company

McKinsey just sized the future of property management. The home version doesn't exist yet.

By Holm TeamMay 21, 2026

Every so often, a piece of analysis comes along that makes a category visible. McKinsey published one this month on the future of agentic AI in real estate, and the punchline is worth reading carefully if you care about how houses, buildings, and homes actually get run.

The argument: the next wave of AI value in real estate does not come from launching individual tools. It comes from redesigning entire domains (maintenance, leasing, asset management, construction) so that software is allowed to actually do the work, with humans in the loop on the moments that require judgment. McKinsey sizes the global opportunity at $430 to $550 billion in annual value across real estate, construction, and development.

If you operate a 12-story apartment building, that framing is a roadmap. If you own an 1,800-square-foot house, you read it differently. Because almost everything in the piece assumes a thing your house doesn't have: a property manager.

The 6:12 a.m. leak

McKinsey opens with a great example. A pipe starts leaking in a 12th-floor apartment at 6:12 a.m. In one version of the future, the resident calls in panic at 9 a.m., the property manager scrambles, and the damage is already done. In the other version, a sensor flags the leak, an AI agent identifies the source, alerts maintenance staff, opens the right doors via a smart lock, dispatches a plumber, and drafts the resident notice. By the time the property manager arrives, the work orders are in motion.

This is a clean vision of agentic AI for a building. Now picture the same leak, in your house, at 6:12 a.m.

There is no sensor. There is no agent. There is no maintenance staff. There is no automated plumber dispatch. There is, at 9 a.m., you, standing in the kitchen in pajamas, trying to remember if the guy who fixed the dishwasher in 2022 was the one with the green truck or the one whose wife had a baby. The contractor list is partly in your iPhone, partly in your spouse's email, partly on a business card that has been in a kitchen drawer for two years.

The McKinsey framing is correct. It just happens to describe a future for commercial buildings that the consumer side has not even started to build.

Meanwhile, in YC's Spring 2026 batch

And in case there is any doubt the commercial side is moving on this: three months after McKinsey published, Y Combinator funded a company that reads like a literal version of the leak example. AquaShield, founded by two robotics researchers out of MIT and Harvard, sells non-invasive sensors plus ML that find leaks inside pipe networks before anyone hears a drip. Per their own YC launch, they claim a 70% reduction in water damage for Europe's largest real estate operator, with $100,000-plus remediation jobs turning into sub-$1,000 targeted fixes. They are already deployed across a residential portfolio above 500,000 units.

McKinsey wrote the white paper. YC just wrote the check.

The interesting thing about AquaShield is who they don't sell to. Their customer is a real estate operator with a procurement department, a multi-building footprint, and the math to justify sensors at portfolio scale. Their customer is not, and won't be, you. There are roughly 85 million owner-occupied homes in the United States. The unit economics that make AquaShield brilliant across a portfolio of 10,000 doors make zero sense across a portfolio of one.

That's the gap. The agentic-AI-for-real-estate thesis is real, it's funded, and it's already deploying. It just stops at the apartment building. AquaShield's customer has a procurement department. Holm's customer was, three paragraphs ago, standing in the kitchen in pajamas, trying to remember which plumber to call.

The four domains, translated

McKinsey identifies four high-value domains where agentic AI can be redesigned end to end: maintenance and facilities, leasing and renewals, investing and asset management, and construction and capital expenditures. Translate each one to a single owner-occupied house and you get an interesting map.

Maintenance. The homeowner version of "from dispatch to done" is "from leak to dry, without losing the rest of your Tuesday." Same workflow McKinsey sketches: signal, triage, access, dispatch, updates, approvals, closeout, learning. None of the supporting cast. (And yes, AquaShield is going to nail the signal step for buildings. You will still be the one who answers the door for the plumber.)

Vendor relationships. Real estate operators have "leasing." Homeowners have "the people who have been inside my house." HVAC technicians, plumbers, the painter, the electrician, the landscaper, the inspector, the chimney person, the appliance repair person who showed up once and disappeared. These are recurring relationships that earn trust over years. None of them are systematized for the typical homeowner. The HVAC company calls you to schedule the annual tune-up. You call the plumber when you remember he exists.

Records and asset management. A REIT runs a portfolio P&L. A homeowner has, on paper, one of the largest assets they will ever own, and treats it like a series of disconnected events. What did you spend on the house last year? What is under warranty? When did you last replace the water heater? Where is the document that proves the roof is twelve years old, not eighteen? For most homeowners, the answer to all four questions is the same word: somewhere.

Major projects. When you renovate a kitchen, you briefly become the project manager of a small construction company that hasn't existed before and won't exist after. The contractor knows that. You don't. Every dropped thread, every miscommunication, every change order, every dropped invoice is on you to catch.

McKinsey describes these four domains for an industry of trained operators. They exist with equal force in every house. The difference is that the homeowner is the operator, and the operator is also working a day job.

Why the McKinsey insight matters more, not less, for the home

The most important sentence in the McKinsey piece is the one about domains versus use cases. They write that "AI use cases tend to be small, bounded tasks that are frequently too narrow to change outcomes." A smart thermostat is a use case. A doorbell camera is a use case. A leak sensor is a use case. Each is fine on its own and changes almost nothing about how you run your house.

What changes how you run your house is the same thing McKinsey says changes how you run a building: a coherent slice of work with a clear owner, a measurable outcome, and a set of connected steps that get redesigned end to end. For a homeowner, the owner is you, the outcome is a home that doesn't surprise you, and the workflow is everything that happens between "something needs to be done" and "it is done, paid for, documented, and the next person knows about it."

This is what we are building Holm to be. Not another smart device. The connective layer that links the leak signal to the plumber you already trust to the warranty you forgot you had to the photo of the shutoff valve you took on move-in day to the invoice you can find again in 2031.

Workflow and feeling

McKinsey makes a point worth keeping in front of you. Quoting their piece: "Real estate is both a workflow and a feelings business. The goal is to automate the friction around the interaction so humans can focus on making sure the brand shows up with more consistency in the moments that matter."

The home version of that sentence is the whole reason we are building Holm. A house is the most workflow-heavy thing you own, and the most feelings-heavy thing you own. The workflow is the friction. The feeling is the home. Automate the friction so the feeling has room to exist. Nobody wants their house to feel like an enterprise software product. Nobody wants to be the property manager of their own life.

The McKinsey piece is, at its core, an argument that the commercial side of real estate is about to be redesigned around that principle, at a scale of $430 to $550 billion in annual value. AquaShield is the first piece of evidence that the bet is rational. The next decade will produce many more like them, all pointed at portfolios. We agree with the analysis. We just notice that on the consumer side, the operator is missing, and we are building the operator.