Field Notes 28 June 2026 8 min read
What NZ property professionals are actually doing with AI
The tools getting traction in real estate and property management look nothing like the early hype.
The tools getting traction in real estate and property management look nothing like the early hype.
The NZ property market runs on trust, timing, and local knowledge. Those three things don’t obviously pair with AI tools, which partly explains why adoption in the sector has lagged behind professional services and retail. But over the past twelve months, the pattern has shifted, and the agents and property managers who are using these tools have worked out something useful: AI is not replacing the local knowledge, it’s absorbing the admin around it.
The gap between what’s being tried and what’s actually sticking is significant. A lot of early experimentation went into AI valuations, automated chatbots on property websites, and “personalised” email sequences that weren’t personal at all. Most of that has been quietly shelved. What’s remained are the tools that cut the time cost of repeatable work without requiring the agent to hand over their judgment.
The most common use case we encounter is listing copy, which makes sense because it’s high-volume, time-sensitive, and has a clear quality floor. A good property listing needs to get facts right, match the price point in tone, and land the search terms relevant to the suburb or property type. Those are tasks a well-prompted AI handles competently, given the right inputs.
The constraint is input quality. Agents who paste in a bedroom count and floor area get generic output. Agents who include notes from their vendor conversation, specifics about the morning light, what the renovation actually cost, and which comparable sales support the asking price, get copy that sounds like it came from someone who’d visited. The tool is fast; the discipline of feeding it well is the actual investment.
One residential agent in Tauranga described spending around four hours a week writing and revising listings at peak. With a consistent prompting workflow, that dropped to roughly 40 minutes, and her appraisal documents, now generated from the same notes, became more detailed rather than less. The time saving went back into vendor contact, not into doing less work.
There’s a longer-term benefit here that’s easy to miss. Listing copy produced with natural language around suburb features, property type, and lifestyle signals feeds AIO work downstream. AI search engines index this content differently to traditional Google, and listings that answer the kinds of questions buyers are actually asking (“three-bedroom house with a study in the Waikato”) are easier to surface accurately. Running this through a repeatable AI automation workflow means quality stays consistent rather than varying with how much time the agent had that day.
Residential property managers have different problems than sales agents. The work is lower on adrenaline and higher on volume: maintenance requests, lease renewals, arrears follow-up, inspection reports, tenant onboarding. The communication volume per property is substantial, and most of it follows recognisable patterns.
The tools getting traction here are not chatbots or virtual assistants in the consumer sense. They’re more like structured templates connected to data. A property manager receives a maintenance request; the AI drafts the acknowledgement, routes the issue to the right trade category, and produces a summary for the landlord. The manager reviews and sends. Turnaround on routine communications drops from a day or two to under an hour.
What doesn’t work is anything requiring genuine discretion: a difficult tenant conversation, a landlord unhappy with a recommendation, a situation where legal exposure is possible. Managers who’ve tried to automate those conversations have found the AI either under-communicates or produces something technically accurate but tonally off. The value is in the routine, not the edge cases.
A property manager at a mid-sized firm in Hamilton was handling a portfolio that previously needed a junior administrator alongside them. That role still exists, but it now handles exceptions rather than first-pass communications. That’s a meaningful shift in how the work is structured, and the cost saving per property is real even when the headcount hasn’t changed.
This is the area with the most visible change over the past six months. AI-assisted property video has moved from novelty to something closer to a production shortcut, particularly for mid-market listings that previously wouldn’t have budgeted for professional video at all.
The hybrid approach works well: real footage of the property itself, edited and graded properly, combined with AI-generated B-roll for lifestyle and location context. A beach suburb listing gets coastal atmosphere shots without the cost of a full aerial session. A new build gets interior warmth shots the unfurnished property couldn’t provide. The AI filmmaking tools available now are consistent enough that the join isn’t obviously visible at normal viewing resolution, provided the real footage is shot to a reasonable standard.
AI avatar content is also gaining traction, particularly for commercial property agents who produce regular market commentary. A 90-second market update, presented by an on-brand avatar reading a script the agent wrote, takes around two hours to produce rather than a half-day of filming, reviewing, and uploading. The volume of content becomes sustainable at that production cost. The agents doing this well are transparent about what the avatar is doing: presenting data, not simulating a relationship. That positioning matters for how it lands with an audience.
Virtual staging remains the biggest disappointment in this space. The tools have improved significantly, but photorealistic staging on real property photography is still inconsistent. Lighting mismatches, furniture at the wrong scale, and the occasional uncanny-valley result mean virtual staging requires human review on every image, and sometimes several regeneration rounds before a usable version appears. The time saving over traditional virtual staging is real, but it’s not the zero-effort solution the marketing implies.
AI valuations are mostly being used as starting points rather than conclusions, which is probably the right posture. The comparable sales data is accurate; the weighting applied to differences in condition, aspect, and local micro-factors is where human judgment still earns its place. We haven’t spoken with any agent who treats an AI estimate as more than a prompt for their own analysis.
Generic AI content, produced without local context, is visibly generic. The suburb blog posts written without reference to actual local knowledge, the social posts that mention a “vibrant community” without naming anything specific, the email sequences that could describe any property in any New Zealand city: these are findable, and agents who’ve published them report they don’t perform. The content that works is specific, named, and locally anchored. That’s not an AI limitation so much as a reminder that specificity was always what made property marketing land.
The clearest benefit metric cited is time returned to client-facing work. Listing copy, property management communications, and market update video all reduce admin without reducing the professional surface area. The harder question, whether AI-assisted listings sell faster or at better prices, doesn’t have a clean answer yet. Market conditions, vendor motivation, and pricing accuracy matter far more than copy quality in most cases, and the confounding factors are significant.
What does show up consistently is that the teams using these tools are handling more volume without increasing headcount. A principal in Christchurch described it as the difference between keeping up and keeping pace. Her team of three is managing roughly 30% more landlord relationships than eighteen months ago. Whether that’s entirely attributable to AI tools or to the broader process improvements she made at the same time is genuinely unclear. The tools are one part of a wider efficiency push, not the whole explanation.
The next meaningful shift for NZ property will come when AI tools get better at aggregating and surfacing local sales intelligence. Agents who build structured data practices now, recording not just sale prices but buyer demographics, time-on-market factors, and objections encountered during campaigns, will be positioned to use that data well as AI retrieval improves. The agents treating their CRM as a searchable knowledge base rather than a contact list are building something with compounding value. That’s the habit worth forming before the tools catch up to it.
From automated guest comms to multilingual booking content, here's where the tools are earning their place.
After a year of hybrid AI video production for NZ clients, here's what works, what doesn't, and where the economics change.
NZ accountants, lawyers, and advisers carry heavy admin loads. Here's where automation is already paying off.