Joseph SfeirSoftware engineerContact
Menu
← Selected work02 / 2026

Evolving mobile product

FitAI

One place for training, nutrition, progress, and useful context.

Status
In progress · Independent project
Type
Flutter mobile product
My role
Product design + mobile engineering

Current Flutter build

FitAI workout library with three complete suggested programs and actions to start or view progress
FitAI daily calorie target, health profile estimates, and breakfast food log
FitAI body progress view with weight, BMI, check-in action, and an honest early trend state

01 / Product idea

Useful guidance starts with connected context.

A Flutter fitness product in progress, designed to connect workout planning, food tracking, and body progress, with a planned advisory AI coach.

The problem

Workout history, food entries, body measurements, and progress views are often spread across unrelated tools, leaving each one without the context of the others.

The direction

FitAI brings those signals into one user-owned system. The goal is not to automate every decision, but to make progress easier to understand and guidance more relevant.

02 / Product surface

Three screens, one continuous fitness loop.

I’m designing and building the product end to end, from the Flutter interface and workout flows to user-owned data and the planned server-side AI boundary.

FitAI workout library with three complete suggested programs and actions to start or view progress
01 / PlanChoose a workout structure

Suggested programs turn a broad fitness goal into a session someone can actually start.

FitAI daily calorie target, health profile estimates, and breakfast food log
02 / TrackLog daily nutrition

Profile-based estimates, food entries, calories, and macros live in one daily view.

FitAI body progress view with weight, BMI, check-in action, and an honest early trend state
03 / LearnTurn check-ins into context

The progress view is useful before a chart is impressive—it explains what the next check-in unlocks.

  1. 01

    Suggested workout programs with clear start and progress actions

  2. 02

    Daily calorie targets, food entries, macros, and profile-based estimates

  3. 03

    Body check-ins and progress views across body, gym, and calories

  4. 04

    A planned advisory coach grounded in relevant user-owned context

03 / Architecture

User-owned data with a deliberate future AI boundary.

Authentication and row-level policies are designed to keep records scoped to their owner. The planned coach boundary assembles only relevant context server-side and remains advisory.

FitAI context path
  1. 01Flutter

    Workout, nutrition, and progress experiences

  2. 02Supabase Auth

    Identity and session management

  3. 03PostgreSQL + RLS

    User-owned fitness records

  4. 04Edge Function

    Planned server-side assembly of coach context

  5. 05OpenAI

    Planned advisory response generation

  • Flutter
  • Dart
  • Material 3
  • Supabase Auth
  • PostgreSQL
  • Row Level Security
  • Edge Functions (planned)
  • OpenAI Responses API (planned)
  • fl_chart

04 / Product decisions

Coherence matters more than feature count.

Challenges

  1. 01

    Keeping a multi-area product coherent instead of building four unrelated trackers

  2. 02

    Protecting user-owned records while still assembling relevant coach context

  3. 03

    Drawing a clear boundary between useful advice and silent automation

What I’m learning

  1. 01

    Context is a product decision before it is an AI capability.

  2. 02

    A mobile data model has to support today’s simple action and tomorrow’s progress view at the same time.

05 / In progress

The next useful milestones.

  1. 01

    Continue refining the active-workout and food-entry flows

  2. 02

    Test progress views with realistic long-term data

  3. 03

    Build and test the advisory coach boundary with transparent context