How a Small Real Estate Investment Team Automated Market Analysis with Gemini Advanced: 15 Hours to 3 Hours Weekly

How a Small Investment Team Cut Weekly Research from 15 Hours to 3 Hours Using Gemini Advanced

A five-person real estate investment team in Seoul was spending 15 hours each week manually collecting property listings, running yield calculations, and compiling market reports. By building an automated workflow powered by Gemini Advanced and the Gemini API, they reduced that time to just 3 hours — an 80% reduction — while improving the consistency and depth of their analysis. This case study walks through their exact workflow, the code they used, and the lessons learned along the way.

The Problem: Manual Research Bottlenecks

The team’s weekly research cycle involved three painful steps:

• Data collection — Scraping listings from multiple portals, copying prices, sizes, and locations into spreadsheets.- Yield calculation — Running cap rate, cash-on-cash return, and gross rent multiplier formulas for each property.- Report generation — Writing narrative summaries and ranking properties for the investment committee.Each step was error-prone and repetitive. The team needed a solution that could handle structured data processing and natural-language report writing in a single pipeline.

The Solution Architecture

The final workflow has three stages, all orchestrated through Python scripts calling the Gemini API:

• Stage 1: Collect and normalize listing data from CSV exports.- Stage 2: Calculate financial metrics with Python, then pass results to Gemini for interpretation.- Stage 3: Generate a weekly investment brief using Gemini Advanced’s long-context capabilities.

Step-by-Step Implementation

Step 1: Environment Setup

# Install the Google Generative AI SDK pip install google-generativeai pandas openpyxl

export GEMINI_API_KEY=“YOUR_API_KEY”

You can obtain an API key from Google AI Studio (aistudio.google.com). A Gemini Advanced subscription gives you access to the latest models with higher rate limits.

Step 2: Configure the Gemini Client

import google.generativeai as genai import os import pandas as pd

genai.configure(api_key=os.environ[“GEMINI_API_KEY”]) model = genai.GenerativeModel(“gemini-2.5-pro”)

Step 3: Load and Normalize Listing Data

def load_listings(csv_path: str) -> pd.DataFrame:
    df = pd.read_csv(csv_path)
    # Standardize column names
    df.columns = [c.strip().lower().replace(" ", "_") for c in df.columns]
    # Ensure numeric types
    for col in ["price", "monthly_rent", "area_sqm"]:
        df[col] = pd.to_numeric(df[col], errors="coerce")
    df.dropna(subset=["price", "monthly_rent"], inplace=True)
    return df

listings = load_listings("weekly_listings.csv")
print(f"Loaded {len(listings)} valid listings")

Step 4: Calculate Financial Metrics

def calculate_metrics(df: pd.DataFrame) -> pd.DataFrame:
    # Annual gross rent
    df["annual_rent"] = df["monthly_rent"] * 12
    # Cap Rate = Annual Rent / Purchase Price
    df["cap_rate"] = (df["annual_rent"] / df["price"] * 100).round(2)
    # Gross Rent Multiplier = Price / Annual Rent
    df["grm"] = (df["price"] / df["annual_rent"]).round(1)
    # Price per square meter
    df["price_per_sqm"] = (df["price"] / df["area_sqm"]).round(0)
    return df

listings = calculate_metrics(listings)
top_10 = listings.nlargest(10, "cap_rate")

Step 5: Generate the Investment Brief with Gemini

def generate_report(df: pd.DataFrame) -> str:
    data_summary = df.to_markdown(index=False)
    prompt = f"""You are a real estate investment analyst.
Below is this week's top property listing data with calculated metrics.

{data_summary}

Write a professional weekly investment brief that includes:
1. Market overview (price trends, notable patterns)
2. Top 3 recommended properties with reasoning
3. Risk factors to watch
4. A comparison table of the top 5 listings by cap rate

Format the output in clean HTML suitable for an email report."""

    response = model.generate_content(prompt)
    return response.text

report_html = generate_report(top_10)

with open("weekly_report.html", "w", encoding="utf-8") as f:
    f.write(report_html)
print("Report saved to weekly_report.html")

Step 6: Automate with a Weekly Cron Job

# crontab -e (Linux/macOS)
0 8 * * 1 cd /home/user/re-analysis && python run_pipeline.py

On Windows, use Task Scheduler to trigger python run_pipeline.py every Monday at 8 AM.

Results: Before and After

Can Gemini Advanced replace a professional real estate analyst?

No. Gemini automates the repetitive data processing and report drafting tasks, but investment decisions still require human judgment about factors like neighborhood development plans, regulatory changes, and relationship-based deal flow. The team in this case study uses Gemini’s output as a starting point that their analysts then review, annotate, and present.

How much does it cost to run this workflow weekly?

For a dataset of 200 listings per week with one report generation call, the Gemini API cost is typically under $1 per run using Gemini 2.5 Pro. A Gemini Advanced subscription ($19.99/month) provides access to the model via AI Studio. Total monthly cost is approximately $20–25 including API overages. This is significantly cheaper than the labor cost of 12+ hours of manual work.

Can this workflow handle listings in languages other than English?

Yes. Gemini supports multilingual input and output. The team originally worked with Korean-language listing data and generated reports in both Korean and English. Simply specify the desired output language in your prompt, e.g., “Write the report in Korean”. The model handles mixed-language data (e.g., Korean addresses with English financial terms) without issues.