AI doesn't give you the price of an item, but it can help estimate it.

Most of the time, this estimate is taken seriously, but unfortunately, it often leads to poor marketing management and misunderstandings down the line. Indeed, estimating a property's value isn't about guessing a price, even with a sophisticated algorithm.

I recently encountered two very concrete situations. In the first, I used artificial intelligence as an estimation tool. In just under two hours, I was able to produce a clear, structured, and well-reasoned analysis. In the second, without using AI, it took me a full day to reach virtually the same conclusion. This comparison is revealing: AI doesn't replace expertise; it simply saves time.

Many of the quick valuation tools offered by real estate portals today rely on methods similar to artificial intelligence. They analyze databases, cross-reference past sales, surface areas, and locations, and generate an average price. These tools have real value: they provide an initial benchmark and allow you to situate a property within the market. The problem arises when this benchmark is mistaken for a definitive value.

A property is more than just an address and a few criteria. An algorithm doesn't visit a home. It doesn't see the quality of the construction, the actual condition of the property, the work needed, or the potential after renovation. It doesn't perceive the natural light, the noise, or the atmosphere of a street. It also can't analyze a condominium, understand its management, or anticipate the impact of a poor energy performance certificate (EPC) in negotiations.

In the current context, this nuance is essential. The market has become more difficult to read. Buyers know they have opportunities, compare prices more, and negotiate more than before. Price has become a central, sometimes decisive, factor.

An overly rough estimate can block a sale for months or lead to a gradual devaluation of the property.

When I use AI, I don't ask it to set a price for me. I provide it with on-the-ground analysis: what I observed during the site visit, the building's technical specifications, truly relevant comparable properties, the current market conditions in the area, the local market dynamics, and the property's strengths and weaknesses. I then feed the AI, which helps me structure this information, making it more readable and informative. But the final decision remains mine.

Quick online valuations and artificial intelligence are therefore not the problem. They only become problematic when sellers or professionals take them as definitive and absolute. For me, they are primarily helpful tools, and certainly not set-in-stone results.

Artificial intelligence cannot predict the price of a property. It can help estimate it, provided it is guided by experience, expertise, and local knowledge. In a more demanding and uncertain market, valuation is no longer a simple number: it is an essential step involving expertise, experience, methods, and on-the-ground experience, and it carries significant implications.

Opinion piece by Nicolas Jeff, Real Estate Agent (LinkedIn).