AI in real estate: how artificial intelligence is changing the market

Artificial intelligence is changing the property sector faster than many in the industry admit. Over the last 3 years, capabilities that used to require teams of analysts with access to exclusive data have become available to any investor with a smartphone. Here is the real picture of what is changing and what is staying the same for now.

Automated valuation: the most mature use

AVMs (Automated Valuation Models) were tools for big banks 10 years ago. Today, Idealista, Fotocasa and dozens of proptech companies offer them free or at low cost. The AI analyses millions of transactions, property characteristics and data on the surroundings to estimate market value with growing accuracy (an average error of 3-7% in liquid markets).

Predicting prices and trends

Predictive AI models can already identify fairly reliably which areas are entering a phase of accelerating prices before it becomes obvious in portal data. They do it by cross-referencing data on: new shop openings, building licences granted, changes in online searches, notarial transaction data and local macroeconomic variables.

Spotting investment opportunities

AI can screen thousands of auction or bank-owned assets and automatically highlight those that meet specific investment criteria: discount to market value, expected rental yield, a clean land registry record, and so on. What used to take hours of manual analysis per asset is now done in seconds.

What AI can't do (yet)

AI can't: accurately assess the real condition of a property without good-quality visual data, detect disputes within the owners' association or problems with neighbours, properly value very unusual assets with few comparables, or replace human negotiation in complex deals. These are areas where the human factor is still irreplaceable.

AI applied to your investment: RadarPro uses AI to analyse auction assets and bank-owned properties: valuation, investment score, analysis of charges and estimated yield. Get free access.