iOS SwiftUI Swift Apple Health Claude Code

Field Log

A personal fitness and nutrition tracker for iOS with Apple Health integration, built with Claude Code as a real test of AI-augmented development in unfamiliar territory.

A personal iOS app built to answer a real question: what does AI-augmented development actually look like when you have no prior experience with the stack?

Overview

Field Log tracks food, workouts, body measurements, and supplements in one place. It reads from Apple Health — body weight, body fat, calories burned, workout sessions — so data from a connected smart scale or Apple Watch appears automatically. All data is stored locally on device using SwiftData. No account, no server, no subscription.

Built entirely with Claude Code as the development assistant — from first SwiftUI view to TestFlight submission — with no prior iOS or SwiftUI experience before starting.

The Build

The premise was straightforward: take the same AI-augmented development approach used on client projects and apply it somewhere genuinely unfamiliar. iOS development provided that. SwiftUI, SwiftData, HealthKit, StoreKit, and the App Store submission process were all new territory.

The patterns that showed up matched what IdeaRoost observes on team projects: rapid initial scaffolding, followed by a gap between working and production-ready that required deliberate architectural review, security consideration, and testing discipline.

Features built:

  • Food logging with USDA FoodData Central and Open Food Facts integration, plus barcode scanning
  • Workout logging with Apple Health sync (active energy, workouts, body metrics)
  • Body measurement tracking with smart scale integration via HealthKit
  • Supplement tracking with daily completion state
  • Macro targets and daily calorie deficit calculation
  • Full data export (JSON and Markdown) via iOS Share Sheet

What It Demonstrates

Field Log is a direct application of the IdeaRoost premise — that experienced engineering judgment plus AI tooling moves faster and produces better outcomes than either alone. It also keeps that premise honest: the architectural decisions, the security review, and the production-readiness work were done by a person, not generated.

The same gaps that show up in AI-assisted team delivery showed up here. Catching them required knowing what to look for.

Status

Currently in TestFlight testing. Not yet publicly available on the App Store.

Privacy Policy →