Building a Rental Listing Agent: AI-Powered Property Marketing and Description Generation

Why AI-Powered Listing Creation Matters

A property manager listing 20 vacant units writes the same type of description 20 times. The result is often generic, repetitive, and fails to highlight what makes each unit unique. An AI listing agent generates tailored, engaging descriptions, captions photos, optimizes for search engines, and distributes to multiple platforms — turning a 45-minute task into a 2-minute review.

Generating Property Descriptions

The core tool takes structured property data and produces a compelling narrative.

flowchart LR
    INPUT(["User intent"])
    PARSE["Parse plus<br/>classify"]
    PLAN["Plan and tool<br/>selection"]
    AGENT["Agent loop<br/>LLM plus tools"]
    GUARD{"Guardrails<br/>and policy"}
    EXEC["Execute and<br/>verify result"]
    OBS[("Trace and metrics")]
    OUT(["Outcome plus<br/>next action"])
    INPUT --> PARSE --> PLAN --> AGENT --> GUARD
    GUARD -->|Pass| EXEC --> OUT
    GUARD -->|Fail| AGENT
    AGENT --> OBS
    style AGENT fill:#4f46e5,stroke:#4338ca,color:#fff
    style GUARD fill:#f59e0b,stroke:#d97706,color:#1f2937
    style OBS fill:#ede9fe,stroke:#7c3aed,color:#1e1b4b
    style OUT fill:#059669,stroke:#047857,color:#fff

from agents import Agent, Runner, function_tool
from pydantic import BaseModel

class ListingInput(BaseModel):
    address: str
    unit_type: str  # studio, 1br, 2br, etc.
    sqft: int
    rent: float
    bedrooms: int
    bathrooms: float
    amenities: list[str]
    neighborhood: str
    pet_friendly: bool
    available_date: str
    unique_features: list[str]  # recently renovated, corner unit, etc.

class GeneratedListing(BaseModel):
    headline: str
    description: str
    bullet_points: list[str]
    seo_title: str
    meta_description: str

description_agent = Agent(
    name="ListingDescriptionWriter",
    instructions="""You are a professional real estate copywriter.
    Write engaging, accurate rental listing descriptions.
    Rules:
    - Never use superlatives like 'best' or 'perfect'
    - Include specific details (sqft, amenities, neighborhood)
    - Follow Fair Housing Act: no discriminatory language
    - Keep descriptions between 150-250 words
    - Write in active voice""",
    output_type=GeneratedListing,
)

@function_tool
async def generate_listing_description(
    address: str,
    unit_type: str,
    sqft: int,
    rent: float,
    bedrooms: int,
    bathrooms: float,
    amenities: str,
    neighborhood: str,
    pet_friendly: bool,
    available_date: str,
    unique_features: str,
) -> str:
    """Generate a marketing description for a rental listing."""
    prompt = f"""Create a rental listing for:
    Address: {address}
    Type: {unit_type} | {bedrooms}bd/{bathrooms}ba | {sqft} sqft
    Rent: ${rent:,.0f}/month
    Amenities: {amenities}
    Neighborhood: {neighborhood}
    Pet Friendly: {pet_friendly}
    Available: {available_date}
    Unique Features: {unique_features}"""

    result = await Runner.run(description_agent, input=prompt)
    listing = result.final_output
    return (
        f"Headline: {listing.headline}\n\n"
        f"{listing.description}\n\n"
        f"Highlights:\n" +
        "\n".join(f"- {bp}" for bp in listing.bullet_points)
    )

Photo Captioning

Property photos need descriptive captions for accessibility, SEO, and platform requirements.

