Real Estate Screener

Paste a set of property listings — copy-paste from Idealista, Rightmove, Zillow, an agent's email, or your own notes. Add a short buyer profile (purpose, budget, priority) and get back a screening report: each listing classified deal / fair / overpriced / speculative / flag, with strengths, concerns, due-diligence items, a fit score, a ranking, and a primary pick tailored to what you're trying to do. Inputs and the most recent report are stored in your browser — no account, no tracking.

Listings (one per block — free format works)

Purpose

Top priority

Budget cap

Currency

Advanced (optional)

Target region / city / neighborhood

Hold horizon

Budget flexibility (%)

Min bedrooms

Property type filter

Paste at least one listing block to begin

Last screen · just now schema · real-estate-screen.v1

How it works

When you click Screen listings, your pasted text and buyer profile go to the same AI model that powers the AI Lab — Kimi K2 via NVIDIA's free inference tier. The model parses each listing block, extracts the key attributes (price, size, bedrooms, year built, condition, property type), classifies it as deal / fair / overpriced / speculative / flag, and scores how it fits your stated profile (purpose, priority, budget, horizon). It then ranks the set, picks one as the primary recommendation, lists what to shortlist vs. reject, and produces cross-listing insights (e.g. "all three are above the local average €/sqm — wait or widen the search"). Your inputs and the most recent screen are stored in your browser under aiapp.realestate.v1 — no account, no cloud storage.

The model uses its training-cutoff knowledge of city-level price norms (Milan ~€4,500–6,500/sqm center, London £8,000–15,000/sqm prime, etc.) to calibrate "fair" vs. "overpriced" — but those numbers can drift. Treat this as a screening pass, not a valuation. The due-diligence bullets on each card are the things you (or your notary / surveyor / broker) should verify before signing anything.

Honest caveat: this is a screening assistant, not real estate advice. The model can't see the property, can't pull live comps, and may have stale price norms. Verify every classification against current local listings, an on-site visit, a structural survey, and a real estate professional before making any offer. Never transfer money based solely on this output.