For years, the property industry has comforted itself with a sort of bureaucratic Darwinism: if someone had the energy to write a long, coherent letter, complete with statute names and bullet points, then at least they’d probably done the thinking. It was never a great filter for truth, but it was a workable filter for effort.
AI has murdered that filter.
You can see it in the numbers and you can feel it in the tone. When the Property Ombudsman is dealing with a reported 55% uplift in complaints in the first half of this year versus the same period in 2025, that’s not a gentle trend. That’s a structural shift. And Propertymark’s phrase “super complaints” is telling: not necessarily more accurate, but more polished, more insistent, and delivered at a volume that makes your average branch manager’s inbox resemble an overflowing redress tribunal.
The killer feature here isn’t that tenants (or vendors, or buyers) are suddenly more malicious. It’s that they can now outsource indignation to a machine that is, by design, obliging. Ask a generative model “have I been wronged?” and it doesn’t respond like a tired Trading Standards officer; it responds like a barrister who’s had three coffees and has decided you’re the main character. Then it produces a document that looks like it took a weekend to craft, complete with legislative references that sound credible until you actually read them.
And that is where the asymmetry bites. The complaint takes forty seconds. The rebuttal takes two hours. The portal lead gen bill doesn’t pause whilst you’re combing through three pages of confidently incorrect citations to a version of the Renters’ Reform Bill that exists only in the machine’s memory. Meanwhile, you’re having to be polite, because you’re a regulated professional and the complainant is not.
A past-president of ARLA Propertymark, Greg Tsuman, floated the obvious mitigation: require complainants to disclose AI use. In a neat bit of institutional reality-checking, the Ombudsman shut it down. Lesley Horton’s point is simple and, frankly, unarguable: the Codes set standards for agents, not consumers, and you don’t start building procedural tripwires for the public without creating exactly the kind of “agents hide behind process” narrative that the industry has spent decades trying to shake.
So, no, we’re not getting an “AI disclosure” box on complaint forms. That door is closed.
What matters more is Horton’s other admission: the Ombudsman is using AI too, partly to cope with volume, and it’s finding the same issue agents are finding — submissions where the actual evidence is buried under repetitive, irrelevant waffle. That’s the first glimpse of the next phase: machines writing complaints; machines triaging them; humans dragged in, expensively, to make decisions when the machines can’t.
If you’re an agent, the temptation is to “fight fire with fire”: use AI to write the response. Fine. But do it like a grown-up. The goal isn’t to out-gasbag the complainant; it’s to extract facts, timelines, and documents faster than the other side can generate prose. Treat AI as a paralegal, not a novelist.
Practically, that means three things. First, tighten your record-keeping so you can produce evidence quickly (tenancy notes, inspection photos, call logs, portal edits, everything). Second, build a standard internal “complaint pack” template so you’re not reinventing the wheel for each case. Third, use AI defensively: summarise, de-duplicate, highlight claims that require proof, and flag legal references that look suspicious.
Because the dirty secret is this: in the age of “super complaints”, professionalism isn’t just about doing the right thing. It’s about proving you did it, faster than a machine can make you look like you didn’t.