Inspiration
This started from Eitan’s idea: short cooking videos are fun to watch, but annoying to cook from in real life. You pause, rewind, miss ingredients, and lose the flow. I wanted to turn that chaotic format into something you can actually use in the kitchen.
What it does
Chef converts cooking videos (mainly TikTok/Instagram links) into a structured recipe with ingredients, step-by-step instructions, and a categorized shopping list.
It also has a Focus Mode for cooking, voice controls (next/back/repeat), AI cooking Q&A, and offline recipe history so you can keep using it without re-processing the same video.
How I built it
I built the app in Flutter with a clean, minimal UI and Provider-based state management.
The app accepts shared links through share intent/deep link handling, sends them to a backend AI pipeline, and receives normalized recipe JSON.
Recipes and user preferences are cached locally with Hive for offline use.
For monetization, I integrated RevenueCat with a free tier (3 scans) and a Pro upgrade for unlimited scans and premium features like voice + export.
I also added:
- shopping list exports through the native share sheet
- analytics + crash monitoring
- haptic interactions to make step navigation feel better while cooking
Challenges
The hardest part was reliability: social video URLs are inconsistent, video quality varies a lot, and AI extraction can fail on fast cuts or vague instructions.
Another challenge was keeping processing smooth while handling mobile edge cases (share extension flow, retries, timeout states, and subscription checks).
Cross-platform behavior between iOS and Android for sharing/export also needed extra work.
What I learned
I learned a lot about product polish under hackathon pressure: making AI output predictable, designing for real kitchen use (big readable steps + hands-free control), and tying together UX, backend AI calls, and subscription logic into one coherent flow.
Log in or sign up for Devpost to join the conversation.