strip out friction, and build the habit in the first week.
The way in. Friction is what kills fitness apps, so the entry is deliberately smooth. A welcome and intro sequence hands off to authentication, then a four-step onboarding: personal info, body metrics, dietary preference, and goal plus schedule. Each step validates on its own, and a summary card closes the flow so nothing feels blind. Guest, user, and admin routes are guarded so people only see what's theirs.
The daily loop. Because the real goal is one habit loop, not scattered features, everything hangs off a dashboard hub with four surfaces reachable through bottom navigation:
Workout: the weekly split and the exercises to complete.
Diet: meals shaped by preference and favorite foods, with calories and protein tracked as you go.
Progress: BMI and category, current streak, workouts per week, average calories and protein, and seven-day charts.
Profile: the personal details and targets that drive the rest.
Resilience by design. Momentum is the motivation, so the streak has to be protected. Tracking data persists locally first and syncs to the backend when it can, so a shaky connection never costs someone their history or their streak.
Interface System
The visual system carries the same insight into pixels: cut vagueness and friction, and make the domain and the numbers readable at a glance.
Foundation. A near-black, dark-only, mobile-first canvas with a token-driven palette, wired through CSS custom properties so color, surface, and border stay consistent everywhere.
Two-color logic. Since training and nutrition are one goal across two habits, a yellow accent carries training and a green accent carries nutrition. Color does the wayfinding, so users read the domain before they read the label.
Components. Cards, chip selects, stat tiles, and progress dots form a small, repeatable kit. Lightweight inline SVG charts put progress in view without pulling in a heavy library.
Motion with intent. Subtle slide and fade transitions, tactile press states, and animated bars give feedback without turning into decoration, so every action feels confirmed rather than vague.
Built for the thumb. Generous targets, single-column focus, and a calm layout keep attention on the numbers that matter, the opposite of the clutter people abandon.
Outcome
GeodeFit answers the exact failure the research named. Fitness apps lose people to friction, vagueness, and a goal split in two, so I built the whole arc as one coherent product, taken end to end by one person across product design and engineering, and shipped in production on Vercel.
It delivers what a fitness app promises and usually pulls apart: personalized onboarding, goal-based calorie and protein targets, a diet library that fits the user's plate, workout tracking, and a progress view that reads at a glance. The design system is deliberate enough to scale, and the resilient data layer protects the one thing users care about most, their streak.
The next chapter is native. The same codebase is being packaged through Capacitor for the App Store and Play Store, carrying the web product into people's pockets without a rebuild.
