I have now met with countless property executives who are genuinely surprised at the low pace of AI adoption inside their own firms, despite having switched on Copilot for all their teams. The pattern is remarkably consistent. A licence is bought, often the cheapest available, little or no real training is provided, and leadership then waits for a transformation that never arrives. Adoption stays low, frustration grows, and the promised step-change is nowhere to be seen. The assumption underneath it all is that putting a powerful tool in people’s hands is the same as changing how they work.
It is not. And the gap between those two things is the single most important story in property’s relationship with AI right now.
I have written before about the apathy of UK property towards technology. That apathy is real, and I will come back to why it is more rational than it first appears. But a new and more dangerous error has taken its place at the firms that consider themselves ahead of the curve: the belief that buying access to AI is the same as adopting it. It is not a strategy. It is a starting line that most firms are mistaking for the finish.
If you are a property leader reading this and feeling that you do not know where to begin, I want to say clearly at the outset that this is the most common position in the industry, not a failing on your part. The technology is moving faster than anyone can comfortably track, the investment is hard to quantify, and the noise is deafening. What follows is an honest account of where the value really is, where the risks really are, and what a sensible first step looks like. The picture is demanding, but it is navigable, and the firms that approach it with clear eyes rather than either hype or fear will do well.
Turning it on is not turning it round
There is genuine value in AI today, and it would be dishonest to pretend otherwise. Most firms that switch on an enterprise assistant capture a first tranche of benefit almost immediately, somewhere between thirty minutes and two hours of time saved per person per day. That is real. It is also shallow because the overwhelming majority of firms stop precisely there.
What they miss is that the tool delivers a fraction of its value without the change management and training that turn a novelty into a habit. The licence is the cheapest part of the journey. The expensive part, the part that actually moves the business, is the work of redesigning processes, training people on what these tools can and cannot do, and rebuilding workflows around new capabilities rather than bolting AI onto old ones.
These firms have also created a risk that most of them have not even considered. When a tool is poorly deployed and people are left frustrated, they do not stop using AI. They go around the firm. They open up free, consumer-grade tools on the side of their desk, paste in client information to get their work done faster, and, in doing so, hand confidential data to models that may use it for training. The firm that deploys AI cautiously and superficially has not avoided AI risk. It has multiplied it.
This is the uncomfortable truth about the “just turn it on” approach. It does not keep you safe, and it does not keep you still. It exposes you.
The value today, and the value six months from now
So where is the genuine value being created? Today, it is in deploying simple task agents in back-office functions: the repetitive, structured, high-volume work that surrounds professional judgement without being professional judgement. That is where the early wins are real and defensible.
Over the next six to twelve months, the centre of gravity moves towards agentic orchestration. This means building specific process automations, stitching tasks together, and, for the firms brave enough to do it, rethinking the process entirely rather than automating a bad one more efficiently. The greatest returns will not come from making today’s workflow faster. They will come from asking whether today’s workflow should exist at all.
I see three areas where serious competitive advantage will be built: agentic deployments, genuine investment in training and change management, and the construction of more resilient technology stacks that move firms away from heavy, brittle legacy systems. I would add a fourth trend worth watching closely. As it becomes dramatically easier to build software, I expect firms to begin replacing rigid, single-purpose third-party systems with their own lightweight versions, configured exactly to how they actually work. The economics of “build versus buy” are shifting under our feet.
Why property resists, and why that excuse is running out
It would be easy to read all of this as a story about a profession that is simply slow and reluctant. There is truth in that. But property’s caution is not pure stubbornness, and it is worth understanding why, because it explains both the resistance and its limits.
Property is not a commodity. No two units are ever identical. They differ by view, by floor, by distance from the lift, by lease term, by rental incentive, by a hundred variables that resist standardisation. The transactions are high in value and low in volume, which makes them illiquid and genuinely hard to compare like for like. The data is generally poor, and much of what we treat as fact is in reality assumption dressed up in confident language. This is a sector built on judgement applied to uncertain information, and that is exactly why it has been slower than others to hand its work to a machine. The caution has a logic.
