AI Real Estate Description Generator for Portals

A major developer with a catalog of 1,000+ properties spent 3 days writing listings for 5 portals. After deploying our AI generator, that time dropped to 2 hours. The problem is familiar to every realtor and developer: manual descriptions are templated, buyers skip them, and managers waste hours on edits. We developed an AI generator that creates unique descriptions in seconds based on characteristics, photos, and geolocation. Our expertise in NLP and Computer Vision allows us to adapt the solution for any portal — CIAN, Avito, Yandex.Realty, and international platforms. The GPT-4o model with Vision API analyzes up to 5 photos, determining renovation condition, finishing materials, and layout features. The result is a specific description with no clichés. We use prompt engineering with a system message that sets the company's tone and excludes empty adjectives. Request a demo version of the generator for your catalog.

How AI Generates a Description from Photos

Input includes structured data (type, area, floor, address, price, amenities) and up to 5 photos. The GPT-4o model combines textual and visual cues: from photos it identifies renovation quality, materials, layout; from data it takes square footage and location. The prompt is configured so the first paragraph contains key information (size, location, condition), and the description includes concrete facts rather than "beautiful apartment."

from openai import AsyncOpenAI
from dataclasses import dataclass

client = AsyncOpenAI()

@dataclass
class PropertyData:
    property_type: str      # apartment, house, commercial, land
    deal_type: str          # sale, rent
    rooms: int
    area: float             # m²
    floor: int
    total_floors: int
    address: str
    district: str
    metro_distance: int     # minutes walk
    price: float
    renovated: bool
    amenities: list[str]    # balcony, parking, elevator, storage...
    year_built: int = None
    ceiling_height: float = None

async def generate_property_listing(
    property_data: PropertyData,
    portal: str = "cian",
    tone: str = "professional"
) -> dict:
    PORTAL_CONFIGS = {
        "cian": {"max_title": 100, "max_desc": 4000},
        "avito": {"max_title": 80, "max_desc": 3000},
        "yandex_realty": {"max_title": 100, "max_desc": 4000},
    }
    config = PORTAL_CONFIGS.get(portal, PORTAL_CONFIGS["cian"])

    amenities_str = ", ".join(property_data.amenities)

    response = await client.chat.completions.create(
        model="gpt-4o",
        messages=[{
            "role": "system",
            "content": f"""You are a realtor writing sales listings for {portal}.
            Style: concrete, no fluff, numbers and facts.
            First paragraph — most important (size, location, condition).
            DO NOT write: "beautiful apartment", "gorgeous view", empty adjectives.
            Title: up to {config['max_title']} characters.
            Description: up to {config['max_desc']} characters.
            Return JSON: {{title, description, key_features (3-5 facts)}}"""
        }, {
            "role": "user",
            "content": f"""
            Type: {property_data.property_type}, {property_data.deal_type}
            Rooms: {property_data.rooms}, Area: {property_data.area} m²
            Floor: {property_data.floor}/{property_data.total_floors}
            Address: {property_data.address}, {property_data.district}
            Metro distance: {property_data.metro_distance} min walk
            Renovation: {'yes' if property_data.renovated else 'needed'}
            Amenities: {amenities_str}
            {'Year built: ' + str(property_data.year_built) if property_data.year_built else ''}
            """
        }],
        response_format={"type": "json_object"}
    )
    return json.loads(response.choices[0].message.content)

Photo Analysis via Vision API

async def describe_from_photos(photo_urls: list[str], property_type: str) -> str:
    """Analyze apartment photos, describe condition"""
    image_contents = [
        {"type": "image_url", "image_url": {"url": url}}
        for url in photo_urls[:5]
    ]

    response = await client.chat.completions.create(
        model="gpt-4o",
        messages=[{
            "role": "user",
            "content": [
                {"type": "text", "text": f"Describe the condition of the {property_type} from the photos. Indicate: renovation condition, finishing materials, layout features, visible advantages. 3-4 sentences, specific."}
            ] + image_contents
        }]
    )
    return response.choices[0].message.content

Why Automating Descriptions Is Profitable?

Manual writing of one listing takes 15–20 minutes — with a flow of 100 properties per month, that's 30+ hours of a copywriter. AI generation processes the same amount in 10 minutes without loss of quality. Moreover, budget savings on copywriting reach 90%. Comparison:

AI generation is 15 times faster than a human with comparable quality. We guarantee compliance with the company's tone of voice and eliminate spelling errors.

How Batch Catalog Processing Works?

