Who holds the admin logins to the systems that run your building, and could you change every one of them this week? If the answer to the first question is a list of vendors, some of whom no longer work on the property, the answer to the second is probably no. That matters to an asset manager because the person holding the login decides how fast a problem can be stopped, and how fast it gets stopped decides what it costs.
Two developments make the question more urgent. People trying to break into systems are finding weaknesses faster. And software agents, programs that act on behalf of a vendor, a manager, or an owner, are now far enough along that NIST, the federal standards body, has started writing rules for them. NIST wrote that sound identity and access management is a precursor to implementing agents at scale, and published an initial set of design tenets for how agents should be identified, authenticated, and authorized. In plain terms, before software is allowed to act for you, you need to know what it is, confirm it is what it says it is, and limit what it can do.
Let's demystify where the risk sits in a building and what it is worth to the asset.
How long you have after a flaw becomes known
The number an owner should care about is the time between a flaw becoming known and someone using it. A recent vulnerability in PHP, a programming language behind a large share of the web, showed how short that time can get. According to radical✦, the flaw was discovered by an LLM and affects potentially 70% of all websites. The newsletter calls the time defenders have to patch before attackers move the "Defender's Window."
The PHP flaw affected websites, but building automation, access control, and camera systems run on software too. If software can find flaws faster, attackers using it can find them faster in any system. Whether your building systems get patched in time depends on who maintains them. If a vendor holds the only admin login, you may not know when a system was last updated, or whether it ever was. You want that answer before an incident, because during one you won't have time to go looking for it.
Every vendor system has its own way in
Think about how systems arrive at a typical commercial property. The access-control installer brings a network for the door controllers. The HVAC contractor sets up the building automation system with its own login. The camera vendor adds a recorder and a remote support connection so its technicians can troubleshoot from their office. Each installer creates an admin account, keeps the credentials, and goes on to the next job. The owner ends up with working systems and very little record of who can reach them.
Each remote connection is a way into the property from outside. Some are well protected, some may still use the password set on install day, and some may belong to a technician who left the vendor years ago. The ownership side can't list them, because the ownership side didn't create them.
What sits behind each connection matters just as much. When systems share one network, or sit on networks that were never separated, someone who gets into one system can often reach the others. An intruder who comes in through a thermostat controller may find the door controllers on the same network segment. Putting each system on its own segment would keep a problem in one system from reaching the rest.
Holding the credentials yourself also changes how an incident goes. The first step after a compromise is to cut off access. An owner with the admin logins can do that directly. An owner without them has to wait on someone else's schedule for every system involved.
What agents add
When a person logs into your building automation system, there is a name attached, a time of day, and usually a reason, and the person works at human speed. An AI agent acting for a vendor or a management company logs in with whatever credentials it was given. It can do everything those credentials allow, many times a minute, without anyone watching each step. Give it a vendor's shared admin account with full rights, and it has full rights.
That is why NIST puts identity and access first. An agent should have an identity of its own, separate from the person who started it. It should prove that identity each time it connects. And it should have written limits: which systems it may read, which settings it may change, and which actions need a person to approve them.
As I put it on PPP Podcast Ep. 50: "Before you ask what AI can tell you, you have to know what information it is seeing, where that information came from and whether you can trust it." The same test applies to what an agent may do. If you haven't set its permissions, whoever configured it set them, and broad access is the easiest thing to configure.
Agents could do useful work in a building, such as adjusting setpoints, flagging equipment drift, and routing work orders. They can do that work safely once their identities and limits are in place.
What containment is worth to the asset
Separated networks and owner-held credentials keep a failure in one system from spreading to the others, and that difference shows up on the operating statement. A camera system cut off within minutes is a small line item. A compromise that reaches comfort systems and door controllers brings emergency vendor hours, staff time spent running the building by hand, tenant complaints that put renewals at risk, and possibly a claim.
Each of those costs comes out of NOI, and NOI is what the asset is valued on. Value is roughly NOI divided by the cap rate, so at a 6 percent cap rate, a dollar of lost NOI takes about seventeen dollars off the asset's value. The same lost NOI tightens debt service coverage, which matters most when the loan is coming up for refinance.
Good records have value of their own. Underwriters reward owners who can show how risk is controlled. An owner who can hand over a current list of every system, every credential holder, and every remote connection, along with how each system is separated from the rest, brings a stronger renewal package than one who has to collect that information from six vendors. The same list answers the question a buyer's diligence team will ask at sale: who can reach the building's systems? A buyer who can't get that answer can lower the offer to cover the uncertainty.
The work, in order, through the PPP 5C™ plan
Clarify comes first, through a PPP Review. We would list every system on the property, who holds each admin credential, and every way a vendor or agent can reach it from outside.
Connect would then move those systems onto networks the owner controls, with each system on its own segment. Our approach is BoT® (Building of Things®), which consolidates building connectivity onto a single, secure, segmented foundation that belongs to the owner, and new systems join that foundation when they are installed.
Collect captures and normalizes the building's data into a consistent model the owner keeps. For access, that would include records of who logged in, when, and what they changed.
Coordinate governs identity, access, privacy, and rules of use. This is where each agent would get its own identity, its own proof, and its own limits, set by the owner and written into vendor contracts alongside export rights and admin credentials.
Control lets any decision engine or agent act only under owner permissions. That happens in the Property Brain™, a vendor- and LLM-agnostic Property Intelligence Layer: a governed data plane and trust plane that sits under the owner's control rather than inside any one vendor's platform.
If you don't own your data & digital infrastructure, your vendors do.
With agents in the picture, the same holds for what software is allowed to do in your building: whoever holds the credentials sets the limits.
Start with one property
Pick one asset this month. List every system on it, who holds the admin login for each, and every way a vendor can reach it from outside the building. Mark which of those connections you would keep, and write down the rules an agent would have to follow before it could change anything. If you would like help, we will run a PPP Review on that property with you and hand you the full list: the logins you don't hold, the connections nobody can explain, and the order to fix them in.