But the logic is running out of road. AI is now raising the baseline across every profession, and property is no exception. Using these tools is shifting from a choice to a necessity. There is a short window in which firms can build real competitive advantage by moving before their competitors wake up, and that window is narrowing every single day. The firms that mistake property’s traditional caution for permission to wait are making a serious miscalculation. The caution was rational when the tools were immature. The tools are no longer immature.
The quiet logic of doing nothing
There is a deeper reason property stalls, and it is worth naming honestly because it sits in the boardroom rather than in technology.
Many of the boards and executives making these decisions are at the end of their careers. They have built their reputations and their equity over decades, and they understandably do not want to rock the boat in the years before retirement and exit. Property has always been a gut-feel business, where judgement is earned over a long career and trusted accordingly, so a technology that few of them truly understand is easy to dismiss as hype. They are frequently told by their CTO that the tech stack is not ready, that there is no AI use policy, no data governance framework, no completed systems migration, and that the AI journey therefore cannot begin until all of this is in place. Perhaps next year. Perhaps the year after. And the market is poor, so it feels like the wrong moment to commit significant, hard-to-quantify investment to an outcome nobody can guarantee while the technology itself keeps shifting under their feet.
I have a great deal of sympathy for the nervousness. It is rational. The investments can be significant, the outcomes are genuinely uncertain, and the pace of change is dizzying. But the conclusion that doing nothing is therefore the safe option is, I am afraid, exactly wrong. Standing still feels prudent. It is in fact the riskiest choice available, for three reasons.
First, standing still does not mean your competitors are standing still with you. If they are embracing AI and you are not, you will fall behind, your overheads will stay high while theirs fall, and the gap will widen to the point where catching up becomes harder and more expensive every quarter. The cost of inaction does not show up on the balance sheet today, but it compounds quietly until it does.
Second, your best people will leave. No one enjoys doing menial, repetitive work, least of all when they know a solution exists and their own firm is the only thing standing between them and it. There is a great deal that surveyors can automate safely, without placing the firm or its data at risk, and a firm so risk-averse that it refuses them even that will create exactly the frustration it was trying to avoid. Those people will turn to side-of-desk tools, and eventually they will turn to competitors who let them work the way they want to. Excessive caution does not protect your talent. It hands it to the firms who moved first.
Third, the opportunity to get ahead is real today precisely because it is still uneven. That will not last. As the market baseline shifts and everyone is using AI, the advantage evaporates and it simply becomes business as usual, in the same way that the internet, Excel and PowerPoint are now business as usual. There was a window in which using those tools well was a genuine edge. That window closed, and the firms that walked through it early were better for it. We are in the same window now, and it is closing at the same pace.
There is one more version of the wait-and-see argument that deserves to be taken seriously, because it is the most intelligent one. The pace of change is faster than anything we have ever witnessed. The Industrial Revolution unfolded over some thirty years; this is happening in quarters. Credible voices suggest we may reach artificial general intelligence within the next three to five years, and the honest truth is that no one knows what that means for society, let alone for property as a sector. So why invest heavily now, the argument goes, when in a year or two the same capability may be cheaper, easier and more mature?
It is a fair question, and it cuts both ways. It is true that those who embarked on the AI journey a year ago may have felt less immediate benefit than a firm starting today, who inherits dramatically better tools and gets a far bigger boost far faster. Consider where agentic AI was twelve months ago compared with now. So yes, waiting does mean buying better technology later. But here is the flaw in treating that as a reason to wait: the firms that started a year ago did not just buy worse tools. They built the data foundations, the governance, the culture and the institutional muscle to absorb each new wave as it arrived. The firm that sits still, hoping the technology will slow down and reach a comfortable maturity, is making a bet that it has the time. Many will not. The unglamorous truth is that some firms will go out of business waiting for the perfect moment that never comes, overtaken by competitors who started learning while they deliberated. The tool gets better if you wait. Your ability to use it does not.