For a large developer with 500+ listings, batch generation is necessary. We use asyncio.gather (see Python documentation) to process objects in parallel:

async def process_real_estate_catalog(
    properties: list[dict],
    portal: str = "cian"
) -> list[dict]:
    generator_tasks = [
        generate_property_listing(PropertyData(**p), portal=portal)
        for p in properties
    ]
    results = await asyncio.gather(*generator_tasks)
    return [{"property": p, "listing": r} for p, r in zip(properties, results)]

Processing a catalog of 500 objects takes 15–30 minutes. The cost per listing is negligible. Implementation time — from 2 weeks for the generation script and CRM integration. Specific cost is calculated individually based on catalog volume and model requirements.

Common Implementation Mistakes?

Ignoring portal limits — title gets truncated, description cut off. As specified in the CIAN API specification, the listing title must not exceed 100 characters. We strictly set character limits in the prompt for each portal. Lack of fallback on API failure — if GPT-4o is unavailable, the system switches to a backup LLaMA 3 model via custom inference. Generation without fact-checking — the model might "assume" floor or area. We add an instruction to the prompt to strictly rely on provided data.

Implementation Process

1. Analysis: audit current listings, gather requirements for format and style.
2. Design: develop prompt, configure Vision API, prepare batch generation scripts.
3. Implementation: write generator code in Python with async support, integrate with CRM via REST API.
4. Testing: A/B test on 50 listings, adjust prompt based on CTR metrics.
5. Deployment: deploy on your server or in the cloud, document, train the team.

Technical specifications:

What's Included

• Prompt engineering for your catalog and tone of voice.
• Generator code in Python (async, batch support).
• CRM integration (REST API, webhooks).
• Adaptation for 3 portals (CIAN, Avito, Yandex.Realty or any others) with Russian language support.
• Documentation and team training (2-hour webinar + recording).
• Support for 1 month after deployment.

Order a pilot project — get a ready script in 2 weeks. Contact us to evaluate your catalog: we'll prepare a demo generation on 50 properties for free. Our experience spans many years in AI development and numerous projects in the real estate sector. Get a consultation on automating your listings today.

Generative AI Development: From Prompt to Production API

We often receive a task "generate a product image" — on the surface it seems simple. But behind this lies a choice between dozens of models, configuring the inference pipeline, manually solving consistency issues, integrating into the product backend, and answering why the model generates hands with six fingers in staging but not in production. Let's break down the directions we work with.

Image Generation: From Prompt to Production API

The current landscape includes FLUX.1 [dev/schnell/pro] from Black Forest Labs and Stable Diffusion 3.5. FLUX.1 [schnell] takes 4 steps instead of 20–50 for SDXL — 5–12 times faster — while maintaining higher quality. On an A100 80GB — 1.2–1.8 s per 1024×1024 image at batch_size=4.

A typical deployment issue: FLUX.1 [dev] requires 24+ GB VRAM in fp16. On A10G 24GB it fits tightly; at batch_size>1 — OOM. Solution: torch_dtype=torch.bfloat16 + enable_model_cpu_offload() from diffusers, or quantization via bitsandbytes to NF4 — minimal quality drop, memory consumption drops to 12–14 GB.

ControlNet and IP-Adapter are key tools for production tasks where controllability is needed. ControlNet with Canny/Depth/Pose maps provides structural control. IP-Adapter (especially IP-Adapter-FaceID) allows transferring character identity to generations — this is the foundation for personalized content. More about ControlNet can be found on Wikipedia.

Case study: e-commerce photography. A retailer with 8000 SKUs needed lifestyle photos for each product. Pipeline: product segmentation (Segment Anything Model 2) → background removal → inpainting with FLUX.1 [dev] using product image as IP-Adapter reference → upscale via RealESRGAN_x4plus. The generation cost is negligible compared to professional photography, providing huge savings. Throughput — 200 images/hour on 2× A100. Our extensive experience from 30+ projects ensures we select the optimal model for your task — an evaluation can be obtained upfront.

Why Is Model Selection Only Half the Battle?

Fine-tuning for a Specific Style or Character

Dreambooth and LoRA are the standard for adapting to a specific visual style or object. LoRA trains in 2–4 hours on 20–30 reference images on a single A100. Rank 16–32 is usually sufficient for style; rank 64+ is needed for precise face reproduction.

A common mistake: training LoRA too long — the model overfits to references, losing the ability to vary. Sign: at cfg_scale=7, all images look like copy-paste of references. Solved by early stopping (usually 1500–2000 steps for 20 images) and prior_preservation_loss.

