Inspiration

Gabby Beckford's Packs Light community embodies something powerful: the courage to overcome self-doubt and pursue transformative experiences. Her audience isn't just interested in travel destinations—they're seeking permission to become the person who takes that trip, asks for that raise, or makes that leap.

We noticed a gap: plenty of apps help you plan what to do, but few address why you're not doing it. GOING was born from a simple insight—the biggest obstacle isn't lack of information, it's the internal blockers that keep us stuck.

What it does

GOING is an AI-powered action coach that transforms overwhelming goals into concrete, completable steps—tailored to your specific blockers.

How it works:

  1. Tell GOING where you want to go (your goal) and what's blocking you
  2. Choose your pace: Minimal (5-15 min), Balanced (20-45 min), or Aggressive (60-90 min)
  3. Get 3 personalized actions you can complete TODAY
  4. Premium users can chat with "Go"—an AI coaching agent that provides deep-dive guidance on any action

The secret sauce: When you tap "Done," GOING doesn't just take your word for it. Our AI validates your progress by reviewing your coaching conversation, celebrating genuine effort and gently pushing back when you haven't fully engaged. It's accountability that actually works.

How we built it

Frontend: .NET MAUI for true cross-platform development (iOS, Android, Mac Catalyst) from a single C# codebase. This enabled a two-person team to ship to multiple platforms without maintaining separate codebases.

AI Backend: AWS Lambda functions calling Claude via Amazon Bedrock. Three endpoints power the experience:

  • /today-actions — Generates personalized action plans based on goals and blockers
  • /action-chat — Powers the "Go" coaching agent conversations
  • /validate-done — Reviews chat history to validate genuine progress

Monetization: RevenueCat handles all subscription logic across both platforms. Free tier gets one journey with action generation. Premium unlocks unlimited journeys, AI coaching, and the validation system.

Voice Input: Native speech-to-text for frictionless onboarding—just speak your goal and blocker.

Challenges we ran into

Apple's Provisioning Gauntlet: iOS code signing is notoriously complex. We battled certificate/provisioning profile mismatches, "Apple Distribution" vs "iPhone Distribution" certificate confusion, and subscription metadata requirements. Multiple rejections taught us that Apple's review process requires meticulous attention to Terms of Use links, EULA references, and localization completeness. As of submission, Apple's subscription products are approved, but Apple's StoreKit servers have not yet propagated the products to devices — the iOS app is live on the App Store, but premium features may show as unavailable. For the complete experience, including working subscriptions, use the Android link in "Try it out" to join the Google Play closed test.

Subscription Configuration: Getting RevenueCat, App Store Connect, and our app all pointing to the same product IDs required careful coordination. Product IDs must match exactly—even minor discrepancies cause "Unavailable" errors.

AI Response Consistency: Getting Claude to return valid JSON reliably required careful prompt engineering. We implemented fallback parsing, code fence stripping, and schema validation to handle edge cases.

Cross-Platform Nuances: Hyperlink handling differs between iOS and Android. We implemented platform-specific solutions using GestureRecognizers and custom renderers.

Accomplishments that we're proud of

🎯 AI-Powered Validation: The "Done" validator is genuinely novel—using AI to assess effort quality, not just completion. This creates real accountability without being punitive.

🎙️ Voice-First Onboarding: Zero friction to start. No account creation, no typing required—just speak your dream and your blocker.

🎉 Celebration System: Confetti animations and audio feedback make completing actions genuinely satisfying. Dopamine hits drive habit formation.

📱 True Cross-Platform: One codebase serving iOS, Android, and Mac Catalyst with platform-appropriate behaviors.

💳 Production-Ready Monetization: RevenueCat integration with working subscriptions, restore purchases, and proper paywall presentation.

What we learned

  1. Apple's process is documentation-heavy: Every link, every localization field, every metadata entry matters. Read the rejection notes carefully—they tell you exactly what to fix.

  2. RevenueCat simplifies enormously: Managing subscriptions across platforms would be nightmarish without it. The SDK abstraction lets you focus on product, not payment plumbing.

  3. AI prompts are products: The difference between a good and great AI feature is prompt engineering. Specific instructions, clear output schemas, and thoughtful examples make Claude shine.

  4. Ship the MVP: We scoped aggressively. Features like photo evidence for validation and social sharing got cut. What shipped is focused and functional.

What's next for GOING

📸 Photo/Evidence Validation: Let users submit photos or screenshots as completion evidence

👥 Community Sharing: Post completed journeys to the PacksLight community for accountability and inspiration

🎙️ Custom Voice Coaching: Imagine Go speaking in Gabby's voice, offering encouragement in her signature style

📊 Progress Analytics: Visualize your journey history, completion rates, and growth over time

🌍 Localization: Expand beyond English to serve Gabby's global audience

Built With

  • .net-maui
  • amazon-bedrock
  • app-store-connect
  • aws-api-gateway
  • aws-lambda
  • azure-devops
  • c#
  • claude-ai
  • google-play-console
  • javascript
  • jetbrains-rider
  • lottie-animations
  • node.js
  • realm-database
  • revenuecat
  • xcode
Share this project:

Updates