Quiet: Where should we go? Let's go. Inspiration

"Where should we go?"

Everyone knows what comes next: 20 minutes of scrolling, cross-checking reviews, dropping pins on a map, polling the group chat... and then someone says, "Honestly, you decide."

We realized the problem isn't a lack of options. It's too many. Every app helps you discover more places, and none helps you actually choose one. So we built Quiet to make that decision effortless.

What it does

Quiet turns a vague intention into a confident decision.

Tell it what you're in the mood for, and instead of a wall of results, you get one focused recommendation with a clear explanation of why it fits. Love it? Save it for later.

Intent → Score → Report → Save

Intent: Say what you want in plain language. Score: Quiet evaluates nearby places against what you actually asked for. Report: You get a short, readable case for the top pick, with no noise. Save: Keep the places worth remembering.

The goal isn't to show you more places. It's to help you stop searching and start going.

How we built it

Quiet is a mobile-first app with an AI decision layer running quietly underneath.

React Native for a fast, smooth mobile experience Supabase for auth, database, and backend services PostgreSQL + PostGIS for location-aware place data and proximity queries AI decision logic that interprets user intent and evaluates candidate places Scoring and reporting layer that ranks options consistently and explains the reasoning in plain language

Rather than trusting a single LLM answer, Quiet combines structured place data with AI reasoning, so recommendations are grounded, consistent, and explainable.

Challenges we ran into Making it simple is harder than making it powerful. Every extra feature was a temptation to add friction. We had to keep cutting. Representing intent. Turning messy human requests ("somewhere chill but not boring") into something a system can score reliably took real design work. Consistent scoring. Two similar searches should give sensible, comparable results, not random ones. Structuring AI output. We had to control what the model returns so the report is concise, trustworthy, and handles uncertainty gracefully. Wiring the full stack. Connecting our local Supabase environment, database functions, API routes, and the React Native frontend meant lots of debugging across layers, and it taught us how a real application fits together. Accomplishments that we're proud of Shipping the complete loop end to end: intent, scoring, report, and save, working together in a real mobile app. Building location-aware logic with PostGIS and database functions that power the recommendations. Designing an AI experience that stays in the background, where the user sees a decision, not a chatbot. Making recommendations that explain themselves, so users can trust the pick instead of just accepting it. What we learned Building an AI product isn't about dropping an LLM into an app. The real work is designing the decision experience around it. Controlling what the user sees matters as much as what the model can do. Structured outputs and honest handling of uncertainty make AI feel reliable. Working with PostGIS, database functions, and full-stack integration gave us a much deeper understanding of how frontend, backend, and data layers communicate. The best AI often feels invisible. What's next for Quiet Group decisions: blend everyone's preferences into one pick, and end the "you decide" loop for good Personalization: learn from saved places and past choices to sharpen recommendations Richer context: time of day, weather, budget, and crowd levels in the scoring Share and go: send the final pick to friends with one tap, with directions built in

Quiet doesn't give you more choices. It gives you your next move.