For deeper customization — full fine-tuning via diffusers + accelerate with FSDP on multiple GPUs. But that already takes 40–80 hours of training and requires a truly large dataset (1000+ images).

Comparison of Image Generation Approaches

Video Generation: Which Models Are Best?

Open-weight video models still lag behind commercial ones in stability and length. Wan2.1 is the best choice for self-hosting: 14B parameters, runs on 2× A100, delivers acceptable quality for short clips.

The main pain of video generation is temporal consistency: the character changes clothing color at the third second, objects "drift." Partial solution — generation with motion_bucket_id and noise_aug_strength in Stable Video Diffusion, or using I2V (image-to-video) instead of pure text-to-video. As noted in VideoPoet research, consistency is achieved by training on long sequences.

AnimateDiff remains a working tool for short loops and motion effects on top of SD/FLUX. Not Sora, but deployable locally and predictable.

Music and Audio Generation

AudioCraft from Meta (MusicGen + AudioGen) is a production-ready stack for music generation. musicgen-large (3.3B) generates 30 s of music in ~8 s on A100. Control via text prompt and melody conditioning — you can specify a melody by humming.

Stable Audio Open from Stability AI is an alternative with length up to 47 s, better structural control (intro/verse/chorus). Deployment is similar: diffusers + FastAPI.

For voice-over and dubbing — ElevenLabs API or self-hosted XTTS v2 (see Speech AI service). For sound design and foley — AudioGen.

3D Generation: Current Practical State

3D generation has not yet reached the same maturity as 2D. But for specific tasks, tools are already working:

TripoSG and Shap-E — text/image-to-3D. Shap-E from OpenAI generates simple 3D meshes in seconds, but geometry is rough. TripoSG gives more detailed results but requires post-processing (remeshing, UV unwrapping).

Wonder3D and Zero123++ — 3D reconstruction from a single image. They work by generating multi-views (6–8 views) and then 3D reconstruction via NeuS or instant-ngp.

Gaussian Splatting (3DGS) — not generation, but reconstruction from a series of photos/videos. For product cards and real estate it's already production: 50–200 photos → 3DGS model in 15–30 min on RTX 4090 → interactive 3D viewer in browser.

What Infrastructure Is Needed for Generative AI Deployment?

Critical for generative models:

• Task queue — Celery + Redis or Ray Serve. Synchronous HTTP for image generation is unacceptable with >5 concurrent requests.
• Caching — similar prompts yield similar results. Semantic cache via embeddings (faiss + sentence-transformers) can reduce GPU load by 20–40%.
• Quality monitoring — CLIP score for text-image alignment, FID for evaluating generation distribution. Integrate into MLflow or Weights & Biases.
• Storage — generated images immediately to S3/MinIO, not on the inference server disk.

What's Included in the Deliverables

We take the project turnkey — from model selection to deployment and monitoring. The result includes:

• Model (or API integration) with performance benchmarks (latency p99, throughput).
• Pipeline documentation (prompt engineering guide, model card, dependency versions).
• Integration with your backend (REST/gRPC, queues).
• Configured monitoring (dashboards, alerts for quality drift).
• Training workshop for the team (2–4 hours).
• Warranty support for 3 months after launch — as part of our quality certificate.

We have completed 30+ projects in generative AI — this gives us the right to guarantee results.

How Is the Generative AI Development Process Structured?

1. Analysis (1–2 days): audit of current architecture, clarification of use case, selection of models and success metrics. We evaluate the project free of charge.
2. Proof of Concept (1–3 weeks): quick prototype on your data — to see real quality, not blog demos.
3. Design (1–2 weeks): pipeline architecture, infrastructure (GPU cluster/API), A/B testing plan.
4. Implementation and fine-tuning (4–12 weeks): development, LoRA/full fine-tuning, integration with queue and cache.
5. Testing (1–2 weeks): load tests, metric validation, edge-case verification (negative scenarios).
6. Deployment and monitoring (1–2 weeks): production deployment, monitoring setup, documentation.

Timeline Estimates

Integration of a ready API (DALL·E 3, Midjourney API, Stability API) — 1–2 weeks. Self-hosted pipeline with fine-tuning — 6–12 weeks. Full platform with UI, queues and monitoring — 3–6 months. The specific cost is calculated individually after analyzing your scenario.

Contact us — order a consultation, and we will select the optimal architecture for your project. Get a preliminary cost and timeline estimate for free.