AI is becoming an early stop in some homebuyers’ research, but real-estate firms are finding that trying the technology is much easier than making it work at scale. A new report puts useful numbers on both sides of that gap, from mortgage calculations to staff training and human oversight.
Watch Desk analysis
What happened
PYMNTS reports that Bank of America found 20% of prospective buyers and current homeowners used AI tools or chatbots for homebuying research in the past year. The share rises to 32% among Gen Z and 28% among millennials. Among AI users, 57% used it to estimate affordability, mortgage payments or closing costs; 55% used it to learn about the process, and 52% to research neighbourhoods, market trends or property values.
The report also says 55% of prospective buyers prefer a human for home tours, while 54% want human help with legal or contractual advice. On the business side, JLL research cited by PYMNTS finds 90% of real-estate companies are testing AI, but only 5% say they are meeting their AI goals. More than 60% are not ready to scale beyond pilots. Just 33% of workers feel adequately trained in AI, and companies with a C-suite sponsor for their technology programme are three times more likely to have a successful one.
Mortgage lenders are also getting governance frameworks. PYMNTS, citing the Urban Institute, says Fannie Mae and Freddie Mac introduced frameworks in early 2026, while MISMO released a voluntary toolkit in June. The frameworks call for accountability, monitoring and human involvement. The Urban Institute says they do not yet require lenders to demonstrate outcomes such as more people becoming, and remaining, homeowners.
Why it matters
Homebuyers are using AI for practical questions about money and place, not just asking it to write a cheerful paragraph about moving house. But the figures also show where people still want human judgement: viewing a home and dealing with legal or contractual advice.
For property and mortgage businesses, high adoption counts are not the same as results. Training, oversight and evidence that a tool improves outcomes matter, especially when AI touches decisions about access to housing or credit. The reported frameworks offer governance expectations, but voluntary standards and pilot projects do not by themselves show that borrowers are better served.
Our read
The interesting divide is not “AI or no AI”. It is between using AI to help with bounded research and treating an experiment as a proven service. Buyers may find a chatbot useful for working through costs or neighbourhoods; lenders and property firms still have to show that their systems are reliable, properly overseen and useful to the people on the receiving end.
A human in the loop is a start, not a magic spell. The tougher test is whether the arrangement produces fairer, clearer decisions, and whether anyone measures that.
What to watch
- Whether Fannie Mae, Freddie Mac or MISMO publish further details on monitoring and accountability.
- Whether property firms move beyond pilots and report measurable outcomes.
- Whether lenders show that AI improves borrower outcomes, not just processing speed.
Discussion spark: Should AI be allowed to shape homebuying and mortgage decisions before firms can show it improves outcomes, or should that evidence be a condition of using it?
Sources and evidence
- Source update (16 November 2026, 09:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.