430 vendors for 30,000 agencies
French proptech is overflowing with AI solutions. The agency market is concentrated and slow to change tools. That gap explains a great deal.
French proptech is overflowing with AI solutions. The agency market is concentrated and slow to change tools. That gap explains a great deal.
Two figures worth setting together. A 2025 Xerfi study counts close to 430 proptech companies offering artificial intelligence solutions in France, across the whole real estate value chain[1]. Against that, France has on the order of 30,000 estate agencies[2].
Roughly one vendor per seventy agencies. In a market where half the branches have fewer than ten people, that ratio is untenable for most of the vendors involved.
A crowded, fragmented offer, where each vendor addresses a narrow slice of agency work. One does valuations, another predictive canvassing, a third the website chatbot, a fourth document analysis.
From the branch manager's chair, the landscape is illegible. They are not comparing three equivalent offers, they are looking at forty products that do different things, none of which covers their week, each of which wants a subscription, a learning curve, and one more tab.
The rational reflex in that situation is to do nothing. And that is largely what the adoption numbers show: 14 percent of real estate businesses declared using an AI technology in 2024[3].
The sector is also highly concentrated on the network side. Orpi exceeds a thousand branches, Century 21 France approaches a thousand, Laforêt counts several hundred, and the Arche group has consolidated a set on the order of 2,800 branches[2].
A vendor that signs a network signs hundreds of agencies at once. The consequence is that commercial effort concentrates on network head offices, and the independent seven-person agency is approached by much smaller players, or not approached at all.
This landscape leaves an empty space, and it is fairly precisely identifiable.
On one side, vertical products that do one thing for everybody, and therefore have to suit the average. On the other, transformation projects sold to network head offices, out of reach for an independent agency on budget and on decision cycle alike.
Between the two, nobody comes to look at how one particular agency works in order to connect what is missing there. This is not a market oversight, it is a consequence of its economics: that work does not replicate identically from one agency to the next, so it does not sell as a product.
An agency waiting for a product to cover its week will wait a long time, because no vendor's economics allow it. An agency stacking vertical subscriptions ends up with four partial tools and a new administrative burden.
The third route is to keep the agency software as the reference, identify what actually costs hours in its own way of working, and have what is missing connected around it. Harder to buy than a subscription. It is the only approach that produces a result resembling the agency rather than the market average.
We are not a software vendor and we do not intend to become one. We take no particular pride in that, it is simply the part of the market where the work was not being done. A product has to suit five thousand agencies. An installation has to suit one, and that is precisely what makes it useful.