PureYumE

Stop losing recipes to the algorithm. PureYumE imports from any URL, matches ingredients you have on hand, and drops fresh curated recipes monthly. Cook more, scroll less.

Comment

Inspiration

I watched my wife screenshot an Eitan Bernath TikTok, say "we should make that," and then order Uber Eats twenty minutes later. That's the gap. Not a lack of recipes — a lack of a system to go from "I saw this" to "it's on the table."

When I read Eitan's brief — "Help people actually cook what they've been meaning to make" — I didn't see a hackathon prompt. I saw the exact problem I'd been trying to solve in my own kitchen.

Eitan reaches 350 million people a year across 7.24 million followers. They save his recipes by the hundreds. They screenshot TikToks, bookmark blogs, save reels. The average home cook owns 47 cookbooks and uses 3-4. The same pattern plays out digitally — save everything, cook nothing.

PureYumE was built from scratch for Eitan's audience. Not retrofitted. Not repurposed. The name, the brand, the product — all designed around one mission: turn saved recipes into dinner.

What it does

PureYumE is a five-step cooking workflow that maps directly to Eitan's brief:

1. Import — Capture recipes from anywhere with one tap. Paste a URL from TikTok, YouTube, or 10,000+ recipe sites. Snap a cookbook page. Scan a screenshot. Dictate from memory. AI extracts the full recipe, generates a photo if the source doesn't have one, and estimates nutrition data. Every recipe gets a complete card regardless of how messy the source was.

2. Match — See what you can cook right now. Log what's in your fridge, pantry, and freezer (voice, camera, or manual). PureYumE cross-references your inventory against your saved recipes and shows you what's ready to make with ingredients on hand. No more "I have nothing to cook."

3. Plan — Tap "Add to Calendar" and choose: schedule a Cook Event or a Shop Event, each on its own day. PureYumE opens the native iOS Calendar with the recipe details or ingredients already filled in. Cook Friday, shop Wednesday — you decide. Maybe you're already at the store and just want the family to see it on the shared calendar. Your meals land where you already plan your life — alongside work meetings, school pickups, and everything else.

4. Shop — Smart grocery lists auto-generated from planned recipes, compared against your On Hand inventory so you only buy what you actually need.

5. Cook — Step-by-step Cook Mode with built-in timers, ingredient checklists, and a dark-mode kitchen UI with large touch targets. When your calendar cooking event fires, you're one tap from the kitchen.

Recipe Drops — The feature no competitor has. Premium members receive curated monthly recipe collections from creators, auto-synced to their library on the 1st. No searching. No importing. Fresh recipes just appear. This is what makes PureYumE a subscription worth keeping — not just tools, but content.

Monetization — RevenueCat-powered freemium. Free tier includes unlimited manual recipes, 20 AI imports, grocery lists, Cook Mode, and calendar scheduling. Premium ($4.99/mo or $29.99/yr) unlocks unlimited AI imports, monthly Recipe Drops, ad-free experience, and premium themes. 7-day free trial on annual.

How I built it

I built PureYumE with one architectural principle: don't reinvent the wheel.

Apple already built authentication (Apple ID), a calendar (EventKit), and a payment system (App Store). RevenueCat already built subscription management. I didn't rebuild any of it. I connected the dots and focused entirely on the one thing nobody else has solved — getting people from "I saved this recipe" to "I made this for dinner."

The result is an app with no backend.

  • No user accounts. RevenueCat creates an anonymous user ID on first launch. No sign-up screen. No email. No password. Users are cooking within 30 seconds of opening the app.
  • No backend database. All recipes, grocery lists, and user data are stored locally on-device. The app works offline, launches instantly, and never depends on server uptime.
  • No custom calendar. iOS EventKit handles meal planning. One tap creates shopping and cooking events that sync to Apple Watch, lock screen, Siri, and CarPlay — for free.
  • No complex API infrastructure. Recipe Drops sync via a lightweight JSON file. One GET request, one version check, done.

The only network calls the app makes are AI API calls at the moment of import — URL parsing, recipe structuring, image generation (DALL-E for missing photos), and nutrition estimation. Once a recipe lands on the device, it's fully local and works offline forever. This is also why the 20 free AI imports isn't arbitrary — each import has a real API cost, making the paywall an honest business decision.

AI Multi-Provider Failover: The import pipeline uses a cascading provider chain — Grok (xAI) as primary (fast, cost-effective, handles ~80% of requests), on-device processing as fallback, and OpenAI GPT-4o as final resort. The user never sees an error. They just get their recipe.