The role of the surveyor, five years out
I have said many times that AI will not replace surveyors, and I stand by it. But it will change where a surveyor’s time and value sit, and anyone who tells you otherwise is selling comfort rather than truth.
The defensible core of our profession is judgment and liability-bearing opinion. No client pays for a valuation because a machine produced a number. They pay for a qualified professional who is willing to stand behind that number and carry the responsibility for it. That does not disappear. If anything, it becomes more concentrated and more valuable.
What changes is everything around that core. The drafting, the research, the data gathering, the first-pass analysis, the assembly of the report: this is where AI compresses hours into minutes, and it is precisely the work that currently fills the days of junior and mid-level surveyors. The role shifts from producing the report to directing and assuring it.
Two further forces will shape the next five years. The first is that proprietary data becomes a genuine and durable competitive advantage. As the models themselves commoditise, and they will, the differentiator is no longer access to AI, which everyone will have. It is the unique data a firm holds and can bring to bear. Years of transactions, inspections, valuations, and asset histories become a moat that competitors cannot replicate. The firms that recognise their data as a strategic asset, rather than as exhaust from completed jobs, will pull decisively ahead.
The second force is that established brand equity and client inertia will slow the pace of change. Clients are slow to switch trusted advisers, and that stickiness will cushion incumbents for a while. But firms should not mistake a cushion for a defence. The brands relying on reputation and client apathy to avoid doing the work will eventually be overtaken by those who used the same window to build their data advantage and retrain their people.
And here I want to be blunt, because the profession needs candour more than it needs reassurance. Serious change is going to be forced on this industry. Many roles will change dramatically, and some will disappear, and a large part of the profession is not ready for that, and pretending otherwise helps no one. The surveyors who thrive will be those who move up the value chain into judgement, client relationships, and oversight. The firms that thrive will be the ones honest enough to retrain their people for that shift rather than quietly hoping it passes them by.
There is a deeper structural risk hiding inside this transition, and it is one the profession has barely begun to confront. Over the next two to three years, as AI becomes commonplace, it is likely that recruitment of junior graduates and mid-level surveyors will collapse because AI will absorb much of the work those roles exist to do. On its own, that looks like an efficiency. It is, in fact, a slow-acting threat to the entire profession.
We all agree that the best use of AI in surveying involves a strong expert in the loop: someone who knows what good looks like, who can tell a sound output from a plausible but wrong one, and who carries the professional judgement to stand behind it. But experts are not born. They are made, slowly, through years of doing exactly the lower-level work that AI is now poised to take away. If we stop recruiting and developing junior surveyors, we stop building the experts of the future. We may right now be living through a golden age, one in which we enjoy both extraordinary AI capability and a generation of seasoned experts able to supervise it. The uncomfortable question is what the field looks like in five to ten years, once those experts begin to retire, and there is no one behind them who was given the chance to build the same depth of judgement. A profession that automates away its own training ground risks eating its own seed corn. The firms that think about this early, and design deliberate ways for juniors to build expertise in a world where the traditional apprenticeship of repetitive work no longer exists will be the ones that still have experts to call on a decade from now. This is a solvable problem, but only for those who see it coming.
Data discipline is not the prerequisite. It is the work.
Every serious conversation about AI in property eventually arrives at the part the industry would rather skip: data. AI is only ever as good as the data and the processes beneath it. You can deploy the most capable model on the market, and it will still produce mediocre output if it is sitting on inconsistent reporting, unstructured data, and workflows that differ from one surveyor to the next.
I want to be precise about this. Structured data, standardised reporting and consistent workflows are not the unglamorous prerequisite to AI adoption. They are the adoption. The firms that have invested in data discipline will find that AI compounds their advantage quickly. The firms that have not will find that AI exposes the mess rather than fixing it.
