/sə.ˈvɑːnt/ — one who knows (taste) deeply.
I eat out constantly. I travel constantly. And I could never answer the simplest question: "Where should we go tonight?"
Not because I lacked opinions — I have 150+ Google reviews, saved lists across 20+ cities, recommendations from friends scattered across text threads. The knowledge existed. It was just everywhere and nowhere at the same time.
Yelp shows you what strangers think. Google shows you what's popular. Neither knows what you actually like.
Savaunt is a personal taste engine. It knows what I love, what I've tried, and what I'd probably love next — anywhere in the world.
Not just restaurants. Savaunt rates and discovers hotels, bars, coffee shops, spas, barbershops, and experiences — anything worth experiencing, scored against the same taste profile. A boutique hotel in Forte dei Marmi gets compared to ones I've loved in Venice and Mykonos. A coffee shop in Paris gets measured against the best I've found in Tokyo and Tel Aviv.
It started with one person's data — mine. Curated over fifteen years of travel and dining. Every review I've ever written. Every saved place. Every city list. Every recommendation from people I trust. All of it, structured into a living brain that scores new places against my actual taste.
This is what makes Savaunt different. You can't describe an unknown place in isolation. But you can describe it through places you already know. Every discovered result gets three-dimensional comparative references:
Each dimension references a different place I've actually been to, using signals extracted from my own words. The taste profile isn't guessed — it's built from real experience.
Search for anything — a cuisine, a city, a vibe. Savaunt hits Google Places to find every matching spot on earth, then scores each one against the taste profile. A sushi restaurant in Montana gets compared to sushi spots I've loved in Tokyo, LA, and San Diego. A new Italian place in London gets measured against the trattorias I fell in love with on Lake Como.
It's not about what's popular. It's about what you'd like.
Savaunt was built in a single sitting by an AI named CC and a human named James who were both tired of not knowing where to go. What started as a handful of places from a friend's list became a global taste engine — thousands of spots across the globe, with a taste DNA built from fifteen years of dining, traveling, and obsessing over the details.
The stack: Next.js, Tailwind, Google Places API, and a custom NLP pipeline that parses natural language reviews into structured vibe/food/service signals. The brain is a static JSON export — no database needed, no server costs, infinitely cacheable. Deployed on Vercel.
Know a place worth knowing about? Savaunt accepts recommendations from trusted contributors. Each submission asks the questions that matter: how's the vibe, how's the food, how's the service, and what should I order. Recommendations are reviewed and absorbed into the brain.