Pure matching and ranking logic calculates readiness, useful unlocks and taste-weighted next-bottle suggestions.
Product / web app
Cocktail Bar
A local-first cocktail PWA that understands what bottles you own, what drinks you can make and which next bottle creates the most useful new options for your taste.
The problem
People buy random bottles, then still cannot make the drinks they actually want.
Most home-bar advice starts with a generic checklist. That creates expensive collections full of bottles that overlap, gather dust or fail to unlock useful cocktails. The product needed to answer a better question: given what I already own and what I like, what should I buy next?
The insight
Optimise the next purchase for useful cocktail unlocks.
The key model separates exact bottle products from ingredient families, then evaluates cocktail readiness against the user's inventory and taste. That makes the recommendation explainable: buy one ingredient and immediately see the specific cocktails it unlocks.
The build
Build the recommendation loop into the product itself.
The product includes inventory onboarding, ready-now cocktails, one-bottle-away views, ranked next-bottle recommendations, recipe builds, tasting states, local backup, offline PWA behaviour and foundations for cloud sync, affiliate offers and native mobile packaging.
Inventory, preferences and tasting history work without requiring an account, with offline support and JSON backup.
Affiliate architecture, entitlement hooks, retailer offers and monetisation planning are separated from the recommendation quality model.
Catalogue validation, engine tests, accessibility checks and end-to-end tooling support the path from prototype to market-ready product.
Why it matters
The strongest app ideas usually come from turning a personal frustration into a repeatable decision system.
Every project is different, but the goal stays simple: understand what needs to work, design it clearly and ship something useful that can be maintained after launch.