🏠 Proppr Property Parser — Production

A production-grade n8n workflow that automatically reads raw property requirement messages from Google Sheets, extracts structured data using a locally hosted Ollama AI (LLaMA 3), validates the output, and saves clean structured records — with full error handling, routing, and logging.

Built by keviv777 as part of the india-ai-agent-toolkit

---

🧠 How It Works

[Start Parser — Scheduled/Manual Trigger]
 │
 ▼
[Read Raw Messages — Google Sheets]
 │
 ▼
[Filter and Classify]
 │
 ▼
[Route Messages]
 ├── Valid ──► [Split Multi Requirements]
 │                 │
 │                 ▼
 │          [Extract with Ollama AI]  ◄── POST http://localhost:11434
 │                 │
 │                 ▼
 │          [Validate and Parse AI Response]
 │                 │
 │                 ▼
 │          [Route Parsed Results]
 │           ├── Success ──► [Save to Structured Output — Sheets]
 │           └── Failed  ──► [Log Errors — Sheets]
 │                           [Save Failed to Structured Output]
 │                           [Mark as Processed — Sheets]
 └── Skipped ──► [Save Skipped Messages — Sheets]

---

✨ Features

• ✅ Production-grade pipeline — full error handling, logging, and routing
• ✅ AI-powered extraction — pulls price, location, BHK, area, amenities from raw text
• ✅ Zero API cost — uses self-hosted Ollama (no OpenAI / cloud AI billing)
• ✅ Smart routing — success / failed / skipped paths handled separately
• ✅ Duplicate prevention — marks processed records to avoid re-parsing
• ✅ Google Sheets as database — no external DB required
• ✅ Handles multi-requirement messages — splits and processes each requirement individually

---

🔧 Tech Stack

Tool	Purpose
n8n (self-hosted)	Workflow orchestration
Ollama (LLaMA 3)	Local AI for data extraction
Google Sheets	Input data + structured output storage
JavaScript (Code nodes)	Data validation + JSON parsing

---

📊 What Data Gets Extracted

From a raw message like:

"Need 2BHK in Andheri West, budget 45k, prefer furnished, available immediately"

The AI extracts:

{
  "bhk": "2BHK",
  "location": "Andheri West",
  "budget": "45000",
  "furnished": "furnished",
  "availability": "immediate",
  "amenities": [],
  "raw_message": "Need 2BHK in Andheri West..."
}

---

🗂️ Google Sheets Structure

Input Sheet (Raw Messages)

Column	Description
message	Raw property requirement text
source	Where message came from (WhatsApp, form, etc.)
processed	TRUE/FALSE — marks if already parsed
timestamp	When message was received

Output Sheet (Structured Data)

Column	Description
bhk	Number of bedrooms
location	Area/city preference
budget	Budget in INR
furnished	Furnished / semi / unfurnished
availability	Timeline / move-in date
amenities	List of required amenities
raw_message	Original message for reference

Error Log Sheet

Column	Description
raw_message	Message that failed
error_reason	Why parsing failed
timestamp	When error occurred

---

🚀 Setup Instructions

1. Prerequisites

• n8n running (self-hosted recommended)
• Ollama installed locally: ollama.ai
• LLaMA 3 model pulled: ollama pull llama3
• Google Sheets API credentials set up in n8n

2. Install Ollama + Pull Model

# Install Ollama (Mac)
brew install ollama

# Or Linux
curl https://ollama.ai/install.sh | sh

# Pull the model
ollama pull llama3

# Start Ollama server
ollama serve
# Runs at http://localhost:11434

3. Import Workflow

1. Download Proppr_Property_Parser_CLEAN.json
2. In n8n → Import from file
3. Connect your Google Sheets credentials

4. Configure Google Sheets

1. Create a Google Sheet with the input structure above
2. In n8n, update the Read Raw Messages node with your Sheet ID
3. Update all append: sheet nodes with your output Sheet IDs

5. Run

• Manual: Click "Execute workflow" in n8n
• Scheduled: Add a Schedule Trigger node (e.g., every 30 minutes)

---

📁 Files

proppr-property-parser/
├── Proppr_Property_Parser_CLEAN.json   # n8n workflow (import this)
└── README.md

---

💡 Use Cases

• Real estate agencies processing WhatsApp/form enquiries
• Property portals structuring raw lead data
• Rental platforms auto-classifying requirements
• Any business extracting structured data from messy text messages

---

⚠️ Important Notes

• Ollama must be running locally on port 11434 for the workflow to work
• For cloud deployment, update the Ollama URL to your server's address
• Google Sheets credentials must be configured in n8n's credential manager

---

📄 License

MIT — free to use, modify, and build upon.