The problem
People new to whiskey don't know what to buy next. Most advice is a ranked list of "best bottles," which ignores what you already own. Three similar bourbons make a worse bar than three different styles.
The recommendation engine
The core of the app is a gap engine. It doesn't rank bottles. It looks at your shelf and finds what's missing in three ways.
- Your shelfThe bottles you own
- Flavor gapSweet, spice, fruity, smoky or rich
- Cocktail gapA classic you can't make yet
- Proof gapNothing easy-drinking, or nothing bold
- Up to 3 picksEach with a plain reason
- Flavor. Every bottle carries flavor tags grouped into five families. If your shelf has no smoky bottle, the engine suggests the most beginner-friendly one.
- Cocktails. It checks six classics, such as the Manhattan and the Old Fashioned, and suggests the most versatile bottle for the first one you can't make.
- Proof. It checks whether you have something under 43% ABV for guests and something at 50% or above for cocktails.
Each recommendation comes with a reason in plain words, so the app teaches as it suggests. It's rules, not machine learning, which keeps every suggestion explainable.
What's in the app
| Part | What it does |
|---|---|
| Shelf | Track the bottles you own |
| Catalog | Browse 174 bottles with flavor, cocktail and price data |
| Preference quiz | Find a starting point from your tastes |
| Recipes | 15 cocktails, matched to what's on your shelf |
| Learn | Guided paths through whiskey styles |
| Tasting log | Record what you've tried |
| Progress | 19 achievements and badges |
How it's built
The app is React Native with Expo and TypeScript, across 15 screens. Supabase provides the Postgres database and user sign-in, and the bottle catalog is defined in SQL. Sign-in tokens are kept in the phone's secure storage.
Status
BarBuildr is a working prototype that runs on a phone through Expo. It hasn't been released to the app stores.