A few weeks ago, we brought together technology leaders from across commercial real estate, including teams from Brookfield Properties, Vornado Realty Trust, Paramount Group, and others, for a working session on AI: where it delivers value today, where it falls short, and what has to change for it to work at scale.
What emerged was a more useful way to evaluate the noise in the market. While firms race to experiment with copilots, chat interfaces, and automation, many are still trying to apply AI inside operating environments that were never built for it.
This article breaks down what to actually look for when evaluating AI in commercial real estate: the conditions that make it work in a real portfolio, where it tends to fall apart in practice, and why the infrastructure underneath it will matter more than any single feature.
The reality generic AI is walking into
One example from the conversation made the challenge obvious.
A single portfolio:
- 189 properties
- 17 operating partners
- 120 brokerage teams across 10 different firms
That’s not a system problem, it’s a coordination problem.
This is the environment AI is expected to operate in. So when it underdelivers, it’s rarely a model issue. It’s that the system around it is fragmented, inconsistent, and impossible to reason across.
Where AI actually falls apart
Three issues came up repeatedly in the discussion. None of them are new. All of them become critical the moment you try to scale AI.
Latency
The data you need often arrives too late to be useful. Third-party inputs lag, and by the time the information is reconciled, the decision window has already moved.
Quality
Core information is not consistently defined. Even something as basic as a lease abstract can vary depending on who created it, which introduces ambiguity into every downstream decision.
Sparsity
Most records are incomplete. Brokers leave breadcrumbs. Teams fill gaps manually. That may work for humans, but it creates weak inputs for systems expected to generate reliable outputs.
Each of these can be managed in isolation. Together, they make it difficult for AI to produce reliable, actionable output at scale.
How to evaluate AI in Commercial Real Estate today
As more solutions incorporate AI, the conversation is naturally shifting beyond feature labels and toward what actually drives meaningful value in practice.
A more practical way to evaluate AI is to look underneath the feature layer and ask a different set of questions:
- Is the data structured enough to support reliable output?
If inputs are inconsistent or incomplete, AI will only amplify those issues. - Can it handle the complexity of a real portfolio?
Many tools perform well in controlled workflows. Fewer hold up across multiple partners, systems, geographies, sizes, and definitions. - Does it create shared understanding across teams?
If leasing, asset management, and operations interpret data differently, the output will be harder to trust. - Does it improve decision-making, not just access to information?
Faster answers only matter if they lead to faster, better action. - Can it scale without adding operational overhead?
If growth still requires more manual work, you haven’t addressed the underlying issue.
This is the thought process more CRE leaders are starting to apply when thinking through their AI strategies.
Why VTS is evolving its role
VTS is shifting from a system of record to an operating partner for how commercial real estate actually runs. The challenge isn’t just access to tools; it’s coordinating data, workflows, and decisions across a fragmented environment.
VTS already sits at the center of that environment, with more than 13 billion square feet on the platform and over 45 thousand users working across leasing, asset management, and operations.
It’s not about adding more features; it’s about building a shared foundation that structures data consistently, connects workflows across teams, and enables decisions with full context.
That foundation is what determines whether AI works in practice and how quickly teams can move from data to decisions.
If you’re evaluating how AI fits into your portfolio, start there.
See how VTS makes AI in commercial real estate work across real portfolios, not just isolated workflows: https://www.vts.com/ai
Be part of what comes next
Asset Intelligence is the first step in bringing these ideas into the VTS platform.
Get started
