AI personality design for Scrappy’s Bitters. Solo 3-month project. 9 features tested down to 3. Saved $20K.
Solo, full ownership from research through final designDuring the pandemic, a friend who's COO at Scrappy's Bitters hooked me up with their full lineup and started sharing recipes he was experimenting with. I already liked to cook, so cocktails were a natural fit. Tweaking ratios, trying different bitters in familiar drinks, figuring out what I liked.
I started making batch cocktails and putting them in mason jars for friends to pick up from our porch. It was a way to share something and feel connected when everyone was feeling really, really isolated (and nervous). Over the months, people started asking me directly for recipes and shopping lists. "What five ingredients can I get to make a few cocktails I'd like in this arena?" I was answering all of that "by hand", so to speak.
That behavior shaped the whole project. I designed Pappy, an AI bartender for Scrappy's Bitters. Three-month solo project. Scruffy, cool bartender energy -- deep knowledge of craft cocktails, not too precious about experimenting, and happy to help you figure out what you actually liked.
Hero shot of Pappy's main conversation UI — chat interface with Pappy's personality on display
The original concept had nine primary features. Through user testing, I watched something that changed the whole direction: what people said they wanted and what they actually engaged with were completely different. In the early interviews, users mentioned all kinds of features they'd want. In mid-fidelity testing, they had little patience for feature investment and onboarding, and wanted to just skip through to the AI chat. Nearly every time. So I cut six features and focused the MVP on three things: a home screen with inspiration prompts, the conversation with Pappy, and a clean recipe viewer.
Before/after comparison — the 9-feature concept map vs. the final 3-feature MVP
3-screen triptych: home with inspiration prompts, conversation view, recipe card
User confidence in cocktail experimentation went from 3.5 to 4.3 out of 7. SUS scores jumped from 73 to 83. People started discovering new uses for ingredients they already owned. That was my favorite part.
Data visualization showing before/after metrics — confidence and SUS scores
OptionalWhen I designed Pappy, I'd never shipped an app. Scrappy's didn't have the budget to build it past prototyping. But about two years later, I have the technical capability to build what I once could only design.