What over 20 housing providers told us about AI
Over the past six months, my colleagues and I have sat down with senior executives, spanning IT directors, heads of asset management, compliance leads and operations directors, from more than 20 housing providers for structured conversations about artificial intelligence, in addition to the countless meetings our customer success and account management teams have had. Not demos or sales meetings, just open conversations about how they’re really using AI, what they want from it and what’s stopping them.
What they told us was remarkably consistent. And it describes a sector in a strange moment and one the national research confirms; from recent housing-sector studies, over nine in ten housing providers have experimented with generative AI, yet only a handful have embedded it into how the organisation actually works and NHF-partnered research found that 44 per cent of housing providers have no AI policy at all. AI experimentation is everywhere but adoption is almost nowhere.
Our conversations explain why and what the few housing providers closing that gap are doing differently.
What we heard – The work, not the technology
Ask housing professionals where AI should help and nobody talks about chatbots in the abstract; they talk about specific, grinding work.
Repairs came up in every conversation, such as backlogs, rising costs, jobs lost in the system and poor information reaching contractors. As one director said, “If you conquer rents and repairs, you’ve conquered 80 per cent of the complexity.”
Reporting was a close second, with one participant describing their weekly ritual: “I spend a lot of my time pulling something off the system, putting it into Excel and then manipulating it.” Others described waiting weeks for reports that answered questions a director had needed that day. The wish was always the same: ask the system a question in plain English and get an immediate answer.
Another area frequently raised was compliance administration, with certificates arriving by email, each opened, read, interpreted, filed and actioned by hand, across thousands of properties. One IT lead put their ambition perfectly: “It would be so nice if all we had to do was leave them in a folder and the certificates just uploaded themselves into the right place.”
Notice what connects these: nobody asked for artificial intelligence. Everyone asked for work to do itself, with a person approving it rather than performing it. Their feedback echoes Aareon’s own internal approach to AI, that “AI is there to support people, not replace them.”
What we heard – Why it stalls
When we talked to housing providers about what was stopping their AI adoption, the answers were never about technology either.
Their answers were about trust and governance. One IT director described a board excited about AI while the data protection lead was, quite rightly, asking questions nobody could answer. Another had restricted use of public AI tools, not from resistance, but because nobody could say where the data went. Several were drafting their first AI policies while others had formed AI councils.
These aren’t signs of a sector resisting change; they are signs of a sector taking its responsibilities seriously. Housing providers manage public assets and serve people in vulnerable circumstances, so the tolerance for poorly-governed technology is rightly low.
The mistake is concluding that caution means waiting. The housing providers moving fastest in our conversations had understood the opposite, that governance is not the brake on AI adoption, it’s the engine of it.
Systems of record & humans in the loop
Underneath both lists (the work that people want done and the fears that stall it) sits the same structural problem.
Housing providers have spent decades investing in software ‘systems of record’. They capture tenancies, repairs, payments, complaints and inspections, and they are essential. But their design assumption is to record what has happened.
A system of record tells you when a resident is in arrears, whereas a ‘system of action’ highlights patterns in payment timing and engagement behaviour that might indicate a resident could benefit from earlier, supportive outreach before the situation escalates. A system of record confirms which inspections were completed, while a system of action monitors throughput against regulatory deadlines and flags the gap while there’s still time to close it.
Crucially, and this directly answers the trust question, a human decides in every case, a principle referred to as ‘human in the loop’. The system surfaces evidence, reasoning and confidence levels then the human professional exercises their judgement. That’s not a limitation of the approach; it is the approach.
What the few do differently
The housing providers in our research who had moved beyond experiments shared a consistent pattern of putting boundaries before capabilities. AI operates within an explicit scope and escalates what it’s not authorised to handle; in a sector where a wrong answer about gas safety has real consequences, knowing when not to act is a design feature.
They make outputs traceable. Staff trust what they can interrogate, and the first question of any AI output is “why?”. They keep accountability human, with clear sign-off points and audit trails. And they start small and prove it with, say, one bounded use-case, measured honestly. As one operations director told us: “Deliver quick wins along the way and the business believes you.”
A leadership question
Our discussions with housing providers have left me convinced the sector is not short of data, ideas or committed professionals. However, it is short of an operating model that connects them, and that model is built through choices about accountability, boundaries and trust that only senior leadership can make.
The adoption gap won’t be closed by (yet) another pilot. It will be closed by housing providers who treat governance as the foundation rather than the obstacle, and who ask of every AI initiative not “what can it do?” but “what work are we redesigning, who is accountable and how do we govern it?”. The organisations answering those questions deliberately will intervene earlier, respond more equitably and be able to show residents, regulators and boards exactly how.
That’s what ‘beyond the hype’ actually looks like.
Dr Mohammad Alomari is the chief AI officer at Aareon UK.
For further details, please see: www.aareon.co.uk.