Tech Stack:

  • React Native + Expo SDK 53 (cross-platform, single codebase)
  • RevenueCat SDK (anonymous auth, subscriptions, paywalls, entitlements)
  • AsyncStorage (local-first, on-device data)
  • iOS EventKit (native calendar integration)
  • Multi-provider AI pipeline (Grok, GPT-4o, Whisper, DALL-E)
  • Built with Vibecode.dev (AI-powered development)
  • Distributed via TestFlight

No servers to maintain. No database to scale. Profitable from subscriber one.

Challenges I ran into

Recipe extraction from social video platforms. TikTok and YouTube recipes live in videos, not structured text. I built a custom extraction pipeline that analyzes video captions, text overlays, audio transcription (Whisper), and video descriptions. When platforms update their structure, the multi-signal approach stays resilient because it doesn't depend on any single data source.

Fridge scanning in real-world conditions. Messy shelves, bad lighting, partially visible labels, and packaged items made ingredient detection genuinely difficult. I built a multi-model approach with confidence scoring — when the AI isn't sure, it asks the user to confirm ("Is this spinach or lettuce?") rather than guessing wrong and ruining a grocery list.

The calendar integration decision. Every recipe app builds their own meal planner. I almost did too. Then I realized: nobody checks a meal planner inside a recipe app. They check their real calendar 50 times a day. Switching to native iOS EventKit was the single best product decision I made — it meant writing less code while delivering a fundamentally better experience. The hardest part of building PureYumE was deciding what NOT to build.

Balancing AI costs with free tier generosity. Each AI import costs real money. Too few free imports and users never experience the value. Too many and the business model breaks. 20 landed in the sweet spot — enough to build a habit, enough to feel the convenience, enough to convert.

Accomplishments that I'm proud of

The complete user journey. Not "save recipes" — but discover → plan → shop → cook. Every step Eitan asked for in his brief is built into the core flow.

On Hand matching. The feature most recipe apps miss entirely. "What can I cook with what I already have?" eliminates the biggest friction point between inspiration and action.

Recipe Drops. The innovation that makes PureYumE a subscription worth keeping. Every competitor sells tools. PureYumE delivers content. New recipes from creators every month give subscribers a reason to stay — reducing churn and increasing LTV in a way that tool-only apps can't match.

The architecture. No accounts, no backend, no servers. Privacy by design, not by policy. An app that works offline, launches instantly, and is profitable from subscriber one. I'm proud of what I chose not to build.

The brand. PureYumE sounds like Pure Yummy. The E stands for Ease, Everyday, Effortless, Elevated, Explore, Enjoy, Endless, Experience, Exclusive. And for the founding creator partner — E is for Eitan.

What I learned

Users don't need more recipes. They need confidence. The real problem isn't finding recipes — it's knowing what you can make with what you have, what you need to buy, and when you should start cooking. PureYumE solves all three.

The best feature is the one you don't build. Using iOS Calendar instead of building a custom planner. Using RevenueCat anonymous IDs instead of building auth. Using Apple's payment system instead of building checkout. Every line of code I didn't write is a server I don't maintain, a database that can't go down, and a password that can't leak.

AI is only useful when it's invisible. Nobody cares about the multi-provider failover chain or the confidence scoring algorithm. They care that scanning their fridge "just works." Hide the complexity, show the value.

Freemium is psychology, not features. 20 free AI imports isn't a number I picked from a spreadsheet. It's enough to build a habit. Users who hit 20/20 are engaged users. Engaged users convert.

What's next for PureYumE

The immediate opportunity is Eitan. PureYumE is purpose-built for his audience and ready for a creator partnership:

  • Eitan's Recipe Drops — His curated collections auto-sync to premium subscribers. When Eitan picks his best recipes for the month, they appear in every subscriber's library. No importing. No searching. Just cooking.
  • Audience activation — A single social post from Eitan to 7.24M followers drives installs. Conservative modeling (2% download, 8% premium conversion) projects $743K Year 1 ARR.
  • Revenue share — Partnership structure benefits both parties. Eitan gets a dedicated app for his audience and a new revenue stream. PureYumE gets distribution.

Beyond Eitan, the Recipe Drops model scales to any creator. Phase 2 adds chefs from Eitan's network — each brings their own audience. Phase 3 brings brand-sponsored drops (food brands sponsor themed collections). Phase 4 becomes the platform — hundreds of creators, monthly drops across cuisines. The Spotify for recipes.

Short-term roadmap: Android version (React Native makes this straightforward), improved content extraction reliability, social sharing of cooking results, and deeper Cook Mode features.

Long-term: A creator marketplace where any food influencer can publish curated collections, and B2B opportunities with meal kit companies, cooking schools, and cookbook publishers.

Every platform Eitan is on was built for someone else. PureYumE was built for Eitan.

Built With

Share this project:

Updates