• 

  Add ingredients by using AI to scan labels

• 

  recipe almost completed

• 

  uploaded url from recipe website

• 

  Groceries Tab (what you need to get from the store for recipes)

• 

  Pantry Tab (what you have in your kitchen)

• 

  Finished recipe uploaded

• 

  Recipe Detail View

• 

  Recipe Tab

• 

  Bottom half of recipe detail view

• 

  Recipes with progress bars on the items you have in pantry

• 

  Home Tab

• 

  more recipe detail view

• 

  bottom of XP tab

• 

  XP badges tab

Inspiration# Inspiration

I cook from TikTok and Reels all the time, but "saving" a video is not the same as having a recipe you can actually use. Ingredients are scattered across captions, comments, and fast voiceovers, so I kept ending up with half-remembered steps and grocery chaos.

StoveReel started as a simple question:
What if any cooking video URL could become a clean, shoppable recipe in under a minute?

---

What I Built

StoveReel turns short-form cooking videos into structured meals you can cook tonight.

Core flow

1. Paste or share a TikTok/Instagram link
2. Extract transcript and context from the video
3. Use AI to normalize the content into:
   • title
   • ingredients
   • step-by-step instructions
   • servings and metadata
4. Instantly generate a grocery list and pantry-aware view

Product highlights

• URL import pipeline with async processing states
• Recipe card UI designed for mobile readability
• Grocery list generation from parsed ingredients
• Pantry features for "ready to cook" signals
• Share Extension so users can import directly from other apps
• Pro paywall foundations with RevenueCat integration

---

How It Works (Tech Stack)

• iOS App: SwiftUI
• Backend: Supabase (Postgres + Edge Functions + Auth + RLS)
• Video/Transcript extraction: Apify integration
• Recipe structuring / normalization: OpenAI models
• Monetization: RevenueCat subscriptions and paywall flow

I focused on a modular architecture so each stage of the import pipeline can be debugged and improved independently (ingest -> parse -> normalize -> persist -> render).

---

What I Learned

• Building AI features is mostly about pipeline reliability, not just model prompts.
• User trust comes from predictable UX: clear loading states, retries, and explicit failures.
• Share extensions are powerful, but require careful handling of auth/session transfer.
• Subscription UX is as much product design as implementation; timing and clarity matter.

---

Challenges I Faced

1) Messy input quality

Video transcripts are inconsistent. Slang, missing measurements, and implied steps made structured output hard.
Fix: Added normalization logic and defensive parsing around ingredient and instruction extraction.

2) Async orchestration

Importing can involve multiple network services and variable latency.
Fix: Implemented queue-oriented processing states and better error boundaries per stage.

3) Data integrity and security

User-specific data and generated content needed strict access control.
Fix: Enabled row-level security and scoped all access by authenticated user.

4) Delight vs. complexity

It is easy for AI apps to feel magical in demos but brittle in reality.
Fix: Prioritized graceful degradation and transparent UI feedback over hidden "magic."

---

Why I'm Proud of This

StoveReel solves a real, daily pain point with a workflow that feels natural:
see recipe -> share URL -> get meal plan and groceries.

It bridges creator content and real-world cooking in a way that is practical, fast, and genuinely habit-forming. This project pushed me to blend product thinking, backend architecture, and AI reliability into one cohesive user experience.

---

What's Next

• Improve ingredient normalization for edge cases (regional terms, optional ingredients)
• Add meal planning and batch grocery optimization
• Expand pantry intelligence (substitutions + "cook now" recommendations)
• Ship full subscription unlock loop and production analytics

---

If food content is everywhere, StoveReel is the layer that turns it into something actionable.

Built With

• apify-(video/transcript-extraction)
• async-job/queue-style-processing-for-imports
• edge-functions
• hosted-postgres)
• languages:-swift
• revenuecat-api/sdk-(subscriptions/paywall)
• revenuecat-database:-postgresql-(via-supabase)-with-row-level-security-(rls)-apis/integrations:-openai-api-(recipe-structuring/normalization)
• revenuecat-ios-sdk-(purchases-ios)-platforms:-ios
• sql-frameworks/libraries:-swiftui
• supabase-client-apis-other-technologies:-share-extension-on-ios
• supabase-edge-functions-(deno-runtime)-cloud-services:-supabase-(auth
• typescript