AI sounds the death knell for legacy HMS
For two decades, the ‘moat’ protecting legacy housing management systems wasn’t exactly built on a foundation of breathtaking innovation. It was constructed from the sheer, soul-crushing cost and complexity of trying to leave.
Many housing providers found themselves essentially ‘tenanted’ to their software vendors, trapped using a partially-useful but frozen platform because the alternative was a multi-million-pound leap into a digital abyss.
But as AI coding assistants tear up the software development world, that moat appears to have silted up.
The catalyst for this sudden disruption isn’t yet another shiny new competitor; it’s a fundamental shift in how we ‘crank out code’. The emergence of AI coding assistants, such as Cursor, GitHub, Copilot and especially Claude Code, has collapsed the cost of custom development and integration by 90 per cent. We’re entering what the AI-verse is calling the ‘SaaSpocalypse’ – a brutal market correction where the value of a ‘black box’ proprietary database is democratised down to nothing. Anyone can make enterprise software now… kind of.
The death of the ‘seat-based’ model
The traditional HMS business model is like a bad gym membership; you pay for ‘seats’ regardless of whether anyone is actually getting any value from the equipment. But when AI agents start handling the heavy lifting of repairs reporting, tenant enquiries and arrears workflows, the ‘per-seat’ metric starts to look irrelevant.
Faced with record stock falls for major software firms, Marc Benioff, CEO at Salesforce, has already hit the panic (sorry, I mean ‘pivot’) button, shifting his entire empire toward ‘Agentforce’. He argues that the future isn’t about selling licenses for humans to log into a CRM, it’s about selling ‘outcomes’ delivered by a digital workforce.
For social housing, this means the value is migrating away from the user interface (those crappy screens your staff love to hate) and toward the data fabric underneath. As Nvidia’s Jensen Huang quipped, coding is increasingly a ‘task’ rather than a barrier. If an AI can build a bespoke, high-efficiency workflow in the time it takes to get a legacy vendor to return a support ticket, why on earth are we still paying a premium for their half-hearted product support and ‘maintenance’?
The ‘SaaSpocalypse’ concept holds that paying for a proprietary black-box database is a terminal strategy. Forward-thinking providers are moving towards ‘sovereign platforms’, standardising on global ecosystems such as Microsoft Dataverse. By adopting a robust, AI-ready foundation, organisations ensure their data is already in the ‘language’ that next-generation agents speak, instantly bypassing a decade of technical debt.
The rise of the composable HMS
We’re moving toward what Ben Thompson (influential author of the Stratechery blog) likens to ‘thin clients’ – all the real grunt is happening elsewhere. You shouldn’t have to navigate a labyrinthine menu to find a tenant record; you just ask your AI assistant to “summarise the last three repairs for Flat 4”.
This requires a ‘composable’ approach. As Gartner’s 2026 ERP Roadmap suggests, the winners will be the ‘geeks’ who can hot swap and customise individual housing functionality modules without the whole house of cards falling over. In social housing, this allows for a strategy of adaptation. You take the core of a global platform and glue together the specific components you actually need, creating a bespoke experience without the bespoke risk.
The most common defence from the legacy giants is the ‘regulatory moat’. They’ll tell you that only their established, closed-end HMS can guarantee the ‘golden thread’ of data required by the Building Safety Act and Awaab’s Law.
I’d argue the opposite. In an era of statutory 24-hour response windows, rigidity is your biggest risk. A ‘data graveyard’ in a legacy system makes it nearly impossible to triage high-risk damp and mould reports with the speed the regulators demand.
A platform-centric architecture treats compliance as a dynamic workflow. By adopting a common data model in your organisation, the ‘golden thread’ becomes a searchable, auditable data fabric. AI agents can scan incoming reports for hazard keywords and trigger emergency workflows instantly, creating a time-stamped, tamper-proof audit trail. In 2026, “the system won’t let me” is no longer a valid excuse.
SaaS is no longer about buying a rigid suite; it’s about ‘composition’. The winners will be those who can adapt their technology stack by swapping individual modules without de-stabilising the core. In a high-regulation sector, the ability to glue together specialised AI agents is the only way to maintain compliance at pace.
The in-house revolution
Tier-one housing providers with their own development teams have been looking for a way to take control of their own destinies for decades. Sadly, a lot of public money has been thrown at global ERP vendors or misguided attempts to become a software company with a single client.
However, AI coding assistants are changing all that. They allow teams to move at a speed unthinkable just a couple of years ago.
But there’s a catch. Cursor CDE Michael Truell recently raised concerns with ‘vibe coding’ (i.e. using AI to build software entirely conversationally, without actually understanding what’s happening). Without an architectural understanding, innovation can quickly turn into a hot mess of solutions to problems that don’t exist. Spaghetti code on steroids.
The solution isn’t to stop building; the solution is to build using validated components. Internal teams can use AI to accelerate the last 20 per cent of the experience that actually matters to residents while relying on proven enterprise architecture for the boring-but-vital 80 per cent that keeps the regulators happy.
For tier-one housing providers, the build vs. buy debate has been settled. The most adept organisations will stop waiting for vendor roadmaps and start building their own, for the first time with confidence of sustainability. The key is to leverage tried and tested housing-specific components to fuel internal AI sprints, combining institutional knowledge with enterprise-grade building blocks.
The new mandate for digital sovereignty
The emergence of AI coding assistants hasn’t just made software cheaper, it has also shifted the power dynamic of the entire market. We’re moving away from an era of vendor-dependence towards an era of architectural sovereignty.
The question for a modern CISO is no longer “which system should I buy?” but “which architecture gives me the freedom to evolve?”. Whether you adopt a stable platform core, adapt its modules or become adept at building your own tools, the goal is the same: ensuring the technology serves the housing mission, rather than the mission serving the technology’s limitations.
And if your current vendor tells you otherwise? Well, they’re probably just whistling past the graveyard. Or planning their retirement.
Aidan Dunphy is the chief product officer at Esuasive.
