Conversational property search engine with vector search, multi-model LLM routing, and financial analytics.
Natural language real estate search, automated property valuations, and interactive mapping powered by RAG.
AI Real Estate Assistant is an open-source web application designed for real estate agencies and platforms. It pairs conversational natural language search with vector retrieval and financial valuation tools. You can deploy it to give clients intelligent property discovery and automated market insights.
The platform parses natural language queries like price limits or location constraints to match listings via ChromaDB vector search with MMR reranking. It integrates 6+ LLM providers to generate property valuation forecasts, neighborhood summaries, and comparative market analysis reports. It also includes financial calculators for mortgage rates, rent-versus-buy decisions, and total cost of ownership.
This tool suits real estate developers, agencies, and tech-savvy brokerages wanting to build custom AI property search portals. It is ideal if you have developer resources to host Python and Next.js applications with database backends. It is not intended for non-technical real estate agents seeking an out-of-the-box desktop app with zero server setup.
Route queries across OpenAI, Anthropic, DeepSeek, Grok, Google, or local Ollama instances with automatic fallback.
Extract filters automatically from plain text queries using hybrid vector and keyword matching with MMR reranking.
Forecast property values across 1-year, 3-year, 5-year, and 10-year horizons with confidence bands.
Display clustered property markers, area metrics, and city overview analytics using Leaflet or Mapbox.
Serve users in 9 languages while applying EU AI Act compliance labels to generated outputs.
The software is free to use under the MIT license. You are responsible for server hosting costs and third-party LLM API fees if cloud providers are used. A managed hosted version called PropVector AI is listed as coming soon.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
Yes, you can configure the backend to use a local Ollama instance for LLM inference alongside local ChromaDB storage.
It uses ChromaDB for vector search and supports PostgreSQL or SQLite for relational data storage.
It offers environment-conditional lazy provider loading to keep baseline memory under 500 MB on constrained environments.
See Zekai first in Google