Last month, I asked whether tech could replace traditional property managers . The economics of automation, I argued, are difficult to ignore. AI adoption that has nearly tripled in a year, predictive maintenance cutting emergency repair costs by up to 30%, and software platforms that cost a fraction of what a traditional manager charges. This raises a rather obvious question: if the economics are this compelling, why hasn't the industry simply completed the transition?

For all the efficiency gains, only 8% of property management firms have fully automated any single workflow, according to Buildium's 2026 Industry Report - the very same report that documented the surge in AI adoption. Most platforms sit on top of existing operations, making staff faster rather than redundant. A chatbot handling 70% of initial enquiries still needs a human for the remaining 30%, and that 30% tends to be the complicated, high-stakes work: the tenant in arrears, the dispute between neighbours, the landlord who needs to be told their asset is underperforming.

McKinsey's March 2026 analysis draws a useful distinction between steps and thoughts. Routing a maintenance request, scheduling a contractor and updating a status log are steps. Deciding whether a tenant relationship is worth preserving when rent is two weeks late, or how to handle a heritage building that doesn't behave like any dataset, is a thought. AI handles steps well. It has made little ground, thus far, on thoughts.

Where the tenant relationship actually lives

Tech-enabled properties show 23% higher tenant retention in some smart building cases, and a one-point improvement in tenant satisfaction is linked to an 8.6% higher lease renewal likelihood . Technology deserves some of the credit. Faster response times help, and so does a well-functioning app. But in high-value residential and commercial portfolios especially, the tenant relationship remains the product.

This is where the hybrid model gains traction. 52% of property management firms already treat AI as a collaborator rather than a replacement for staff. Perhaps this reflects not a reluctance to automate, but a clear-eyed understanding of where the technology creates value and where it stops.

The costs that don't show up in fee comparisons

Then there is the risk argument. Whilst there is a fee gap between traditional management and software platforms, what those comparisons rarely capture is the cost of getting things wrong. Algorithmic bias in tenant screening is a live regulatory issue in several US states, and GDPR compliance adds complexity that pure-tech platforms have been slow to manage consistently. Training gaps also compound the problem: only 28% of real estate professionals have received formal training in the tools their firms have adopted. When technology outpaces the people operating it, efficiency gains tend to shrink - and errors risk changing character, from mistakes a manager makes and owns to mistakes a system makes that have the potential to go unnoticed.

The firms getting this right are not defending their model against technology. As in many other sectors, they are using automation to take on the routine work - enquiries, scheduling, reporting - and redirecting that time toward the work that builds long-term value - landlord relationships, tenant retention, portfolio strategy. 85% of institutional investors now expect AI tools to be standard in asset management. Increasingly, the value a manager adds lies in making sense of what those tools produce.

The competitive logic points to pressure on mid-market firms - those built on being reasonably good at everything. It is a familiar pattern: automation absorbs the middle of the quality range first, because that is where performance is most standardised and least personal. Reasonably good is precisely the standard software reaches first.

The replacement question was never really a yes-or-no. It was always which parts, at which price points, and for which kinds of assets. Automation in property management is no longer new, and no longer rare. Its benefits are cost and speed. The challenge now is what it looks like, in practice, to spend the hours the software has returned on the thoughts it cannot have.