There is a hard truth here. Much of surveying has tolerated bespoke, individual ways of working as a mark of craft. That craft is real, and I would never diminish it. But inconsistency is the enemy of anything that scales. The unglamorous work of standardising templates, structuring asset data and tightening workflows is what determines whether AI delivers a marginal ten per cent or a transformational multiple. Firms that treat data as an asset to be governed, rather than a by-product of getting the job done, will separate themselves from the rest.
Responsible adoption is not a brake. It is the enabler.
As co-chair of the RICS working group that wrote the Responsible Use of Artificial Intelligence in Surveying Practice standard and as Vice-Chair of the Professional Group Panel on Valuation, this is the question I care about most. The standard, which came into effect on 9 March 2026 and applies to all 150,000 chartered surveyors, gives the profession something it did not have a year ago: a clear basis for accountability and governance, sitting alongside wider frameworks such as ISO 42001.
What firms need to do today is straightforward to state and harder to do. Establish clear accountability, so that a named human remains responsible for professional output. AI assists; it does not absolve. Understand and document where and how AI is used across the workflow, so that it can be explained to a client or a regulator. Attend properly to data security and confidentiality, because client and asset data cannot be allowed to leak into tools that were never assessed for it. And train people not only on how to use these tools but on their limits, so that professional scepticism is applied to AI output rather than quietly assumed away.
The firms getting this right do not treat the standard as a compliance burden. They treat it as the thing that protects client trust and professional liability at the same time. In a profession whose entire value rests on trusted, accountable opinion, governance is not a brake on AI adoption. It is what makes responsible adoption possible at all.
The survey report is no longer the endpoint
For most of its history, the survey report has been a destination. Commissioned, delivered, read once, filed, and forgotten. That model is breaking down, and rightly so.
Survey data is increasingly the raw material feeding facilities management systems, lifecycle planning, compliance platforms, and portfolio analytics. The value of the survey now extends far beyond the document itself, and that reframes what a surveyor is actually producing. They are no longer producing a report. They are producing structured asset data that lives on, gets reused, and compounds in value across the entire life of the building or portfolio. A survey captured as a static PDF is worth a fraction of the same survey captured as structured, queryable data that flows into the systems managing that asset for the next twenty years.
The organisations pulling ahead are those treating their estate as a live data asset rather than a series of one-off inspections. As estates become more data-native, the relationship between surveying and asset management becomes continuous rather than transactional. The implication for surveyors is significant. Those who can produce data, and who understand how that data is used downstream will be far more valuable than those who deliver a beautifully written report that goes nowhere.
Where the money is, and where it is not
There has been significant investment in AI for design, engineering, and construction technology, with companies in that space attracting serious capital. Broadly, I agree that the money is concentrated at that end of the market, and the reason is not accidental.
Design and construction sit on rich, machine-readable data in CAD and BIM, with clearer inputs and outputs. They are solving more structured problems, which makes them more tractable and more obviously investable. Capital follows structured problems with measurable returns, and that is what it has found upstream.
The implication for surveying is double-edged. The risk is that our profession is treated as a downstream, lower-priority market, underserved by serious tooling built specifically for it. The opportunity sits in exactly the same place. The surveying-specific space that is being underinvested in is the structuring, governance, and intelligent reuse of asset and condition data, and the judgement layer that sits on top of it. Valuation, condition assessment, risk and liability-bearing opinion are harder to capitalise precisely because they are less structured and carry professional accountability. That is exactly why generic construction-tech tools will not serve them well, and exactly why the firms that build or adopt tooling aimed at that layer, rather than waiting for the construction wave to wash down to them, will capture the value.
The fee compression trap
There is a sentiment building among clients that, because their surveyors now use AI, fees should fall. The logic seems intuitive: if the work is faster, it should be cheaper. I think this is one of the most important commercial traps the profession faces, and surveyors should resist it firmly rather than quietly conceding it.
