CBRE released its 2026 Office Occupier Sentiment Survey on July 30, and two of its numbers sit badly together. Eighty nine percent of the companies surveyed now ask for three days a week in the office. A year earlier it was 78 percent. Only 14 percent are making major improvements to their space.

So your tenants are pulling their people back in, and most of them are doing it without a renovation budget. John Morris, who leads Advisory Leasing for the Americas at CBRE, framed the stakes well. "The market has shifted from whether the office matters to whether it delivers." Delivery happens in the daily contact between your team and the people in the suites.

AI is what many buildings are hoping gives their office that edge. A lot of what gets pitched for tenant engagement aims at output, meaning more announcements, faster newsletters, and better event copy. The survey data points somewhere else. AI pays off in tenant engagement when it shortens the trip from a tenant's question to a real answer. That means the place to start is the request already sitting in your inbox.

Attendance Rose Faster Than Workplace Investment

Sixty two percent of respondents call enhancing employee experience a priority. Yet 47 percent of the 97 companies surveyed rate their own workplace experience as average or below average against their peers. Your tenant contact carries a mandate and a thin budget at once.

That gap is good news for a landlord who can help. When a tenant can't buy its way to a better office, service quality becomes the main method of closing the gap. And the tenant's own technology plans are raising the bar on what your building has to deliver. Nearly a quarter of firms already see AI changing how they plan space. Another 30 percent expect that within two years. Among the changes they expect, 36 percent point to a need for better amenities. Julie Whelan, CBRE's Head of Occupier Research, called AI "a catalyst for office redesign." That lands on office landlords and on anyone building a case for tenant retention.

Tenant Relationship Management Is Where AI
Has a Record

Deloitte surveyed more than 850 senior leaders at real estate owners and investment firms. Every firm held at least $250 million in assets under management, or AUM. The survey asked where AI was producing results and where it was producing headaches. The top area for positive results was tenant relationship management, at 41 percent. The top area for trouble was property operations and management, at 42 percent.

Read those two together and you get a clear steer. The tenant conversation is where this technology has a track record. And the plant room is where it keeps stalling. The same survey shows how fast expectations have reset. The share of firms reporting a transformative impact from AI fell from 12 percent to 1 percent in one year. Reports of trouble with rollout climbed from 16 percent to 27 percent. So the sober read on ai in property management is simple. It delivers in narrow lanes, and the tenant relationship is the lane with the best evidence.

There's a catch worth naming here. Almost every tenant question is an operations question underneath, like the suite that's warm on the north side every August. So your front end can only answer as well as the operating record behind it allows. That's why those two Deloitte findings belong together. When AI sits on a unified property management system that holds both the tenant experience and the building operations, it can reach the work order history that turns a question into a real answer. And that pairing is where you'll get the most out of it.

Start With the Question a Tenant Already Asked

Every building has a front door for tenant contact. It's the service request, the email to the property manager, the text message to the chief engineer. Start there, because a tenant remembers the exchange they started far better than the one you started. Four jobs sit at that front door and none of them need a person:

• Reading an inbound request and sorting it to the right trade
• Answering on the spot when the answer already exists in your records
• Confirming receipt with a clear next step and a time
• Drafting the reply for the requests that do need your judgment

Propmodo made the point well on August 11, in a piece on why building software is turning to chat. Nic Halverson, the chief executive of Occuspace, said facility teams want answers rather than data. He argued that chat is replacing the dashboard because it removes a step. The same holds one level out, with your tenants. Somebody in suite 1200 who asks why the north side is warm wants a sentence back. So tenant communication tools that give a sentence will beat a portal that gives a form.

Judgment Stays With Your Team

Halverson was candid about the limit. "AI is smarter than all of us but we have more domain specific knowledge," he said. "Now we have to think about how we can get that knowledge into AI." The gap is context. Software can see that suite 1200 filed four requests this month. It can't see that the contact there is under pressure before a board visit on Friday. And it can't see that the fourth request was really a plea for calm.

So draw the line on purpose. Software handles lookups, sorting, drafting, summaries, and scheduling. Your team handles promises, exceptions, money, and anything that touches the lease. Two rules keep that line in place. A fast wrong answer costs more than a slow right one. So nothing goes out unchecked if it commits you to a date or a dollar.

Proof That It's Working

Four numbers will tell you whether any of this is landing. Track median first response time on inbound requests. Track the share of requests closed in a single exchange. Track your reopen rate, because a request that comes back was never really answered. And track requests per thousand square feet by suite. That's the number that flags a quiet tenant before the quiet turns into a notice. Those four make a workable key performance indicator set. And they map onto whatever service level agreement you've promised.

The renewal meeting is where the work pays. Ask your system for a short monthly summary per tenant. It should cover what they asked for, what you fixed, and how long it took. Read it before you walk in. That summary turns a rent talk into a service record. And it gives you an early read on tenant rollover risk while there's still time to act. Teams that already use analytics to steer building operations will know the pattern. The reporting is only as good as the record underneath it.

Where the Work Lands

AI answers from the records it can reach, so what you connect decides what it can say. Say a tenant asks whether their 9 a.m. guest has arrived and when the warm north side gets looked at. Your visitor management log holds one answer and your work order history holds the other. In one tool, AI reads across both and replies once. In separate tools, somebody on your team becomes the integration layer.

That's the case for a single tenant experience platform over a stack of point tools. It's also how Cove is built. Service requests, announcements, bookings, visitor management, and building operations sit in one system. So your tenant engagement software and your operating record are the same record. Point ai for property management at that and it has something real to work from. And your tenants get a front door that answers them instead of one that guesses.

FAQ Recap

What is the best use of AI for tenant engagement?

Put it on the inbound path first, meaning the service request and the message a tenant sends you. That's where AI has the strongest record in commercial real estate. Deloitte's 2026 survey found tenant relationship management to be the top area where owners report positive results. Sorting, instant replies, and drafted answers give you a fast win. And your team keeps deciding the hard parts.

Can AI replace a tenant experience app?

No, and framing it that way misses the point. The value comes from what the software can reach rather than the interface a tenant taps. So a chat layer over separate systems will answer confidently and wrongly. Connect your requests, work orders, and asset history first. Then add the chat front end.

Is AI in property management actually delivering results yet?

In narrow places, yes, and the honest picture is more modest than it was a year ago. Deloitte's survey shows the share of firms reporting a transformative impact from AI falling from 12 percent to 1 percent in a year. Property operations and management ranked as the hardest area to roll out. Tenant relationship management is the exception. So it's the sensible place to start.

How do I measure tenant engagement?

Measure response and resolution rather than activity. Track median first response time, share of requests closed in one exchange, reopen rate, and requests per thousand square feet by suite. Those tell you far more than open rates on an announcement. Bring those four to your owner and to your renewal meetings. They describe service you actually delivered.