Hey PH community
Kooking hits the App Store on April 30th and I wanted to share a behind-the-scenes look at what we've been building and the thinking behind it.
The core idea: most recipe apps are graveyards. You save hundreds of recipes and cook maybe 5. We built Kooking's AI match score to fix that it ranks recipes by how likely you are to actually make them, based on your skill level, pantry, and time.
But beyond the AI, Kooking is a community a feed of real home cooks sharing what they're actually making, not polished food blog content.
Hey Product Hunt! π Deepak here, maker of Kooking.
One of the core design decisions we made early on was the swipe-to-discover feed β and I want to share why we went this way and how it actually works under the hood.
π€ The problem with recipe discovery today
Every recipe app I tried had the same broken loop: you search, you scroll, you bookmark 200 things, and you never cook any of them. The problem isn't a lack of recipes β it's that none of them feel personally relevant in the moment.
Searching requires intent. But most nights, you donβt have a dish in mind β you just want something to catch your eye.
π Why swipe?
Swiping maps perfectly to how people actually decide what to cook. Itβs visual, low-friction, and mood-driven. Right means "save this," left means "not tonight." No search bar, no filters, no rabbit holes.
But the real power is what happens behind each swipe. Every interaction β save, skip, cook, import β feeds a live taste profile that makes the next card more relevant than the last.
β¨ How the AI match score works
Each recipe card shows a personal match score (0β100). Itβs not a global popularity rank β itβs built entirely around you. Hereβs what drives it:
Taste profile β cuisines, ingredients, and cooking styles you consistently save or cook
Cooking behaviour β quick weeknight meals vs. weekend projects, minimal prep vs. complex techniques
Meal plan context β if youβve already got chicken twice this week, the feed surfaces variety automatically
Time & season signals β a slow braise scores higher on a Sunday afternoon than a Tuesday at 6pm
Creator affinity β recipes from creators you consistently engage with get a natural boost
The score is intentionally visible on every card. We want it to feel transparent β not like a black-box algorithm deciding what you eat, but a smart suggestion you can trust or override.
The more you swipe, the sharper it gets. Early users have told us it starts feeling "weirdly accurate" after just a couple of sessions.
Would love to know β what would make a recipe discovery feed feel truly personal to you? Drop it below π