Deepak Sharma

Why we built Kooking around swipe-to-discover — and how our AI match score actually works

by

When we started building Kooking, we asked ourselves one honest question: why do people stop using recipe apps?

The answer kept coming back to the same thing — friction. You open an app, see a wall of search results or a generic feed, and spend more time hunting than cooking. Discovery felt like work.

So we rethought the whole experience from scratch.

🃏 Why swipe-to-discover?

Swipe mechanics aren't just a design trend — they match how people actually make decisions about food. When you're figuring out what to cook tonight, you're not searching for a specific dish. You're browsing. You're in a mood. You want something to catch your eye.

Swiping right (save it) or left (not today) gives users an effortless way to signal their taste in real time. No typing, no filters, no rabbit holes. Just a fast, visual feed that feels more like flipping through a magazine than querying a database.

It also creates a feedback loop. Every swipe teaches Kooking more about you — your cuisine preferences, dietary habits, complexity tolerance, and the kinds of dishes you actually make vs. just bookmark.

🤖 How our AI match score works

Every recipe in Kooking gets a match score that's personal to you — not a global popularity score, but a relevance score built around your profile.

Here's what feeds into it:

1. Taste profile — built from your swipes, saves, and completed cooks. We track which cuisines, ingredients, and cooking styles you gravitate toward.

2. Cooking behaviour — do you prefer quick weeknight meals or weekend projects? High ingredient counts or minimal prep? The AI weights recipes accordingly.

3. Meal plan context — if you've already planned chicken twice this week, the AI will deprioritize more chicken recipes and surface variety instead.

4. Seasonal & time signals — recipes are weighted based on time of day, day of week, and season. A slow braise scores higher on a Sunday than a Tuesday at 6pm.

5. Creator affinity — if you consistently save from certain creators, their new content gets a boost in your feed.

The result is a score from 0–100 shown on each recipe card. It's not magic — it's transparent. We want users to understand why a recipe is being shown to them, not just feel like an algorithm is controlling their plate.

We're still early and the model keeps improving with every interaction. Would love to hear from the PH community — what signals would YOU want an app like this to learn about you? 👇

11 views

Add a comment

Replies

Be the first to comment