ChefSnapAI

ChefSnapAi helps you turn videos, photos, and handwritten notes into clean, structured recipes you can cook from instantly, then save, organize, and share with friends in one simple flow.

Comment

Inspiration

Cooking ideas are everywhere, but usable recipes are not. We kept finding great meals in reels, screenshots, and handwritten notes, then losing them when it was time to cook. ChefSnapAI was inspired by that gap: turning messy inspiration into reliable, repeatable recipes.

What it does

ChefSnapAI captures recipes from images, videos, and typed notes, then converts them into clean, structured, cook-ready steps with ingredients, quantities, and timing.

It helps users:

  • Extract text with OCR
  • Normalize ingredients and instructions
  • Save and organize recipes in a personal library
  • Revisit and share recipes quickly
  • Unlock advanced import/AI features via subscription

We also use a confidence model to rank extraction quality, e.g.
[ S = 0.5\,C_{\text{ocr}} + 0.3\,C_{\text{parser}} + 0.2\,C_{\text{ai}} ] so low-confidence outputs can be flagged for quick edits.

How we built it

  • Flutter app for cross-platform UI and flows
  • Firebase Auth for sign-in and identity
  • Firestore for user and recipe data
  • Cloud Storage for uploaded media
  • OCR + AI parsing pipeline to transform raw text into structured recipe JSON
  • RevenueCat integration for subscription/paywall management
  • Firestore security rules and indexes for safe, fast querying

Challenges we ran into

  • OCR noise from low-light photos, angled pages, and stylized fonts
  • Normalizing inconsistent ingredient formats (fractions, units, shorthand)
  • Keeping recipe parsing stable when source quality is poor
  • Designing a paywall that protects premium features without hurting onboarding
  • Balancing flexibility in data models with strict validation for reliable rendering

Accomplishments that we're proud of

  • End-to-end “capture to cook” flow that works across multiple content sources
  • Clean recipe structuring from messy real-world input
  • Production-style subscription architecture with entitlement checks
  • Thoughtful schema/security setup that can scale beyond MVP
  • A product direction that solves a real daily pain point for home cooks

What we learned

  • Good AI UX needs strong guardrails, not just strong prompts
  • Structured outputs and schema validation are critical for reliability
  • OCR quality directly affects downstream AI quality, so preprocessing matters
  • Subscription logic should be designed early, not bolted on later
  • Fast iteration in Flutter + Firebase is powerful for shipping product ideas quickly

What's next for ChefSnapAI

  • Smarter auto-fixes for quantities, units, and missing steps
  • Grocery list and meal-planning generation from saved recipes
  • Better multilingual OCR and ingredient normalization
  • Collaborative recipe spaces for families/teams
  • Deeper personalization based on dietary preferences and cooking history

Built With

  • cloud-firestore
  • firebase-storage
  • for
  • ios
  • languages:-dart-frameworks:-flutter-state-management-&-routing:-riverpod
  • local-notifications
  • share-api-networking:-http/rest-apis-target-platforms:-android
  • sign-in-with-apple-media-&-device-apis:-image-picker
  • web/desktop
Share this project:

Updates