noms - save the spot

Save your next spot from Instagram Reels, TikTok, and more instantly. Never lose a great restaurant again: TestFlight

Screenshot 1 Screenshot 2 Screenshot 3

Overview

While scrolling, I screenshot and 'save to noms' a video of incredible crab fried rice from Chicago. It finds a match - a modern Vietnamese eatery.

It tells me its other must try items, that a 20-30 minute wait is typical, and I must arrive between 4-6pm for a bar happy hour. I am more than excited to now for it.

A moment of food discovery is defined as that split-second of pure intent when you see something delicious and think "I need to go there." This happens almost daily for users across Instagram Reels, TikTok, YouTube Shorts, and other forms of short-form content. These moments don't typically resurface.

noms bridges the gap between social media discovery and real-world dining.

It captures restaurants at the moment of peak interest, enriches them with comprehensive data from Yelp's AI API, and organizes them into your personal food map - ready exactly when you need them.


Features

Feature Description
iOS share extension Share restaurants with any app through a screenshot
AI restaurant matching Finds exact restaurant from constructed smart queries and structured data (e.g., geocoding)
Comprehensive enrichment Auto-populates hours, reviews, photos, prices
Highlights section AI synthesized content of business attributes and reviews of essential tips to know about a restaurant before going
Visual food map Interactive map of all your saved noms
Shared lists Collaborate on noms with family and friends
Smart caching Ensuring compliance with Yelp API terms of use (<24h TTL), while reducing load times and costs

How It Works

diagram

Yelp AI API Integration

Restaurant matching

body: JSON.stringify({
  query: extraction.search_query,
  user_context: {
    latitude: extraction.coordinates.lat,
    longitude: extraction.coordinates.lng,
  },
  request_context: { skip_text_generation: true },
}),
  • Accurate matching of structed data extracted from a screenshot to a restaurant
  • Resolves variations, misspellings, and partial names to find the right match
  • Provides a summary about the restaurant

Highlights generation

const query = `Based on reviews for ${businessName}:
RESTRICTIONS:[age/dress/payment requirements]
SCENE:[who goes here and atmosphere, max 8 words]
SIGNATURE:[must-try dishes/drinks]
WAIT:[weekend dinner wait]
PARKING:[difficulty and solution]
INSIDER:[pro tip from regulars]

If there isn’t enough reliable info for a category, omit that line entirely. Format exactly as shown.`;
  • Synthesizes available business information, attributes, and reviews into acionable insights
  • Extracts practical information not available via standard APIs
  • Returns a structured format for consistent UI presentation

This positions the use of Yelp AI API as an enabler of instant restaurant intelligence at scale. It directly supports transforming messy social media discoveries into comprehensive, actionable dining information without manual curation.

Development

This project is built using:

Category Technologies
Application React Native, Expo
Integrations Yelp AI API (primary data), Nominatim OpenStreetMap (geocoding), Google Gemini (OCR)
Infrastructure Supabase (Postgres, Edge Functions)

Why This Matters

The shift in behavior

Discovery has fundamentally moved from search to social. "restaurants near me" turned to:

  • Friends' Instagram stories
  • Food influencer reels
  • Viral TikTok reviews
  • YouTube travel vlogs

We don't ask users to change behavior. We integrate into their existing flow:

  • Native iOS share extension works from any app
  • Background processing means no waiting
  • Push notifications confirm saves
  • Visual map interface they already understand

noms becomes infrastructure to convert discovery into action. It captures intent at its peak and preserves it for when its needed:

  • First-mover on AI-powered food capture
  • Network effects through shared lists

Built With

Share this project:

Updates

Submission history