A Flutter home-training app that generates personalized weekly plans, tips, and reports via the Gemini API, with human-trainer booking on Cloudflare Workers, R2, and D1.
Online Fitness Coach is a mobile app for home fitness training that uses generative AI through the Gemini API to generate personalized training plans from the user's data (age, weight, height, goals, activity level), deliver daily tips and periodic progress reports, and answer questions in natural language. It was carried out by a team of students with technical assistance from Techno Enjaz, and it brings together in one platform an AI coach and human trainers whom users can book sessions with and message.
The app's interface is built with Flutter and Dart, with a back end on Cloudflare Workers, Cloudflare R2 storage for reference exercise videos and files, and a relational Cloudflare D1 database. Personalization relies on text and numeric data only: the app does not analyze the user's photos or videos and needs no extra sensors, and the videos it shows are for reference and instruction only.
| Item | Details |
|---|---|
| Project type | Mobile fitness-training app powered by generative AI |
| Field | Generative AI, natural language processing, digital health and home training |
| Project status | Implemented app with documented screens, without a field trial with users |
| Techno Enjaz role | Technical support and assistance to the students during project development |
| Core technologies | Flutter, Dart, Gemini API, Cloudflare Workers, Cloudflare R2, Cloudflare D1 (SQLite) |
| Outputs | Generated weekly training plans, daily tips, monthly reports with charts, trainer session booking, inquiry system |
Many people are turning to home training and digital apps, but the lack of direct supervision raises the risk of injuries from incorrect technique and reduces the effectiveness of workouts because there is no feedback. Most traditional fitness apps also rely on static video content that does not adapt to the user, while personal trainers and gyms are expensive.
On the other hand, commercial apps that analyze movement with computer vision (such as Onyx, Kemtai, and Tempo) depend on high-end phones, LiDAR depth sensors, or costly subscriptions. The project therefore chose a lighter approach: interactive, text-based fitness consultation that runs on mid-range Android phones and addresses three gaps the project identified:
The project was carried out academically by the students, with technical assistance and support from Techno Enjaz during development. It is presented among the office's work as an example of its support for mobile apps that combine generative AI models with edge cloud infrastructure.
| Role | Tasks |
|---|---|
| Regular user | Create an account, enter personal and health data, request an AI plan, log activities and meals, book sessions, view reports, send inquiries |
| Personal trainer | Manage availability, review booking requests and accept or reject them, follow up on inquiries, give feedback and guidance |
| AI coach (Gemini) | Generate training plans automatically, give tips, analyze daily activity data, generate reports and summaries |
| Administrator | Manage trainer data, monitor users, oversee content and services, and monitor system operations |
The system uses a multi-tier architecture with four components:
The entity-relationship (ER) diagram links the users table to workouts, meals, and monthly reports, trainers to appointments, inquiries to their messages, and training plans to their days.
| Screen | Function |
|---|---|
| Home screen | App introduction, gender selection, sign-up or login, with admin and trainer entry points |
| Personal data entry | Collects age, height, and weight to build plans |
| AI coach / trainer selection | Shows the AI coach and the list of human trainers with their specialties |
| User dashboard | Summary of current status, the daily plan, and progress level |
| Weekly training program | Completion percentage and the daily exercise schedule with each exercise's name, duration, and status |
| Daily tips | Logging meals and workouts and showing tips generated from activity and health data |
| Monthly reports | Charts of weight change and the number of meals and activities logged during the month |
| Inquiries | Lists the user's inquiries and creates a new one with a choice of trainer |
The project documents the implementation of the app and its core screens, its integration with the Gemini API to generate plans, tips, and reports in JSON, and a platform with four roles covering planning, follow-up, and evaluation. Challenges it addressed include separating the reference videos stored on R2 from the AI-generated written tips, and limiting generic or inaccurate answers through prompt engineering and post-processing of responses before display.
The project includes no quantitative measurements of its own, such as actual response time, plan quality, or a trial with trainees. The Cloudflare figures it cites (such as edge latency and storage pricing) are general platform specifications, not results measured in this app.
According to the project's recommendations, the app could be extended through:
The app processes personal and health data such as weight, meals, and activities, so any real deployment needs clear data-protection policies, user consent for sending data to cloud AI services, and compliance with local and international regulations.
If you are working on an AI-powered fitness app or smart coach and need technical help developing the idea or building a prototype, you can contact the Techno Enjaz office to discuss your project's requirements and the appropriate scope of support.
It is a mobile app for home fitness training built by a team of students with technical assistance from Techno Enjaz, using the Gemini API to generate personalized training plans, tips, and reports. It combines an AI coach with human trainers whom users can book sessions with and send inquiries to.
The user enters data such as age, height, weight, goals, and activity level, and the Cloudflare Workers back end sends it to the Gemini API with engineered prompts that require answers grounded in fitness references. The response comes back as JSON and is shown as a weekly program and daily tips that adapt to the meals and workouts the user logs.
The interface is built with Flutter and Dart with Arabic and RTL support, the back end runs on Cloudflare Workers, the database is SQLite-based Cloudflare D1, and Cloudflare R2 stores reference exercise videos. Google's Gemini API handles the intelligent functions such as generating plans, tips, and reports.
No. Personalization relies only on text and numeric data, the app does not process the user's photos or videos, and the videos it shows are for reference. It offers written guidance such as rep ranges, rest times, and the importance of warming up, while movement analysis is outside the scope of the current version.
The project documents the implemented app and its screens (weekly program, daily tips, monthly reports, bookings, inquiries) and its Gemini API integration, but includes no quantitative performance measurements or trial with trainees. Its limits include dependence on the internet and cloud services, a focus on Android and core bodyweight exercises, and no scientific evaluation of plan quality against a human trainer.
Reach out to Techno Enjaz engineering team to discuss technical feasibility, prototype hardware, and development scope.
Discover more engineering prototypes and academic implementations
A Flutter app and Laravel admin dashboard on MySQL combining Syrian destinations, a city guide, hotel booking, a handicraft store, articles, and visitor experiences.
A Flutter university guide app with a Laravel admin dashboard covering majors, study plans, courses and search, plus a virtual assistant matching FAQs via Google Gemini.
A kindergarten management system with a Flutter parent app and a Laravel web dashboard for administration, covering attendance, health, meals, events, and messaging.
