I've been around long enough to watch entire technology cycles play out. From the early days of automated rent setting, through the BI revolution, through COVID-era leasing transformation and now into what everyone is calling “the AI era.” And I'll be honest: the signal-to-noise ratio right now is the worst I've ever seen it.
"AI" has become a word that means everything and therefore means nothing. Talk to ten vendors at any industry conference and every single one will tell you they have AI. Some of them do. Most of them have something much closer to a chatbot wrapper on a database they haven't touched in five years, dressed up in a very convincing press release.
For operators, COOs, CTOs and asset managers making real technology decisions with real budget and real accountability, that matters. A lot.
Most of what's being marketed as AI in multifamily falls into one of three traps.
The first is the AI veneer: legacy vendors who've slapped a modern interface on old infrastructure. It appears cutting-edge in a demo. It doesn't solve your actual operational problems.
The second is general-purpose tools used without domain expertise. ChatGPT and its cousins are genuinely impressive. They're also dangerous when applied to multifamily decisions without the data foundation and domain logic to back them up. They'll give you a confident answer regardless of whether it has any grounding in your actual portfolio.
The third, and most insidious, is dirty data. In construction, engineers seek clean fill dirt, free of contaminants that could compromise a structure's foundation. In proptech, most AI is being fed the equivalent of junk fill: siloed, inconsistent, outdated data from disparate property management systems. The old rule applies more than ever: garbage in, garbage out.
If there’s one thing I've learned about our industry, it’s this: when something isn't working, it usually comes back to the same root cause...we're not asking the right questions.
So, here's a question I'd encourage every operator to ask of any AI tool they're evaluating:
Can it explain why?
Not "what" because any tool can summarize a trend. Not "how" because that's just reporting with a new interface. But why. Why is the suggested rent $2,045 instead of $2,035? Why this concession? Why does this renewal increase?
If your AI can't answer those questions in a way your team can stand behind in front of a board, an investor, or a regulator — you don't have AI that matters. You have AI theater. "The AI made me do it" is not a legal defense. It's not a professional defense either.
There's a spectrum worth understanding clearly.
At the baseline, Insights AI recaps data, identifies trends and produces forecasts. Useful, but it still requires significant human effort to act on and the risk is that it parrots your own data back to you in a way that feels like intelligence but isn't.
Above that, Conversational AI understands intent and context, letting you ask natural language questions and get workflow-aware responses. Better, but still vulnerable to poorly worded prompts and incorrect assumptions.
The destination is Agentic AI: action-centric and autonomous. It doesn't just identify a problem; it resolves it. This is where the category gets genuinely transformative. And it's also where the stakes for data quality are highest. You cannot build agentic AI on a dirty data foundation. The ball needs to be on a tee for AI to swing with consistent accuracy.
We're at that dangerous inflection point where the hype is cresting and the disillusionment hasn't fully set in. Executives are under pressure to adopt AI quickly, vendors are happy to sell them something and the gap between what's promised and what's delivered is widening by the quarter.
The operators who win the next five years won't be the ones who adopted AI first. They'll be the ones who adopted AI right: built on a clean, governed data foundation, with the transparency to defend every recommendation complete with systems capable of actually executing, not just advising.
"Beyond the Hype: AI That Matters" covers the three risks catching operators off guard, walks through the three phases of AI maturity and makes the case for why confidence comes from the data layer, not the recommendation layer. It also includes a practical checklist for evaluating any AI partner you're considering.
It's not long. It's not padded. It's the sober perspective that's hard to find when everyone has something to sell.
Download "Beyond the Hype: AI That Matters" here →