Flutter app
Cross-platform iOS and Android app built with GetX: guided sessions, coaching conversations, progress tracking, and multi-device session management.
- AI Product Engineering
- Forward-Deployed Services
Kaya Functional Health
Kaya is an AI functional medicine health coach. It guides people through structured wellness sessions — intro, foundation, follow-up — with real-time coaching, habit, goal and mood tracking, and personalised journeys that don't need a human coach for every interaction.
Live — App Store & Google PlayView on Google Play →
Mobile app interface
Snapshot
Production status & delivery details.
Verified production status, stack, and maintenance model.
- Client / Context
- Kaya — AI functional medicine health coach for iOS and Android, published by GenAILabs
- Engagement
- End-to-end build: Flutter app, backend API, AI coaching engine, Firebase auth and payments — then a fixed price plus an ongoing retainer for maintenance
- Status
- Live on the App Store and Google Play (production)
- Technology
- Flutter, GetX, Firebase Auth, Firebase FCM · Python FastAPI, PostgreSQL · OpenAI / LLM streaming over SSE · Stripe · Google Sign-In, Apple Sign-In · Flutter Secure Storage, Codemagic CI/CD
The challenge
What high-stakes problem had to be solved?
The business context, constraints, and risk surface that shaped every decision.
The client needed a personalised AI health coaching app that guides people through structured wellness sessions, tracks habits, goals and mood, and answers in real time — without requiring a human coach for every interaction.
Kaya — AI functional medicine health coach — had to work as a cross-platform product: coaching conversations, guided sessions, progress tracking and personalised health journeys, with sessions that follow the user across devices. It also had to ship through both app stores, carry the medical disclaimer at onboarding, and stay aligned with GDPR while coaching guidance remained clearly separate from medical advice.
Contributions
What GenAILabs worked on.
Factual breakdown of our direct engineering contributions.
AI coaching engine
Real-time coach responses streamed from OpenAI/LLM over SSE — under one second to the first token — with personalised health journeys.
Structured health journeys
Intro → foundation → follow-up sessions, a habit and goal system, mood tracking, and weekly analytics.
Backend API
Python FastAPI service on PostgreSQL for sessions, habits, goals, mood, progress, and uploads of up to 10 images per session.
Auth, payments & delivery
Firebase Auth with Google and Apple sign-in, FCM notifications, Stripe payments, Flutter Secure Storage, and Codemagic CI/CD.
System / solution
Technical design of the custom system.
Flutter client, FastAPI backend, LLM streaming, Firebase auth, Stripe payments, and Codemagic delivery to both stores.
System architecture
Flutter app with GetX on iOS and Android → Python FastAPI on PostgreSQL for sessions, habits, goals, mood and uploads → OpenAI/LLM responses streamed over SSE. Firebase Auth handles Google and Apple sign-in; Firebase FCM delivers notifications; Stripe handles payments; Flutter Secure Storage holds credentials; Codemagic CI/CD ships builds to the App Store and Google Play. Firebase Analytics, an internal dashboard, and App Store Connect cover monitoring.
Flywheel impulse
Deployment, integration, and real-world use.
Verified real-world deployment, maintenance, and scale protocols.
Kaya is live in production on the App Store and Google Play. The engagement was a fixed price for the build, followed by an ongoing retainer for maintenance: bug fixes, model updates, and new features each sprint. Public store links are not available yet.
- Integration pattern
- Firebase Auth (Google and Apple sign-in) and FCM, Stripe payments, OpenAI/LLM streaming over SSE, Codemagic CI/CD shipping to the App Store and Google Play
- Scale protocol
- Multi-device session management with conflict resolution, biometric login with refresh token rotation, and up to 10 images per session
- Maintenance mode
- Fixed price plus an ongoing retainer: bug fixes, model updates, and new features each sprint, watched through an internal dashboard, Firebase Analytics, and App Store Connect
Outcomes / evidence
Verified results from the shipped product.
Every result listed is verified and sourced from the Kaya case data.
Verified evidence
Under 1s to first token
Real-time coach responses streamed over SSE from the LLM
Verified evidence
Biometric login, rotating refresh tokens
Multi-device sessions with conflict resolution and secure storage
Key outcome
Live on the App Store & Google Play
Production release built end to end by GenAILabs, maintained on a fixed price plus retainer
Capabilities involved
How core competencies were applied.
AI Product Engineering and Forward-Deployed Services were both demonstrated in production.
AI Product Engineering
The whole product: Flutter app, FastAPI backend, AI coaching engine with sub-second SSE streaming, Firebase auth, Stripe payments, and structured intro → foundation → follow-up journeys.
Forward-Deployed Services
Live on both app stores through Codemagic CI/CD, with multi-device session handling and a maintenance retainer covering bug fixes, model updates, and new features each sprint.
Related work
Other production builds.
Two more engagements from the same team.
AI Product EngineeringPilot
Mingcheji
Global car-history and parts platform for buyers outside China — six languages, USD pricing, piloted on a preview site.
Agentic Layer EngineeringProduction
NexAgents
Production AI agent platform for teams — per-person model access, agent builder, data stores with spending caps, and metered token costs.
Let's talk
Have a similar challenge?
Let's discuss the problem you're trying to solve.
Talk to our team
