FitGenie: an all-in-one fitness platform

Users currently stitch together four apps to manage fitness: MyFitnessPal for food, a workout app for training, YouTube for exercise form, and a diet planner or a nutritionist for meals. Each swap loses context. Your calorie budget does not know your leg day just wrecked you, and your training plan does not know you are 400 calories under target. The wedge is one shared data model across eating and training, not any single feature on its own.

That bundle is also the trap. FitForge, Myfit AI, MyFitnessCoach and others have already built the same four-panel bundle, so having all four features is not a moat by itself, it is table stakes to enter the category. The differentiator has to be sharper personalization: the AI coach and planner actually adapting to logged behavior week over week, instead of generating a plan once and leaving it static, plus a content layer of form videos that is genuinely better than a stock exercise API.

Estimates vary by research firm and scope, but they agree on direction: a large market still compounding at double digits.

The incumbent, MyFitnessPal, doubled its monthly price from $9.99 to $19.99 after its 2020 acquisition by Francisco Partners. That is a pattern common after private equity ownership: extract more from an entrenched user base rather than out-innovate. It leaves a pricing and product gap directly below the incumbent for a bundle that is cheaper and does more, which is exactly the seam FitForge and Myfit AI are already working. Expect a crowded, well-funded field where differentiation and retention execution matter more than market size alone.

Four bundled competitors, one unbundled incumbent. All four bundles cover roughly the same feature set. The differences show up in emphasis and execution quality.

None of these five has won decisively yet. This reads as several strong products splitting a growing pie, not one dominant winner. That is good news for entry and bad news for anyone hoping category creation alone will carry them.

Four features, sequenced by how much they depend on each other and on user data that does not exist yet.

Build the tracker and the content library first. They are commodity-infrastructure problems solvable with existing APIs (Section 07), and they generate the logged data the two AI features need to personalize instead of guessing on day one. This core scope stays first in the build order regardless of geography. Section 05 below is a scoped addition for a specific segment, not a change to that sequence.

Every calorie tracker on the market, including the four competitors in Section 03, is built around a hidden assumption: food comes with a label or a menu. That assumption breaks for a huge and underserved segment, and it is worth naming without turning the whole roadmap around for it.

Most Indian households, especially urban middle-class ones, eat home-cooked food most meals. That food is cooked in bulk for the whole family, not portioned per person, and it is almost never weighed. Asking that user to weigh a katori of dal on a kitchen scale before logging it is asking them to change how they live, not just adopt an app. A meaningful share of that segment will try logging once, find it tedious, and quietly churn. This is not a small edge case. It is the default eating pattern for hundreds of millions of potential users, and no direct competitor in Section 03 has built for it.

What breaks today

Features

For a user with no saved family recipe yet, fall back to a regional dish database seeded from India's National Institute of Nutrition food composition tables, covering common dishes such as dal, sabzi, poha, idli, paratha, and sambar, so the app is useful from the first log rather than waiting for setup.

Rough screens for the MVP surfaces plus the home-cooked-food logging flow, with the backend call each screen depends on called out underneath. These are wireframes for scoping, not final visual design.

Buy the commodity data, build the reasoning layer. Nobody wins by re-cataloguing 14 million foods from scratch.

Data and content sourcing

AI layer

• Coach and diet planner reasoning: a Claude or GPT-class model via API, prompted with structured user state (goals, logged history, constraints) rather than free-form chat, which keeps output consistent and auditable.
• Food photo estimation: a vision model for a first-pass estimate, always paired with an easy manual-correct UI. Accuracy for mixed dishes runs roughly 70-85% industry-wide, so design for correction, not perfection.
• Voice and free-text meal parsing: speech-to-text feeding an LLM parser that checks saved household presets first, then the regional dish fallback, then a generic estimate, in that order of trust.
• Personalization loop: store structured logs (sets, reps, RPE, meals, adherence) and feed a rolling window back into plan-generation prompts. This loop is the actual product moat, not the underlying model choice.

Stack

• Mobile: React Native or Flutter, cross-platform given a lean team, native performance is not yet a bottleneck at MVP scale.
• Backend: managed Postgres plus a lightweight API layer (Node/NestJS or Python/FastAPI). Avoid microservices until there is a scaling reason that justifies the overhead.
• Video delivery: CDN-backed adaptive streaming (Mux or Cloudflare Stream) rather than self-hosting raw files.
• Compliance: treat all logged health data as sensitive by default. If any biometric or health-record integration ships (Apple Health, wearables), review HIPAA-adjacent handling even though a consumer wellness app is not itself a covered entity.

A lean team (2 mobile, 1 backend, 1 designer, fractional PM) building on the sourcing strategy above, not from scratch.

Roughly 5 to 6 months to a beta-worthy MVP, all-in around $180-275K including the commissioned video content, before the optional home-cooked-food add-on. That add-on is worth pulling forward into Phase 1 rather than running after Phase 3 if the initial launch market is India-first, since it changes the core logging flow rather than sitting on top of it. If the launch market is US or Europe-first, it can wait and slot in as a later regional expansion.

Freemium, priced just under the incumbent's post-price-hike tier, undercutting on price while over-delivering on feature bundling.

Undercut MyFitnessPal's $79.99-99.99/yr Premium and Premium+ tiers while bundling training and diet. The pitch to a price-sensitive switcher is simple: pay less, and replace two more apps. In an India-first launch, expect the whole price ladder to be rebuilt around local willingness to pay rather than a straight currency conversion of the figures above.

• Content and SEO first, paid second: exercise-form and calorie-count queries are high-intent, high-volume search terms. A content arm built on the same exercise and food data can capture organic traffic before any ad spend.
• Micro-influencer seeding early, not just as future work: free Pro accounts to 20-30 mid-tier fitness creators (10K-200K followers) in exchange for authentic use, ahead of the formal creator marketplace phase. Cheaper trust-building than the eventual paid platform.
• Referral loop: shared workout plans and progress cards are natural, low-cost virality mechanics for a fitness app. Design for screenshot-shareability from day one.
• Retention is the real growth lever: category-wide 30-day retention for fitness apps is notoriously weak, commonly cited in the single-digit percentages. Winning here depends less on acquisition spend and more on the coach and planner actually adapting. Churned users are the biggest cost in this market.

Sequenced deliberately. Each stage needs the retained audience the previous stage built, not the other way around.

1. Creator marketplace: once DAU and retention are provable, invite the influencers already seeded in Section 10 to sell structured programs through the app. FitGenie takes a platform cut, creators get distribution and the personalization data layer for free.
2. Commerce, supplements and equipment: a logged-nutrition user base is a targetable audience for relevant, non-intrusive product recommendations. Sequence after trust is established, not before, to avoid reading as a bait and switch.
3. Yoga vertical: the lowest lift of the three. It largely reuses the existing video-content pipeline and AI-coach architecture with a different content library and program templates.