A boiler starts making a noise three weeks before it fails. A damp patch spreads slowly behind a wardrobe for a month before anyone reports it. The signs are almost always there before the emergency call comes in. The question is whether anyone saw them in time.

That's the promise behind the new wave of AI in property maintenance: predictive systems that read service history, equipment age and usage patterns to flag problems before they become emergencies. Industry estimates suggest predictive maintenance can cut emergency repair events by 40 to 60 percent and extend equipment lifespan by 15 to 25 percent. Emergency repairs typically cost three to five times more than planned ones, so the financial case is strong. But for most repairs teams in social housing, the barrier isn't whether the technology works. It's whether the information ever reaches the person who needs to act on it. Job details sit in one system, photos in another, resident history in a third. A pattern that would be obvious on one screen gets lost across three logins.

The smarter path isn't necessarily more sensors or more data. It's making the data that already exists visible at the point of decision. When a surveyor or contact centre agent can see a job's full history, prior visits, photos and flagged risks in one place, patterns become obvious without needing a predictive algorithm to spot them first. Triage becomes proactive rather than reactive, because the person triaging has the full picture rather than a single ticket.

This is where Alix fits. Alix brings every piece of job information, photos, notes, resident context and history, into one system that's built for triage, not just ticketing. Instead of a contact centre agent seeing an isolated report of "damp in bathroom," they see the full pattern: previous callouts, associated repairs, any vulnerability flags, and the photos attached to each one. That visibility means routine repairs get prioritised properly instead of piling up around whatever's loudest that day, and genuine emergencies get identified faster because the context is already there. It's not about predicting the future. It's about not losing sight of the present.

AI in maintenance will keep getting more sophisticated. But the housing providers who benefit first won't necessarily be the ones with the fanciest algorithm. They'll be the ones whose teams can already see the whole picture, because that's what makes any signal, predictive or otherwise, worth acting on.