Real estate is a sector in which decisions often rest on two axes at once: a financial valuation and a relationship with a tenant, buyer, investor or resident. That combination makes AI governance in this sector different from sectors where it is purely about process or production. A valuation model that estimates a value too high or too low affects a transaction. A chatbot that gives a tenant an incorrect commitment about maintenance or a rent increase affects a relationship and possibly a legal obligation. The ratio between operational AI applications and applications that communicate directly with customers, tenants or investors is generally more skewed in real estate than in sectors with fewer external contact moments. There is often less production AI and more AI that offers text, advice or a figure to a third party.
On top of that, real estate organizations often work with a network of parties: brokers, managers, contractors, legal advisors. Each of those parties can deploy AI themselves within the work being done for your organization, without you having any visibility into it. Governance does not stop at your own employees.
AI in this sector concentrates around a number of recognizable tasks: drafting valuation reports or valuation justifications, summarizing rental contracts and legal documents, answering questions from tenants or buyers, generating property descriptions and marketing texts, and analyzing market data for investment decisions. Some of these applications are visible and approved. Others arose because an appraiser, a leasing broker or a customer service employee found a tool that saved time and kept using it.
The IT list of approved applications rarely describes in practice what is actually happening. An appraiser who uses an AI tool to draft a valuation report, an employee who has contracts summarized by a free chat service: that use arises at the desk, not in a procurement process. Similar patterns occur in other sectors with a lot of customer or client contact, as described in how AI governance looks in the installation sector and in how AI governance looks in wholesale.
Governance starts with making distinctions. An AI application that organizes internal reports requires different treatment than an application that presents a valuation figure to a customer or automatically responds to a tenant complaint. The risk level depends on who uses the outcome, whether a figure or advice is presented directly to a third party, and whether it carries a financial or legal consequence. An AI tool that only edits photo material for a property listing sits elsewhere on that scale than a tool that helps determine a valuation figure.
This classification is not a one-time exercise. New tools arrive as soon as someone finds them useful, and the risk assessment of an existing application can change as usage shifts, for example when a tool that started as an aid for internal draft text is gradually used to communicate directly with a tenant.
The core of the problem is rarely a lack of rules, but a lack of visibility. Anyone who wants to know what is actually being used must ask the people who work with it every day. That only works if the question is asked without consequence. An appraiser who admits to using a non-approved tool for draft reports should not feel that he is thereby confessing to a violation. As soon as that feeling exists, the usage disappears from view rather than from practice, and the risk continues to exist unchanged, only invisible. How this questioning can be set up in practice is described in how you prevent employees from using a tool nobody knows about and in how you prevent company data from ending up in a free chat window, two risks that often coincide in real estate because contract data and valuation figures are sensitive.
The similarities with other sectors with many external parties and document flows, as described in how AI governance looks in construction and in the healthcare sector at how AI governance looks in healthcare, are greater than the differences: everywhere it holds that usage first settles at the edge of the organization, before anyone gains visibility into it.
A governance approach for real estate maps out which AI applications exist, who uses them, which risk level applies, and what oversight is needed: who checks an outcome before it influences a valuation, a tenant letter or an investment recommendation. This connects to existing risk structures within the organization, without placing a separate AI regime alongside existing risk management. The specific rule content, with the precise obligations that apply to specific applications, is covered elsewhere; here it concerns the mechanism through which you gain visibility and can demonstrate what is going on.
Once it is clear which AI applications are running in your real estate organization and what risk level applies to them, another question follows: how much of the work that appraisers, managers and customer service employees do every day is actually suitable to hand over to AI, and which part is not. The work scan from FTE TO AI calculates this per task, so that governance does not stop at risk alone, but also makes visible where AI actually frees up capacity.
Vraag maar. Governance begint bij weten wat er draait — ook wat niemand heeft goedgekeurd.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.