Behind the Build: How We Created an AI-Powered Luxury Ranch Property Advisor

Introduction

Imagine searching for your dream luxury ranch or property, but instead of sifting through endless listings, you’re guided by a friendly, knowledgeable AI assistant—one that understands your needs, asks the right questions, and recommends properties tailored just for you. That’s the experience we set out to create with the Henry AI SoulMachine platform.

Built for real estate professionals, property seekers, and innovative agencies, this project solves the problem of impersonal, overwhelming property searches. By combining conversational AI, real-time data, and a visually engaging frontend, we deliver a unique, human-like digital advisor that makes property discovery both delightful and efficient.

General Overview

Henry AI SoulMachine is an end-to-end platform that:

  • Lets users interact with a digital “ranch specialist” (Henry) via chat and voice.
  • Gathers user preferences conversationally (budget, location, property type, amenities).
  • Recommends matching luxury ranch properties from a live database.
  • Provides an interactive map and property details.
  • Seamlessly blends AI-driven conversation with real estate expertise.

Key Features:

  • Conversational AI assistant with a Western persona.
  • Real-time property recommendations.
  • Interactive property map and details.
  • Voice and chat interface.
  • Scalable, API-driven backend.

Tech Stack

Frontend:
- Next.js (React) for fast, modern UI. - Tailwind CSS for styling. - Leaflet for interactive maps. - SoulMachines Web SDK for digital human integration.

Backend:
- FastAPI (Python) for robust, async APIs. - Uvicorn for ASGI server. - Prisma ORM with SQLite for database access. - Modular API design (OpenAI, property, SoulMachine, RAG endpoints).

Database:
- SQLite (via Prisma ORM) for property data.

AI/ML:
- OpenAI GPT-4o for conversational intelligence. - LangChain, HuggingFace, and Torch (for future extensibility). - SoulMachines digital human for lifelike user interaction.

Infrastructure/DevOps:
- Poetry for Python dependency management. - Ngrok for secure local development and public access. - Vercel for frontend deployment.

How AI Powers the System

AI is at the heart of Henry AI SoulMachine:

  • Conversational Intelligence:
    The backend uses OpenAI’s GPT-4o model, guided by a custom system prompt that gives Henry a warm, Western personality. The AI gathers user preferences, asks follow-up questions, and keeps the conversation focused on property discovery.

  • Real-Time Recommendations:
    User inputs are parsed and used to query the property database. The AI then summarizes and presents the best matches, adapting its suggestions based on user feedback.

  • Digital Human Integration:
    The SoulMachines SDK brings Henry to life as a digital human, enabling natural voice and facial interactions.

  • Integration Method:
    The backend communicates with OpenAI via secure API calls, passing structured prompts and tool definitions. The frontend connects to the backend via REST and WebSocket APIs.

  • AI Challenges & Solutions:

  • Maintaining Persona: A detailed system prompt ensures Henry’s responses are always on-brand.
  • Real-Time Data: The AI is augmented with up-to-date property data via tool integrations.
  • Seamless Handoffs: The system gracefully handles off-topic queries, always steering users back to property search.

Technical Breakdown

Database Structure:
- The Property model stores all relevant details (price, type, location, amenities, images, etc.). - Prisma ORM provides async, type-safe queries.

Backend Architecture:
- Modular FastAPI app with routers for OpenAI, property, SoulMachine, and RAG (retrieval-augmented generation). - WebSocket endpoints for real-time chat. - REST endpoints for property data and AI session management. - Tooling layer defines functions the AI can call (e.g., fetch properties).

Key API Routes: - /openai/session: Starts a new AI chat session. - /db/property/fetch: Returns property listings based on filters. - /soulmachine/ws/simple_chat_with_rag: WebSocket for live chat with the digital human.

Engineering Decisions: - Used FastAPI for async performance and modularity. - Prisma ORM for easy schema evolution and type safety. - SoulMachines SDK for best-in-class digital human experience. - Ngrok for secure, public development URLs.

User Journey Walkthrough

  1. Landing on the Platform:
    The user visits the web app and is greeted by Henry, the digital ranch specialist, via chat or voice.

  2. Conversational Onboarding:
    Henry introduces himself and asks about the user’s budget, preferred locations, and property types.

  3. Preference Gathering:
    The user responds naturally (text or voice). Henry asks follow-up questions, one at a time, to clarify needs.

  4. AI-Driven Search:
    The backend parses user preferences and queries the property database for matches.

  5. Property Recommendations:
    Henry presents tailored property options, complete with images, descriptions, and an interactive map.

  6. User Feedback Loop:
    The user can ask for more options, refine their criteria, or request details. Henry adapts recommendations in real time.

  7. Closing the Conversation:
    When the user finds a property they like, Henry offers next steps and thanks them warmly.

  8. Seamless Experience:
    All interactions are smooth, with instant feedback and a lifelike digital human interface.

Text-Based Flowchart / Architecture

User
  ↓
Frontend (Next.js)
  → Loads SoulMachines digital human
  → Renders chat/voice UI and property map
  ↓
User sends message (text/voice)
  ↓
Frontend triggers API/WebSocket call
  ↓
Backend (FastAPI)
  → Receives user input
  → Passes input to OpenAI GPT-4o with system prompt and available tools
  → AI interprets, asks clarifying questions, or calls property fetch tool
  ↓
If property data needed:
  → Backend queries SQLite via Prisma ORM
  → Returns property data to AI
  ↓
AI generates response (recommendation, follow-up, etc.)
  ↓
Backend sends response to frontend
  ↓
Frontend updates UI (chat, map, property cards)
  ↓
User continues conversation or ends session

Conclusion

Henry AI SoulMachine reimagines luxury property search by blending advanced AI, real-time data, and digital human interaction. The result is a platform that feels personal, engaging, and highly effective—showcasing our team’s expertise in AI, full-stack engineering, and user experience design.

Future Plans:
- Expand to more property types and regions. - Add multi-language support. - Integrate advanced analytics for agents. - Enhance the digital human’s capabilities.