Real estate underwriting has been done roughly the same way for 30 years. An analyst collects data, builds a spreadsheet, models assumptions, and writes a report. The spreadsheet is usually enormous. The data collection is manual. The model is often a black box that only its creator fully understands.
AI is breaking all three of those parts at once, and the industry is just starting to figure out what that means.
The data problem, first
The biggest bottleneck in underwriting has never really been the modelling. It's been the research. Finding comparable sales, rental rates, vacancy trends, construction costs, and demographic shifts that work is slow, repetitive, and expensive when you're paying an analyst to do it.
The new generation of real estate AI tools (FeasibilityPro.ai, Cherre, HelloData.ai, others) is building what you might call "intelligence layers" unified data stacks that pull from proprietary research, third-party market sources, and live feeds and make that data queryable in natural language.
Instead of an analyst spending half a day pulling comps, the AI surfaces them in seconds, with citations you can trace back to the source. The analyst's job shifts from data retrieval to data interpretation.
That's a meaningful shift.
What's actually changed in the modelling layer
The financial models themselves are getting smarter too, but in a specific way, they're becoming auditable rather than just automated.
The old problem with AI-generated financial models was trust. You'd get a number but have no idea how the system arrived at it. That's fine for a rough estimate, but no lender or investment committee is going to approve a deal based on a model they can't interrogate.
The better tools now are generating models where every cell is cited. You can click a value and see exactly what formula drives it, what assumption it's based on, and where that assumption came from. FeasibilityPro.ai does this through AI-labeled cell references and calculation chain tracing inside Excel.
That's the thing that makes AI models actually usable in professional contexts, not the speed, but the auditability.
The Excel situation (nobody wants to talk about this, but it matters)
Every few years, someone declares that Excel is dying and will be replaced by some purpose-built platform. It never happens. Real estate development firms run on Excel. Investment committees review Excel. Lenders want Excel.
The AI tools that are actually gaining traction understand this. They're not trying to replace the spreadsheet; they're embedding AI inside it. Excel Add-ins, AI-assisted formula debugging, and natural-language cell navigation. The interface stays familiar, and the capability increases.
This is probably the most pragmatic design choice an AI tool in this space can make.
Where the human still matters
Worth being clear about what AI underwriting tools don't do:
They don't know about the informal planning relationship between a developer and a local council
They don't factor in a site-specific contamination issue that's not in public records
They don't understand that a particular submarket is weird because of one dominant landlord who never drops rates
They don't replace a senior analyst's gut check on whether the numbers feel right, given everything they've seen
AI tools in real estate are really good at compressing the mechanical work. They're not good at replacing the domain expertise that takes years to build. The analysts who will win are the ones who use AI to eliminate the grunt work and spend their time on the judgment calls.
The tools worth knowing in 2026
FeasibilityPro.ai:- Excel-native AI pro-forma generation + market research layer
TestFit:- site-to-yield design feasibility for developers in early site testing
Zenerate:- AI-generated floor plans and site layouts for land development
Feasibly:- multi-agent AI for bank-ready feasibility reports (higher price point, human oversight)
Cherre/Cotality:- institutional-grade real estate data platform
AIRE Software:- AI feasibility studies with market-driven financial analysis
None of these tools is doing the same thing. The market is fragmenting by use case which is actually a healthy sign. It means the tools are getting specific rather than trying to be everything.
What to watch for in the next 12 months
The next frontier is AI agents that can run multi-step underwriting workflows autonomously. Not just generate a model but update it when market conditions change, flag when an assumption has drifted from current data, and surface deal risks without being asked.
Whether that's exciting or concerning probably depends on what seat you're sitting in.
Top comments (0)