Where is AI actually used in real estate today?
Across the transaction, in bands of maturity. Valuation is the oldest: automated models price properties from comparables, market data and property features, and every major portal runs one. The front office came next: chat and voice agents answer listing inquiries, book viewings and qualify leads, the property version of the AI receptionist. Content tools draft listings and stage rooms virtually. And the paperwork layer, leases, applications, invoices, is standard document AI work. The adopted uses all share one shape: high-volume routine work with a person still owning the decision.
What does each use actually require?
Data before tools, in most cases. A valuation model is only as good as the transaction and property data behind it; a lead-qualifying chatbot needs current listings, availability and pricing wired in, or it becomes a polite dead end; maintenance prediction needs equipment and ticket history. The tools themselves are rentable, which is why agent-facing products dominate the search results. The differentiator is not access to AI but the state of your records, and the firms that struggle are the ones automating on top of spreadsheets no two of which agree.
What should a buyer be skeptical about?
Vendor percentages and silent failure. Claimed savings on repairs or conversion rates are rarely measured the same way twice; pilot one building or one office and count yourself. Watch prediction claims that lack the years of history machine learning needs, assistants that answer confidently beyond their data, and anything touching fair-housing-sensitive decisions, where screening and pricing models carry real discrimination risk and regulatory exposure. Let AI do the volume and keep people on the judgment, and the category pays for itself one saved hour at a time.