Reel Meal

Instantly turn cooking videos & photos into recipes with AI. Organize cookbooks, auto-generate shopping lists, and get step-by-step help from your personal AI Sous Chef.

Comment

Inspiration

Every day, millions of people discover amazing recipes on TikTok, Instagram, and YouTube - and then lose them in an endless scroll. There's no easy way to go from a 60-second cooking video to an actual recipe you can follow in the kitchen.

Reel Meal came from the frustration of bookmarking dozens of cooking videos and never being able to find them again - let alone cook from them without constantly pausing and rewinding. I wanted to bridge that gap: one link in, a full recipe out.

What it does

Reel Meal turns any cooking video or photo into a structured, searchable recipe using AI. Just share a link from TikTok, Instagram, or YouTube (or snap a photo of a recipe from a cookbook), and Reel Meal automatically extracts ingredients, steps, cook times, and a thumbnail (or AI generates one!).

From there, you can:

  • Organize recipes into cookbooks and share them with friends or family via iCloud
  • Generate shopping lists from your recipes with smart ingredient grouping
  • Cook hands-free with step-by-step cooking mode
  • Ask your AI Sous Chef for substitutions, scaling, dietary tweaks, or cooking tips mid-recipe

How I built it

Reel Meal is a native iOS app built with Swift and SwiftUI, following Clean Architecture with MVVM. The AI pipeline runs through a Cloudflare Worker proxy:

  1. Apify scrapes video metadata and transcripts from social platforms
  2. Gemini AI transcribes audio when no transcript is available, then extracts structured recipe data
  3. Imagen generates thumbnails when the source video doesn't provide one
  4. CloudKit syncs everything across devices
  5. RevenueCat manages the freemium model - 5 free AI imports, unlimited with Premium

A Share Extension lets users send links directly from any app without opening Reel Meal, and imports happen in the background. The app uses zero third-party UI dependencies - every component is built with native SwiftUI.

Challenges I ran into

  • Video platform scraping is fragile. Each platform structures content differently, and some don't expose transcripts. I built fallback paths: scrape transcript → transcribe audio → extract from thumbnail as a last resort.
  • AI extraction accuracy. Getting Gemini to output structured recipe data reliably required significant prompt engineering and validation logic.
  • CloudKit sync edge cases. Handling offline edits, deletion conflicts, and concurrent syncs with UserDefaults persistence led to subtle bugs. Debounced saves and immediate metadata writes don't mix - I learned that the hard way.
  • Share Extension sandboxing. iOS share extensions run in a separate process with strict memory limits. Coordinating with the main app via App Groups required careful state management.

Accomplishments that I'm proud of

  • The import just works. Paste a link, get a full recipe in seconds - it feels like magic every time.
  • Zero third-party UI dependencies. Every view, animation, and component is native SwiftUI.
  • Seamless share extension. Sharing a video from TikTok or Instagram directly into Reel Meal feels like a native platform feature.
  • AI Sous Chef context. The chat assistant knows your recipe and can answer specific questions about it - not providing just generic cooking advice.
  • Freemium that's fair. Free users get 5 AI imports and unlimited manual recipes. Premium unlocks everything without paywalling basic functionality.

What I learned

  • Prompt engineering is as important as code architecture when AI is core to your product
  • Shipping an MVP means saying no to features that feel essential - I cut meal planning, active pantry management, and nutrition tracking to focus on the core import-to-cook flow
  • Native SwiftUI can produce polished, performant UIs without reaching for third-party libraries
  • Building for real content (messy video transcripts, missing metadata, foreign languages) is a completely different challenge than building for clean test data

What's next for Reel Meal

  • Android version to reach a wider audience
  • Meal planning with weekly calendars and automatic shopping list generation
  • Nutrition tracking by extracting calorie and macro data from recipes
  • Social features - follow friends, discover trending recipes, and build a community cookbook
  • Smarter AI - learn user preferences over time to suggest recipes, substitutions, and portion adjustments automatically

Built With

Share this project:

Updates