AI is already here — just not evenly distributed
In social housing, AI is no longer a distant promise — it’s already here, woven into everyday work. The latest NHF AI Survey (July 2025), HAILIE User Intelligence Report (July 2025), and Aspiration and Applications of AI in Social Housing (Sept 25) reveal that more than 90% of housing providers have begun experimenting with generative AI tools like ChatGPT or Microsoft Copilot. Yet only 4% have embedded them at scale. This gap between enthusiasm and execution defines the sector’s current moment. AI’s presence is everywhere, but its impact is uneven. To understand what that looks like in practice, let’s follow a typical housing association through the eyes of four people: a tenant, a housing officer, a CEO, and a board member, each living a different day in the life of AI.
From pilots to people
Across the sector, adoption has outpaced assurance. The NHF, DASH (Oct 2025), HAILIE, Service Insights surveys and studies show common themes: staff-led experimentation, weak governance, limited AI literacy, and data that is often incomplete or inconsistent. Only around 20% of organisations have a formal AI policy or strategy in place. Boards are learning as they go, and frontline staff are often teaching themselves. Tenants rarely know when AI is shaping their interactions, and few feel its benefits yet. The following snapshots bring these statistics to life, showing how AI feels to those who live, work, and lead inside housing.
The Tenant – Pauline’s Day
For Pauline, AI is mostly invisible until it isn’t. When damp reappears in her bedroom, a predictive repairs system quietly flags it as urgent, yet no one calls. Later, a chatbot texts to confirm an appointment that’s been cancelled the next morning. She sighs: “It’s clever, but no one listens when it gets it wrong.” Like most tenants, she welcomes technology that promises faster fixes, but only if it delivers, and if someone follows up. Research shows fewer than one in five tenants feel digital tools make repairs easier, and trust drops sharply when automation replaces human contact. Pauline doesn’t want algorithms; she wants assurance. She wants to know when AI is being used, why decisions are made, and who is accountable when they’re wrong.
So what should you do (Tenant view): Design AI around empathy, not efficiency. Tell tenants when and how AI is used. Keep human routes open. Track ‘trusting AI’ alongside repairs KPIs. Tenants like Pauline will believe the technology only when they feel the difference in their homes. For more details on the tenants’ voice on AI, read Tenants, Trust and AI: What housing leaders need to hear
The Housing Officer – Amir’s Day
Amir spends his mornings triaging tenant emails and case notes. Between home visits, he uses Copilot to summarise meetings and draft follow-up letters. It saves him time, a small but welcome win in a job that rarely pauses. Yet he isn’t sure what happens to the data he uploads, or whether he’s even allowed to use these tools. Like many frontline staff, Amir is learning by doing. Sector data shows that while over 90% of providers are experimenting with AI, only 2% provide formal training or guardrails (NHF July 2025). Enthusiasm is high, but confidence is low, and without structure, every prompt carries potential risk. Without clear guardrails or audit trails, this “DIY adoption” risks data errors and compliance slips.
So what should you do (Staff view): Turn experimentation into structured learning. Provide prompt libraries, practical training, and clear guidance for everyday tools. Encourage staff to share what works and what doesn’t. Capture lessons before enthusiasm outpaces policy.
The CEO – Sally’s Day
For Sally, AI has become both a promise and a puzzle. Her dashboard displays colourful analytics, sentiment from tenant surveys, repair completion data, and Copilot-generated board summaries. It looks impressive, but she still wonders: how much of this can I trust? Her teams are piloting chatbots and predictive models, but evidence of real impact remains patchy. According to the research, only 4% of organisations report widespread AI use, while 80% cite lack of skills and confidence as major barriers. Data quality is improving, but governance remains underdeveloped. Sally sees the potential for proactive services, better insights, and fewer complaints, but knows that ambition without assurance risks trust. She knows the Regulator and Housing Ombudsman are watching closely, AI assurance will soon be as expected as data protection or safeguarding policies
So what should you do (CEO view): Link every AI pilot to a strategic priority, safety, satisfaction, and sustainability. Demand evidence, not anecdotes. Invest in AI literacy and data readiness before scaling. Build assurance around impact, ethics, and purpose.
