I’ve sat through plenty of change in this industry over the last four decades, but nothing has landed on the lettings industry quite like the combination of the Renters’ Rights Act on one side, and artificial intelligence on the other.
On its own, the Renters’ Rights Act is vast but manageable. We understand what it is there for; we agree, largely speaking, with its purpose – even if not always fully with its precise delivery; and in all cases, we understand as professional property sector businesses that legislation changes and evolves, and ultimately rigorous processes enable it to become part and parcel of life for agents, landlords and tenants.
There is always a period of adjustment to go through whenever the industry faces any significant legislative change.
But we face an added challenge in this current era that, as an industry, we haven’t encountered previously – or at least not to the same degree. It is something that is significantly adding to the burden and workload of agents adapting to the changes.
It is not the structure of the legislation itself, nor the newness of some of its elements, that we haven’t had to consider previously.
It is the existence of AI.
It is reshaping what it actually takes to manage a rental property properly – and most landlords have no idea it’s happening.
Tenant complaints haven’t just increased, they’ve changed shape.
Since the Renters’ Rights Act came into force, tenant complaints have soared across the sector.
But it is not so much the increased volume as how those complaints are now presented that really adds to the workload.
In the past, a tenant might have called up, discussed a problem, and nine times out of ten a quick conversation with a proposed resolution would sort it out.
Now, increasingly, the first a letting agent hears of an issue is a lengthy, formally structured letter – often threatening legal action, escalation, reporting the agent to Trading Standards or – perhaps – threats to take the problem straight to a First-tier Tribunal.
Multiple points. Legal-sounding language. References to clauses and rights. It reads like it’s been drafted by someone who knows the law inside out.
It hasn’t been. It’s been drafted by AI.
And therein lies a problem – because a little knowledge is a dangerous thing – and AI, convincing as it might look, doesn’t know everything.
Tenants start an AI query to ask a simple question, but find themselves engaging with it as if it is equivalent to free legal aid. The trouble is, when you boil it down, AI is neither neutral nor reliably accurate.
Whether or not they intended the chat with their AI to take them there, tenants seem often to reach a point where they are essentially asking “do I have a case?”, and out comes a complaint letter, ready to send.
There are several issues at play here.
For one, AI – by its nature – tends to offer a bias towards the person asking it the question. If you jump onto ChatGPT, Claude or whatever your platform of choice might be, and feed it one side of a potential dispute, it will very often validate that side – regardless of whether the full facts actually support it.
It’s not lying. It simply doesn’t have the other half of the story, and it cannot take account of facts it hasn’t been given.
Take a point from the examples I mentioned above – a threat to take an issue to Trading Standards. There are avenues open to tenants who have issues with their letting agents – but for an ordinary service complaint, Trading Standards is unlikely to be one of them. The relevant Ombudsman or other redress body the agent is under, yes, perhaps – once a complaint process has been followed through the agency itself first and that process exhausted. But straight to the ombudsman? No. And Trading Standards? Very unlikely unless very specific circumstances are in play – such as the agent not being registered with such a scheme.
Nevertheless, it turns out this is not where the tenant ended up after half an hour of chatting to their AI, and now they’ve arrived, letter written in what appears to be full legalese – but much of it is either incorrect or wholly disproportionate to the original issue.
This isn’t just my observation. It echoes what senior figures across the industry are flagging.
Greg Tsuman, a past president of ARLA Propertymark, has described a rise in complex, multi-part complaints – many without real grounds – and warned that AI’s tendency to validate the person asking it can leave tenants believing they have a stronger case than they do.
Ben Stokes, president-elect of ARLA Propertymark, has raised a similar concern: that AI is increasingly pointing tenants towards outdated legal information: old proposals, superseded bills – advice, indeed, to contact the local Trading Standards team. It will be dressed up as current reality, and the tenant has taken it all at face value. Because why wouldn’t they?
Nevertheless, we have to talk about where this leaves us as a sector.
Every complaint, however it was generated, still has to be treated as genuine. Every point still has to be checked, responded to and documented – because not just legally, but reputationally too, a letting agent can’t afford not to.
The workload this creates is enormous
Here’s an example of what this actually looks like from where we sit as agents on the ground dealing with this issue.
A well-worded, AI-generated letter can raise six or seven separate issues in one go – some legitimate, some based on a misunderstanding, some simply incorrect. What may have started as a straightforward enquiry to ChatGPT has spiralled into a litany by the time the complaint reaches us.
Each issue needs proper investigation.
Each response needs to be accurate, evidenced, and recorded, because The Property Ombudsman will expect to see that trail if it ever does escalate.
And the reality is that, had the tenant only called in with whatever the original issue was, the whole thing could probably have been resolved within a few hours.
For example, an issue starts as an online enquiry about how to resolve a dripping tap in a rental property.
Two hours later, it has become an entire epistle of a complaint, now submitted in writing, citing excessive noise, sleep deprivation, untold water bill increases and reference to deteriorating mental health.
Submitted in writing as it is and marked, on the advice of the tenant’s ChatGPT, as a formal complaint, this now has to be dealt with under the terms of the ombudsman or redress scheme – even if all that is really needed is to make a call to the plumber to visit and replace a washer.
Now multiply this kind of scenario across a growing caseload of tenants, and you start to see why “spike in complaints” barely captures what’s happening. It’s not more of the same work; it is a new category of work, requiring the kind of legal literacy and documentation discipline that simply wasn’t part of day-to-day lettings management a few months ago.
A word of warning for self-managing landlords
If you’re a landlord managing your own property without an agent involved, this is worth taking seriously. You may not have encountered an AI-drafted complaint yet, but the odds are shortening. When it lands, you won’t have a compliance process, a complaints procedure, or a team trained to separate genuine issues from misunderstandings, inaccuracies and scattergun claims.
You’ll be facing it alone, without the benefit of legal knowledge, working out in real time whether a claim actually holds up.
If you do already have a managing agent, you may not realise how lucky you are.
And you probably don’t realise most of this is even going on.
The complaint arrives, gets triaged, investigated, responded to and closed. From your side, it barely registers. You only know that we need to send a plumber to replace a washer – which, invariably, you sign off on without question! You have no idea that getting to that outcome involved three-quarters of a day’s work.
That invisibility is the point. It’s what a properly resourced agency is for.
At Petermans, we’re fortunate. We have been in the game long enough and built a team over time that is strong enough to absorb this new, added pressure. Structured escalation, solid documentation processes, and yes, using our own tools sensibly to manage the volume without losing the human judgement that actually resolves disputes fairly. AI can help us, too, to process a long letter and get to the source of what the problem is that needs addressing and how.
But AI can’t replace the experience needed to know whether the underlying complaint actually holds water in the first place.
That balance, between speed and judgement, is going to matter more, not less, as this trend continues.