import base64
from openai import AsyncOpenAI

async def caption_property_photo(
    image_path: str,
    property_context: str,
) -> str:
    """Generate an SEO-friendly caption for a property photo."""
    client = AsyncOpenAI()

    with open(image_path, "rb") as f:
        img_b64 = base64.b64encode(f.read()).decode()

    response = await client.chat.completions.create(
        model="gpt-4o",
        messages=[
            {
                "role": "user",
                "content": [
                    {
                        "type": "text",
                        "text": (
                            f"Write a concise, descriptive caption for this "
                            f"property photo. Context: {property_context}. "
                            f"Keep it under 20 words. Include the room type."
                        ),
                    },
                    {
                        "type": "image_url",
                        "url": {"url": f"data:image/jpeg;base64,{img_b64}"},
                    },
                ],
            }
        ],
    )
    return response.choices[0].message.content

SEO Optimization

Rental listings need to rank on apartment search engines and Google. The agent generates optimized metadata.

def generate_seo_metadata(
    listing: GeneratedListing,
    city: str,
    state: str,
    unit_type: str,
    rent: float,
) -> dict:
    """Generate SEO-optimized metadata for a rental listing."""
    keywords = [
        f"{unit_type} for rent in {city}",
        f"apartments in {city} {state}",
        f"{city} rentals under ${int(rent + 200)}",
        f"pet friendly apartments {city}" if "pet" in listing.description.lower() else None,
    ]
    keywords = [k for k in keywords if k is not None]

    return {
        "seo_title": listing.seo_title,
        "meta_description": listing.meta_description[:160],
        "keywords": keywords,
        "og_title": listing.headline,
        "og_description": listing.description[:200],
        "structured_data": {
            "@type": "Apartment",
            "name": listing.headline,
            "address": {"@type": "PostalAddress", "addressLocality": city},
        },
    }

Multi-Channel Distribution

Once the listing is generated, we push it to multiple platforms.

from dataclasses import dataclass

@dataclass
class PlatformConfig:
    name: str
    max_description_length: int
    supports_html: bool
    photo_limit: int

PLATFORMS = {
    "zillow": PlatformConfig("Zillow", 5000, False, 30),
    "apartments_com": PlatformConfig("Apartments.com", 3000, True, 25),
    "craigslist": PlatformConfig("Craigslist", 2000, True, 12),
    "facebook": PlatformConfig("Facebook Marketplace", 1000, False, 10),
}

def adapt_listing_for_platform(
    description: str,
    platform: str,
) -> str:
    """Adapt listing content for a specific platform's requirements."""
    config = PLATFORMS.get(platform)
    if not config:
        return description

    adapted = description[:config.max_description_length]
    if not config.supports_html:
        # Strip any HTML tags for plain-text platforms
        import re
        adapted = re.sub(r"<[^>]+>", "", adapted)
    return adapted

@function_tool
async def distribute_listing(
    listing_content: str,
    platforms: str,
) -> str:
    """Distribute a listing to specified platforms."""
    platform_list = [p.strip() for p in platforms.split(",")]
    results = []
    for platform in platform_list:
        adapted = adapt_listing_for_platform(listing_content, platform)
        results.append(f"Published to {platform} ({len(adapted)} chars)")
    return "\n".join(results)

The Complete Listing Agent

listing_agent = Agent(
    name="RentalListingAgent",
    instructions="""You are a rental listing specialist.
    Help property managers create and distribute listings.
    Always ensure Fair Housing compliance — no language
    that discriminates based on protected classes.""",
    tools=[generate_listing_description, distribute_listing],
)

FAQ

How does the agent ensure Fair Housing Act compliance?

The description agent's instructions explicitly prohibit discriminatory language. Additionally, a post-processing step scans for flagged terms (e.g., "perfect for families", "walking distance to church") that could imply preference for protected classes. Flagged content is revised before publishing.

Can the agent update listings when rent prices change?

Yes. The agent can regenerate descriptions with updated pricing while preserving the rest of the content. A update_listing tool would take the existing listing ID and new parameters, regenerate only the changed sections, and republish to all platforms.

How do you handle duplicate listings across platforms?

Each listing gets a unique internal ID. The distribution system tracks which platforms have received each listing and their platform-specific IDs. Updates and deactivations are synchronized across all channels through these mappings.

---

#RentalListings #PropertyMarketing #SEO #AgenticAI #Python #LearnAI #AIEngineering