The error lies in misunderstanding what clients are actually paying for. They are not paying for the mundane tasks that surround a valuation, the drafting, the data gathering, the assembly. They are paying for the analytical rigour and the professional judgement of someone who has trained for twenty years to provide an accurate opinion of value, an opinion they can stand behind and, if necessary, defend in court. That expertise has not become cheaper because AI exists. If anything it has become more valuable, because it is now the scarce and defensible part of the service.
So the messaging needs to shift, deliberately and confidently. The position is not “we use AI, so we cost less.” It is “we deliver the same analytical rigour and depth of expertise you have always relied on, only far faster.” The judgement is unchanged. The liability the surveyor carries is unchanged. What has changed is speed, and speed is a benefit to the client, not a discount the client is owed. Firms that fail to make this argument will find themselves competing away the value of their own expertise, training clients to expect a premium service at a commodity price. The firms that hold the line and articulate clearly why they are holding it will protect both their margins and the perceived worth of the profession itself.
The business case is not as clean as the sector pretends
I want to add a note of realism that runs against the prevailing enthusiasm, because the profession is being sold a business case that is cleaner than reality.
The current pricing of AI is not a stable foundation to build on. The major AI firms are moving towards public markets, and that changes their incentives profoundly. The strategy until now has been to be first and best regardless of cost, to make the world fall in love with these models, to embed them as deeply as possible into how we all work. Once that embedding is complete and customers have little practical ability to switch, the commercial logic shifts towards raising prices and demonstrating returns to investors on a quarterly basis. It is classic oligopolistic behaviour, and firms that have built their entire operating model on today’s pricing will find themselves hostage to it.
There is a second dynamic that is widely misunderstood. Even if the cost per token falls over time, and it probably will, the number of tokens consumed is expanding exponentially. As models grow more capable and take on more complex tasks, usage explodes. The naive assumption that AI simply gets cheaper is wrong. One of the most important capabilities a firm can build is teaching its people to use the right model for the right complexity of task, and to question whether AI is the right tool at all. A great deal of AI use today is AI for its own sake rather than because it genuinely saves time.
So my counsel is this. Embrace AI where it makes sense, but be resilient and clear-eyed. In the short to medium term you may need to carry both your existing staffing overhead and a significant increase in spend on tokens, models, and the consultants needed to deploy and manage them. The efficiencies should arrive over time. But the business case is messier and more demanding than much of the sector is currently willing to admit, and firms that plan for a clean, linear return will be caught out.
What this actually requires, and where to start
The thread running through all of this is that the hard part was never the technology. It is the judgement about where to apply it, the discipline to fix the data and processes beneath it, the governance to do it responsibly, and the honesty to prepare people for genuine change. These are not problems you solve by buying a licence.
If you take one thing from this article, let it be that you do not need to solve all of it at once, and you certainly do not need to wait until every policy and migration is perfect before you begin. The firms that succeed start small and deliberately. A sensible first step is to pick one or two real, contained processes where the pain is obvious and the data is reasonably good, put a proper AI use policy and a few clear guardrails around them, train the people involved properly rather than handing them a licence and hoping, and measure honestly what changes. That is a contained, low-risk way to build the institutional muscle, the evidence, and the confidence to do more. It is the opposite of both reckless deployment and paralysed waiting, and it is entirely achievable within the next quarter.
This is the work we do at Artefact. We sit with boards and executive teams across the property sector and help them navigate precisely these questions: where AI creates real value and where it does not, how to build the data foundations that make it work, how to govern it responsibly under the RICS standard, and how to plan for a commercial reality that is more complex than the hype suggests. We are not a software vendor. We are technology-agnostic, which means our advice is shaped by what is right for the firm rather than by what we are trying to sell. And because we have done this across a range of property actors, we bring the comparative perspective that any single firm, working alone, simply cannot have.
The window for advantage is real, and it is narrowing. The firms that move now, properly rather than superficially, will define the next decade of this profession. The ones that mistake turning it on for turning it round will spend that decade wondering why the promise never arrived. The good news, for any leader feeling daunted by all of this, is that moving properly does not mean moving recklessly or all at once. It means starting, with clear eyes, now.
