Cooked Recipe App

Turn any recipe link, image, or video into an instant structured recipe, smart grocery list, and guided cooking workflow.

Comment

Inspiration

Cooking today often starts on social media — TikTok, YouTube, Instagram, blogs. But while inspiration is easy to find, actually using those recipes is frustrating. Videos are unstructured, ingredients are buried in captions, and grocery planning is manual and time-consuming.

I wanted to bridge the gap between inspiration and execution.

The goal was simple:

Turn any recipe link into a structured, shoppable, cook-ready experience in seconds.

What it does

Users can paste a recipe URL (video or webpage), and the app:

Extracts transcript or recipe text

Uses AI to parse and structure ingredients + steps

Automatically generates a grocery list

Lets users mark recipes as cooked and restart them with “Cook Again”

Instead of pausing and rewinding videos, users get a clean, structured cooking workflow.

How I built it

-Multi-Provider AI Architecture

The backend uses a fallback AI hierarchy:

Anthropic (Primary)

OpenAI (Fallback)

Gemini (Final Fallback)

This ensures reliability and avoids vendor lock-in. If one provider fails, the system gracefully falls back to the next.

-Video Handling

For video links:

If a transcript exists → use it directly

If not → transcribe audio using OpenAI Whisper

Then pass transcript into structured recipe parsing

This converts unstructured media into structured data.

-Smart Grocery Flow

Instead of treating recipes as static, the app treats each cooking event as a session. When a user marks a recipe as cooked, it can be reactivated with “Cook Again”, regenerating the grocery list.

This allows:

Reusability

Clean grocery state management

Future expansion into cooking history and analytics

-Subscription Infrastructure

The app integrates:

RevenueCat for subscription management

Apple App Store subscription handling

Development builds for native testing

This allows a scalable freemium model while keeping business logic separated from app logic.

Challenges we ran into

-Expo Go vs Native Modules

RevenueCat doesn’t work inside Expo Go due to native module limitations. I had to move to development builds to properly test purchases and paywalls.

-Apple API Key Complexity

Apple has multiple API key types (App Store Connect vs Subscription Keys). Using the wrong key results in upload failures. Generating the correct SubscriptionKey_XXXX.p8 was required to connect RevenueCat properly.

-Video Transcript Reliability

Not all platforms provide accessible transcripts. Designing graceful fallbacks — including transcription and manual input — was necessary to ensure reliability without breaking the user experience.

-Prompt Engineering for Structured Parsing

Recipe content varies wildly in format. Getting consistent structured output required refining prompts to handle:

Missing measurements

Implicit ingredients

Casual spoken instructions

Different international measurement systems

Accomplishments that we're proud of

Multi-Provider AI Architecture

One of the most significant accomplishments is building a resilient, multi-provider AI fallback system. Instead of relying on a single model, the backend intelligently falls back across providers (Anthropic → OpenAI → Gemini) to ensure consistent parsing reliability.

This design reduces downtime risk, prevents vendor lock-in, and improves system robustness under real-world conditions.

Turning Unstructured Content into Structured Workflows

The core technical achievement is reliably transforming messy, unstructured recipe content (videos, transcripts, captions, blogs) into:

Clean ingredient lists

Ordered step-by-step instructions

Structured grocery lists

Handling inconsistent formatting, spoken instructions, missing measurements, and varying writing styles required careful prompt engineering and validation logic.

Video Transcript + Transcription Pipeline

Supporting video links required a layered approach:

Use transcript when available

Fall back to audio transcription when necessary

Parse the resulting transcript into structured recipe data

Designing this pipeline to fail gracefully while maintaining usability was a major milestone.

Subscription Infrastructure

Successfully integrating subscription logic required:

Native module handling outside of Expo Go

Development builds for proper paywall testing

Correct Apple subscription key configuration

Linking subscription logic cleanly to app entitlements

This ensures the app is monetization-ready and scalable.

Clean Cooking Session Logic

Rather than treating recipes as static objects, the app models cooking as repeatable sessions. The “Cook Again” functionality allows users to regenerate grocery lists without losing historical context.

This lays the groundwork for future features like cooking history, usage analytics, and personalization.

What I learned

How to design a multi-provider AI fallback system

How to integrate subscriptions properly in a React Native/Expo environment

How to handle native module limitations in development workflows

How to architect AI features for resilience instead of single-point dependency

How to transform unstructured media into structured, usable data

What's next for Cooked Recipe App

Customizable Batch Sizing

Users will be able to dynamically adjust serving sizes. Ingredient quantities will automatically scale. This makes recipes adaptable for meal prep, families, or single servings.

Clear Premium vs Free Breakdown

A dedicated screen will transparently display feature differences between free and premium tiers, improving conversion clarity and user trust.

AI Cooking Assistant

An interactive AI assistant will allow users to ask contextual questions such as:

“Can I substitute this ingredient?”

“How long will leftovers last?”

“Can I make this dairy-free?”

“What can I cook with what I have?”

This transforms the app from a static tool into a dynamic cooking companion.

Reminder Push Notifications

Users will be able to:

Set cooking reminders

Schedule meal prep days

Receive grocery shopping reminders

This increases retention and supports habit formation.

Automatic Recipe Image Generation

Each recipe will automatically generate a clean visual image for display in the app. This improves:

Visual organization

User engagement

Shareability

AI-generated imagery ensures consistency even when the original source lacks usable images.

UI and UX Improvements

Planned refinements include:

Smoother transitions

Improved grocery checklist interactions

Better visual hierarchy

Enhanced loading states

More intuitive navigation

The goal is to make the experience feel polished, premium, and frictionless.

Long-Term Vision

Future expansion may include:

Pantry inventory tracking

Smart meal planning

Nutritional breakdown analysis

Grocery delivery integrations

Shared family cooking accounts

The long-term objective is to become the operating system for modern home cooking — structured, intelligent, and effortless.

Built With

Share this project:

Updates