Back to all articles
Stop implementing resident apps
Written byErwin Buckers
Published onOct 14, 2026
CategoryInspiration
Read time5 min

Stop implementing resident apps

This article originally appeared in the 7th edition of Unlocked by GSL.

For fifteen years, every tenant platform on the market competed on the same thing. The interface. Whoever built the cleanest home screen, the smoothest booking flow, the notification people actually opened, won the tender.

That competition is over, and nobody won it. Interfaces kept getting better right up to the point where they became optional.

A resident with a question no longer has to find your app, remember a password and hunt for the right screen. They just ask.

Sometimes they ask your platform's own AI assistant. Ours is called Ask, and it is not confined to the app. The same assistant answers in WhatsApp, in email and by voice, so a resident uses whichever one they already have open. Sometimes they never come to you at all, and put the question to the AI assistant on their own phone, which can reach into your building only if you have given it a way in.

Either way, the interface that won your last tender is now one of several ways in, and not the one anyone reaches for when there's no hot water at ten on a Sunday evening.

So the question worth asking about your resident app is no longer what it looks like. It is what it is connected to, what it knows about the person asking, and what it is allowed to do.

An operator asked me recently which resident app they should move to. Engagement on the one they had was flat, and the assumption was that a better interface would fix it.

I asked what it was connected to. The PMS, partly. Nothing else.

A better interface would not have moved that number, because the interface was never the constraint.

None of that means the app no longer matters. It does, and the nuance is the whole point. Residents open an app on purpose for the things that are worth opening it for: what's on this week, who else is going, the gym slot, the study room, the marketplace, the neighbour with a spare bike. Browsing is how belonging starts, and it is also where the ancillary income actually gets sold. Nobody discovers a community in a chat reply.

But discovery is the smaller half of what a resident needs from a building. Most of it is errands. Is my rent confirmed? Why is my invoice higher this month? Any update on my leaking tap? Where is my parcel? Those errands don't need an interface at all, and increasingly they won't have one. They'll be handled by an AI assistant, wherever the resident already is.

The product, in other words, was always the layer underneath.

What you are actually buying

Look at where the money goes in a managed residential portfolio, and the picture is unambiguous: people are sixty to seventy percent of the management bill. Software is a rounding error next to staff time, and a large share of that time goes on something nobody put in a job description: being the integration layer.

The PMS doesn't talk to access control. Access control doesn't talk to the parcel locker. The utility platform doesn't talk to anything. So a person does it. Someone rekeys the same new tenant into four systems on move-in day. Someone else opens six tabs to answer one question about one resident, then updates two of them by hand. Every system added over the last decade made the stack richer and that work heavier. That is what being the integration layer means, and it is the stage most of the market is stuck at: software-powered, not connected.

An AI assistant inherits that problem. Drop a copilot on top of a disconnected stack and you get a very fluent guess. What makes it work is what sits underneath, and it comes down to two things.

Context. An AI assistant can only answer from what it can reach, and only if what it reaches is true today. That means breadth: every system in the building, not the three that happened to have an API. Depth: one full profile per resident, covering lease, invoices, tickets, bookings, access and community, not six partial views. And freshness: the AI assistant has to know where someone is in their tenancy. A question about the end of the contract is idle curiosity in week two. In week forty the same question is a renewal decision with a deadline attached, and the answer should say so.

Permission. An AI agent that acts needs a principal: who is asking, what they may do, and who is accountable afterwards. That is an identity model. Accounts provisioned from the tenancy rather than raised as IT tickets, scoped to a role, and revoked when the tenancy ends, across every connected account.

Without the first, an AI assistant guesses. Without the second, it can only ever suggest.

image

Student housing is the hard version of this

Every residential asset class has this problem. PBSA has it on a compressed clock.

A build-to-rent block turns over maybe a third of its residents in a year. We turn over nearly all of them, in a fortnight, in September. Every arrival needs an account, a door credential, a laundry login, a parcel profile, a utilities record. Every departure needs all of that taken away, and the "taken away" half is the one that quietly doesn't happen. Ask your team how many former residents still have a working credential somewhere in the stack. The honest answer is usually that nobody knows.

No amount of interface design fixes that. It is a provisioning problem, and it either happens automatically off the lease or it happens by hand in the worst two weeks of your year.

The upside of getting it right is equally structural. On our own modelling, an operator running five thousand homes today can run around seven thousand with the same team once the coordination work is removed, and the freed hours go to the community programme and the service quality that make people renew. The headcount stays where it is. The portfolio grows past it.

Ask a different question

So when the next platform conversation starts, park the demo of the home screen. Ask instead:

How many of my systems do you already integrate with, in production, not on a roadmap? Does a resident have one profile across all of them, or one per module? Where does an account come from: the tenancy, or a spreadsheet? What happens to that account, everywhere, the day the tenancy ends? Can my own AI tools reach this platform directly, or is your AI assistant the only way in?

Those five answers tell you more than any screenshot will. They also happen to be the questions that separate a supplier you'll replace in three years from infrastructure you'll build on.

The app still earns its place. It's just no longer the thing you're buying.

Stop implementing resident apps. Start implementing the layer they run on.

image