Property guidesevimibul
Product guide

How evimibul.org Works — Technical Deep Dive

How we aggregate 133K+ listings, how the AI finds your home, how price scores work. A peek behind the curtain.

Published March 24, 20264 min read
On this page
  1. Data Collection: 133K+ Listings from 3 Sources
  2. AI Chat: Find Homes by Talking
  3. Price Scoring
  4. Nearby Places (POI)
  5. Comparison: 10 Metrics Side by Side
  6. Tech Stack

How evimibul.org Works — Technical Deep Dive

evimibul.org is an AI-powered real estate search engine covering all 81 provinces of Turkey. We aggregate 133,000+ listings, analyze them, and find the best matches in seconds. Here's what happens behind the scenes.


Data Collection: 133K+ Listings from 3 Sources

SourceListingsMethod
Emlakjet~127,000HTTP scraping
Hepsiemlak~900HTTP scraping
Sahibinden~5,600API

For each listing, we extract: title, price, rooms, square meters, location, coordinates, up to 20 photos, agent info, building age, and more.

Data is refreshed daily. Removed listings are automatically cleaned up via Vercel Cron jobs.


AI Chat: Find Homes by Talking

The core of evimibul.org is the AI chat system. You describe what you want in natural language (voice or text), and the AI extracts structured search filters.

How It Works

  1. You say: "I'm looking for a 2+1 rental in Kadikoy, max 20K"
  2. AI extracts: city=Istanbul, district=Kadikoy, rooms=2+1, type=RENT, max_price=20000
  3. Location verified: Google Maps API confirms the coordinates
  4. Boundary applied: Kadikoy's actual administrative borders (bounding box) are used
  5. Results returned: Matching listings from PostgreSQL, with price scores

Location Precision

Understanding "Batikent" or "Konyaalti" requires local knowledge. We solve this with a chain:

  1. AI estimates coordinates
  2. Google Maps Geocoding verifies
  3. Turkey boundary data (81 provinces, 1,013 districts, 12,895 neighborhoods) provides exact borders
  4. Search happens within these borders

Price Scoring

Every listing gets an automatic price-performance score:

  1. Calculate price per m² for the listing
  2. Find the median price/m² for similar listings nearby (same type, similar rooms)
  3. Compare
ScoreMeaningDifference
DealWell below market>20% cheaper
AffordableBelow market5-20% cheaper
MarketNormal price±5%
OverpricedAbove market>5% more

Nearby Places (POI)

For every listing with coordinates, we show nearest:

  • Schools (primary, middle, high) — 5km radius
  • Hospitals — 8km radius
  • Transit (bus, metro) — 5km radius
  • Markets/Grocery — 3km radius
  • Malls — 8km radius
  • Pharmacies — 3km radius

All from OpenStreetMap's Overpass API — completely free and unlimited.


Comparison: 10 Metrics Side by Side

Compare up to 4 listings with weighted scoring:

Property Quality (70%): Price affordability (25%), price/m² (20%), area (15%), rooms (10%)

Location (30%): School (8%), transit (8%), market (5%), hospital (4%), pharmacy (3%), mall (2%)


Tech Stack

LayerTechMonthly Cost
FrontendNext.js 16 + React 19$0
DatabasePostgreSQL (Neon)$19
CacheRedis (Upstash)$0
AIClaude API (Anthropic)~$16
MapsOpenStreetMap$0
GeocodingGoogle Maps$0 (10K/mo free)
HostingVercel Pro$20
EmailResend$0
Total~$56/mo

133K+ listings, AI search, 10 languages, 81 provinces — for $56/month.


Try it yourself at evimibul.org — tell the AI what you're looking for. It's free, no ads, no tricks.

Happy house hunting!

ShareWhatsAppXLinkedIn

Was this guide helpful?

Start your home search

Comments (0)

Sign in to comment.

No comments yet. Share your experience.