Case Study: Wolverhampton Homes Predicts Damp & Mould
Wolverhampton Homes and NEC used AI to predict damp and mould risks across 21,000 homes — combining property, repairs, and environmental data within NEC’s maintenance system. The model reached 98% accuracy with complete data, but fell to around 70% when records were incomplete. That insight drove a 15% reduction in unplanned repair visits. The lesson is clear: AI is only as good as its data. For CEOs like Sally, it’s proof of potential; for boards like Bob’s, it’s a reminder that governance and audit must evolve as models mature.
The Board Member – Bob’s Day
Bob opens his board pack to find a line about ‘AI-enabled efficiencies in customer service’. It sounds promising, but what does it really mean? During the meeting, the finance director mentions Copilot trials, the CEO references predictive repairs, and the conversation shifts to risk. Are data protection checks in place? Who owns AI oversight? How do we know this aligns with our values? The HAILIE and NHF reports show that fewer than one in four providers have a named AI lead or policy. Boards are being asked to approve innovations they barely understand. Bob doesn’t want to slow progress; he wants confidence that experimentation is happening within a framework, that someone is accountable, and that bias is being checked. Bob sees governance as the missing spine of AI adoption; assurance frameworks, audit routines, and bias reviews need to catch up before innovation runs ahead of control.
So what should you do (Board view): Treat AI as both an opportunity and a governance responsibility. Build digital fluency through short briefings and scenario-based discussions. Ask clear assurance questions: who owns AI risk, how are decisions audited, and how do outcomes support tenants? Integrate AI oversight into existing risk and ethics frameworks.
Crosscutting Reflection – What These Lives Tell Us
Viewed together, these four stories and examples like Wolverhampton’s predictive damp and mould model sketch the real state of AI in housing: hopeful but half-built. Pauline’s experience shows tenants still living at the edge of digital progress. Amir’s story reveals enthusiasm without structure. Sally’s day exposes the credibility gap between ambition and assurance. And Bob’s reflections highlight the need for informed oversight. Across all levels, the same pattern emerges: enthusiasm outpacing readiness, progress outpacing governance. The sector has moved from ‘if’ to ‘how’, but confidence, coordination, and capability have yet to catch up. Governance isn’t red tape; it’s what turns good experiments into sustainable practice. The message is clear: technology alone won’t transform housing; trust, strategy, and stewardship must grow together.
Closing Call to Action – From Pilots to Proof
AI in housing is no longer about awareness; it’s about readiness. The HAILIE, NHF AI Survey, Service Insights and DASH research show the same trend: enthusiasm outpacing assurance. The challenge for leaders isn’t to innovate faster, but to adopt smarter linking experimentation with evidence, and technology with trust. The stories of Pauline, Amir, Sally, and Bob all point to one truth: AI’s success in housing depends less on algorithms and more on leadership, literacy, and lived experience.
For Executives: Three Moves to Make AI Work
1. Start with Purpose – Align each AI initiative with measurable outcomes: satisfaction, safety, or sustainability.
2. Build Fluency, Not FOMO – Invest in literacy and data quality before scaling.
3. Measure Trust as Rigorously as Time – Pair efficiency metrics with fairness and transparency indicators.
For Boards: Responsible AI Oversight Checklist
• Strategy – Is our AI purpose aligned to our mission?
• Skills – Have members received AI risk briefings?
• Stewardship – Who owns AI oversight, and how are ethical decisions recorded?
Integrate these questions into annual assurance cycles — the Regulator and Ombudsman will soon expect it.
Download our DASH Insight Briefing brings together findings from the NHF, HAILIE, and DASH surveys to reveal where housing associations really stand on AI readiness, governance, and leadership.
