All Case Studies
    AI Fitness Web App

    Gymmie AI

    Rules decide. AI explains.

    Visit Live Site Health & Fitness / SaaS · 2026
    100%
    Rule-checked plans
    Versioned
    Replayable engine
    No black box
    Auditable AI
    Real-time
    Streamed insights
    // overview

    Gymmie is deterministic fitness intelligence. It builds auditable training plans from your equipment, readiness, recovery signals and training history — keeping every workout decision in versioned, transparent rules, and using AI only to explain the plan in plain language.

    // the brief

    Challenge & Approach

    01

    The Challenge

    Most AI fitness apps are black boxes — they generate plans you can't trust or audit, and can quietly recommend unsafe loads. Gymmie needed the opposite: training decisions that are deterministic, versioned and replayable, with AI contributing clarity rather than unpredictable judgement.

    02

    Our Approach

    We built a server-side, versioned workout-prescription engine where readiness gates and safety checks constrain training before any plan is shown. AI is layered strictly on top as narrative — explaining, never deciding. A cached dashboard snapshot delivers fast first paint with streamed insights, transactional completion keeps data integrity, and a deep internal tooling suite gives the team full observability over the system's decisions.

    // what we built

    Features & Capabilities

    Deterministic Training Engine

    • Server-side, versioned, replayable workout prescription
    • Readiness gates from recovery, fatigue & check-in signals
    • Deterministic safety checks on workload & fatigue limits
    • Adaptive progression — right stimulus at the right time
    • Transactional workout completion for data integrity

    AI Narrative Layer

    • Plain-language, reason-coded explanations of every plan (OpenAI)
    • Streamed insights for fast perceived performance
    • AI strictly explanatory — never the decision-maker

    Member Dashboards

    • Personalised workout & full workout history
    • Nutrition + weekly nutrition planning
    • AI coach & mastery progress views
    • Profile & system status pages
    • Guided onboarding flow

    Analytics & Evidence

    • Training-load / workload history charts
    • Recovery & readiness ('ready score') gauges
    • Muscle-group fatigue heatmap
    • Decision-evidence views backing every recommendation

    Internal & Admin Tooling

    • Internal intelligence, patterns & trajectories dashboards
    • User management & per-user inspection
    • Corrections, variety & system tooling
    • Admin beta & ops consoles

    Platform & Reliability

    • Supabase auth (login, register, password reset)
    • Drizzle ORM on Postgres with managed migrations
    • Inngest background jobs & Upstash Redis caching
    • Sentry monitoring; Stripe billing
    // the work

    Screens & Experience

    Dashboard on mobile
    Nutrition on mobile
    // built with
    Next.jsTypeScriptDrizzle ORMPostgresSupabase AuthOpenAIInngestUpstash RedisSentryStripeRecharts
    // Trusted by
    DSYNAIEvoJoveSafeSenseGrangerJeoumFertil Pharmacy
    // next steps

    Build Something Like This

    Have a product in mind? Let's make it the next case study.