← Back to Projects

LA DOLCE VITA

Premium villa experience — real client, real deployment

The Vision

La Dolce Vita is a private luxury villa in Las Terrenas, Samana Peninsula, Dominican Republic. The client is listed on Airbnb but wanted a dedicated branded site that gives potential guests a richer, more personal experience — one that reflects the quality of the property and gives them direct marketing control.

The site is designed to feel like the property itself: unhurried, elegant, and genuinely helpful. When guests are ready to book, they're redirected to the Airbnb listing. The goal is brand differentiation, not Airbnb replacement.

Core Engineering

1. AI Concierge Chatbot

An embedded chat widget powered by OpenAI GPT-4o-mini. Guests can ask anything — amenities, house rules, local restaurants, activities, transportation, or availability — and get a natural, helpful response. The bot automatically detects and responds in English, Spanish, or French. Property details live in a Google Sheet the client manages directly; the server fetches and caches it with a 5-minute TTL. No code deployments needed when content changes.

2. Real-Time Availability

Integrates with the live Airbnb iCal feed to show real-time booking availability. A custom natural language date parser understands inputs like "Is March 15th available?", "What about March 15–20?", or "First week of April" — extracts the dates, checks the live calendar, and responds naturally in the chat.

3. Smart Response Routing

The backend runs a layered decision tree on every message: keyword matching handles common questions instantly (no AI call, no latency, no cost); date queries hit the iCal checker; everything else fetches Google Sheet context, calls OpenAI, and logs to MongoDB. This keeps the experience fast and the API bill low.

4. Visual Identity

Glassmorphic UI with warm platinum brushed-metal accents. Ken Burns photo gallery with keyboard and touch navigation. Custom availability calendar with blocked-date rendering. Scroll-driven parallax transitions between page sections. Mobile-first, fully responsive across all breakpoints.

Architecture

Keyword Routing Common questions answered instantly via pattern matching — no AI call, zero latency.
Date Parser Natural language date extraction checked against live Airbnb iCal feed in real time.
Google Sheets Cache 5-minute TTL cache on property knowledge base — client updates a spreadsheet, server picks it up automatically.
Conversation Logging Every chat exchange saved to MongoDB Atlas with session ID and timestamp — client can review what guests are actually asking.
Deployment React on Vercel, Express on Render, full CI/CD pipeline. Vercel rewrites /api/* to Render so no hardcoded URLs.

Project Type

Freelance Client Project — Live Production

Tech Stack

Frontend: React 18, Vite, CSS
Backend: Node.js, Express.js
AI: OpenAI GPT-4o-mini
Data: MongoDB Atlas, Google Sheets API v4
Availability: Airbnb iCal (node-ical)
Deploy: Vercel + Render

Role

Full-Stack Engineer & Product Lead

Highlights

Multilingual AI chatbot (EN/ES/FR) • Real-time iCal availability • Client-managed content via Google Sheets • Conversation logging • Full CI/CD