Use-cases, value and priorities
While many housing providers are experimenting with AI, we believe that few have fully unlocked its potential. Recent research reveals pockets of innovation, widespread uncertainty and a growing need for strategic clarity.
Our research, ‘Aspirations & applications of AI in social housing’, was based on 220 survey respondents and 50 in-depth interviews from 10 housing providers across England. This article explores those findings, reviewing the current state of AI in social housing, highlighting real-world use-cases, such as tenant complaint tracking and predictive maintenance, and introducing a practical framework for prioritising AI projects based on their difficulty and value.
The study found that although 31 per cent of respondents use AI tools, only 22 per cent were aware of AI being available for specific roles. However, trust remains low, with just 20 per cent believing AI consistently delivers accurate information and only 42 per cent feeling it aligns with their organisation’s values. The study also flagged a major problem, that of poor data quality across the sector.
What’s working?
The informal adoption of AI in housing is good, with around 31 per cent of employees reporting using AI in their roles, although often without formal organisational awareness. Among these users, 94 per cent reported benefits such as time savings and improved communication.
Two-thirds of respondents believe AI will improve service quality (65 per cent), boost personal productivity (67 per cent) and deliver cost efficiencies (68 per cent). And AI is being used to streamline repetitive tasks, such as grammar-checking communications and analysing complex datasets.
What’s lagging?
The majority of staff aren’t aware of how AI being used in their organisations, with a striking 64 per cent of staff unsure what AI tools their organisation had.
When it comes to having the right AI strategy and policies, only 14 per cent were aware of an AI policy and fewer than four per cent knew of a formal strategy. Training is also lagging, with just six per cent reporting access to AI training, indicating a reliance on informal learning and experimentation.
Unsurprisingly, given the other findings, only 44 per cent of employees believed AI supports good decision-making and just 20 per cent trusted AI to provide consistently accurate information. And confidence in AI’s ability to support equality, diversity and inclusion (EDI) was low. Fewer than half believed AI could help identify tenant vulnerabilities or deliver personalised services to marginalised groups.
This fragmented landscape suggests that although AI is being used, its deployment is often ad-hoc, lacking strategic oversight and alignment with core social housing values. However, despite the challenges, AI is already proving its worth in specific operational areas. Two standout use-cases are automating the tracking of tenants’ complaints and enabling predictive maintenance.
Tracking tenants’ complaints
Tenant complaints are a critical touchpoint in housing services. Traditionally, these are handled manually, often leading to delays, misrouting and inconsistent follow-up. AI can now streamline this process through natural language processing (NLP), by analysing emails, messages and voice inputs to automatically categorise complaints. Through using sentiment analysis, AI can understand emotional tones within different communication formats to help prioritise urgent or sensitive problems.
The use of smart routing means AI can assign complaints to the most suitable team based on type, severity and historical resolution data. This not only improves operational efficiency but also enhances tenants’ satisfaction by ensuring timely and appropriate responses.
Predictive maintenance
For housing providers, reactive maintenance is expensive and disruptive. AI is enabling a shift towards predictive maintenance by analysing sensor data, historical repairs and environmental conditions.
For example, by using anomaly detection, AI can flag unusual patterns in equipment performance and predict failures before they happen. AI is also being used to optimise scheduling so that maintenance can be planned to minimise downtime and extend asset life. Meanwhile, budget forecasting is improved with predictive AI models that help to allocate resources better, reducing emergency costs.
Unlocking AI’s potential
While there are many strong use-cases for AI in housing, there remain problems around gaps in strategy, data readiness and staff training. To see effective use of AI in housing, guardrails must be used to ensure the responsible use of AI.
Success with AI starts with clean, accessible data. If systems contain outdated or duplicated data, or if your infrastructure prevents AI access, any AI efforts will stall – many organisations falter at this stage. Reviewing your data alongside your AI goals helps to identify what needs fixing before value can be extracted.
But how do you prioritise which AI projects to start with?
Choosing the right projects
With limited resources and myriad potential applications, housing providers need a strategic approach to AI adoption. A ‘difficulty vs. value’ framework offers a practical way to evaluate and prioritise projects.
Step 1 – Define value
Value should be assessed across multiple dimensions:
- Operational efficiency – Time and cost savings.
- Tenant experience – Improved satisfaction and engagement.
- Compliance and risk reduction – Better adherence to regulations.
- Social impact – Alignment with EDI and sustainability goals.
Step 2 – Assess difficulty
Potential difficulties include:
- Technical complexity – Data requirements, integration challenges and model sophistication.
- Organisational readiness – Skills, training and cultural acceptance.
- Data quality – Availability and reliability of data inputs.
To give some simple examples, managing shared inboxes and automating email responses are low-effort, high-impact use-cases. Tenant complaints, which are often complex and time-consuming, can be streamlined with AI. By analysing complaints’ data, AI can triage problems and send automated updates, freeing staff to handle more nuanced cases.
AI can also detect patterns, such as identifying recurrent problems in a housing block. Temperature data from buildings can be matched with tenants’ feedback to optimise heating or cooling, improving comfort and reducing costs.
Proving value and RoI
Starting with manageable, high-value projects to build confidence and demonstrate AI’s benefits is key to success. As teams gain experience and savings grow, more complex initiatives can follow, with proving value and RoI being top priorities.
Working with an experienced AI services provider will help housing providers create an AI roadmap that will guide them through identifying opportunities, estimating RoI and creating actionable plans. Any AI strategy should include short- and long-term goals, cross-departmental collaboration and change management to build trust and ensure successful adoption.
AI adoption is at a pivotal moment in housing. Technology can improve tenant services and optimise operations, but its adoption is uneven and often unstructured. Setting a strong AI strategy is key to successful implementation of AI and to ensure return on investment.
Mark Rotheram is the chief technology officer at BCN.
