Housing Wealth Agent: Full Implementation Plan

Overview

This project builds an AI-powered housing advisor for Saudi Arabia that helps buyers find the best neighborhoods, the right time to buy, and whether a property is overpriced. It uses a multi-agent workflow in Swarms.

Business Goal

Turn housing search from emotional browsing into data-driven decision support.

Core Users

• Saudi homebuyers.
• First-time buyers.
• Families comparing neighborhoods.
• Investors looking for fair valuation.

Product Modules

1. Data ingestion.
2. Market valuation.
3. Project and infrastructure analysis.
4. Buyer affordability.
5. Final report generation.

Agents

1) Market Scout

Collects property and market data.

2) Asset Valuator

Compares asking price with fair market value.

3) Project Monitor

Scans projects and growth indicators around the neighborhood.

4) Financial Strategist

Calculates monthly burden and financing fit.

5) Report Composer

Writes the final recommendation report.

Inputs

• Property title.
• Area.
• Asking price.
• Neighborhood.
• Age of property.
• Buyer income.
• Buyer commitments.
• Down payment.

Technical Stack

• Python.
• Swarms.
• PostgreSQL.
• FastAPI.
• Next.js or Streamlit.
• OCR / document parsing tools later.

Implementation Phases

Phase 1

Build the data model and manual data sources.

Phase 2

Implement the five agents and connect them in sequence.

Phase 3

Build the user interface.

Phase 4

Add real market feeds and inference scoring.

Phase 5

Pilot with real users.

Prompts

Each agent should receive a strict Arabic system prompt to keep the output structured and consistent.

Next Step

Use the starter script in output/housing_wealth_agent_starter.py as the base and extend it with real data sources and Swarms orchestration.