The environment changed before the surveying profession did

The public started using AI over three years before the profession did. That gap is the source of almost everything below.

• Clients arrive already briefed. They ask a model what survey they need, what a defect means and whether your fee is fair, before they contact you.
• Search traffic is moving into AI answers. Some large surveying firms have lost around a quarter of their website traffic in six months. The question is no longer where you rank, it is whether you appear at all.
• Complaints and letters of claim now cost minutes to produce. Defending them costs more than ever, because volume is rising and each response has to be read line by line. That cost eventually lands on premiums, including those for firms that have never used AI.
• Larger clients in construction increasingly hold the data and the analysis. They want a surveyor to sign. The work reduces. The accountability does not.
• Most adoption inside firms is shallow. Faster report writing, same process underneath. The value sits in redesigning the process, and almost nobody has started there.
• Vendors are arriving weekly with unclear answers on data location, training use and what happens if they fold. The exposure sits with the firm, not the tool.
• Shadow AI is already inside firms that believe they have none.
• Agents are next. A chatbot is a conversation. An agent is closer to an employee you hand a task to. The oversight problem gets much harder.
• Graduates arrive fluent in firms with no tools and no policy. Leadership near retirement has a rational reason to change nothing. APC assessors are judging competence in something many have not used.

The opportunities are large and so are the risks. The difficulty is the speed. Nobody is keeping up, including me.

You do not have to adopt any of this. You do have to understand what it is doing to your clients, your exposure and your firm.

The public got AI tools first.

No procurement process. No training manual. No policy to write. Years of free access, and a population that now reaches for a model the way it used to reach for a search bar.

Surveying is somewhere near the start of its own AI adoption curve. Careful, regulated, slower by design, and in many firms still deciding whether any of this applies.

Both of those things are true at once, and the gap between them is where almost every problem I am seeing sits.

A firm can decline the technology. No firm can decline the environment.

This edition is the whole picture as I see it.

Clients turn up already briefed

The instruction decision is now partly made before anyone contacts you.

Take residential surveying. A homeowner asks a model what level of survey they need. A buyer describes a crack in a wall and receives an explanation, a likely cause and a list of questions to put to a surveyor. Somebody checks whether a quoted fee is reasonable. Somebody else asks what the difference is between two firms in the same town.

None of that used to happen, or it happened slowly through friends, agents and solicitors. It now happens in ninety seconds, at no cost, at any hour.

The output is not always right. That is almost beside the point. It arrives with confidence and structure, and it shapes what the client expects from you before you have spoken.

Whether you appear at all

Search traffic is being pulled into semantic AI answers – both on Google homepage and through AI chatbots such as Claude, ChatGPT, Gemini and Copilot.

For years the question was where a firm ranked on Google. The question now is whether the firm features at all in the answer a model assembles when somebody asks for a surveyor in their area, or for the difference between a chartered surveyor and an unregulated one.

This is not theoretical. I know large surveying firms that have seen website traffic fall by around a quarter in the last six months. A sustained drop, with nothing broken and nothing changed at their end.

The traffic did not go to a competitor. It went into an answer somebody read instead of clicking through.

If your website has been built to satisfy a search engine, it has been built for a system that is losing traffic. If your name, your discipline and your coverage area are not clearly stated somewhere a model can read them, you are absent from a growing share of the market’s first question. Above all else, your website needs to answer the questions people ask – not simply keywords.

Very few firms have looked at this. It is the least discussed item in this edition and one of the most consequential.

The cost to complain has collapsed. The cost to defend has not

Anyone with a grievance can now produce a structured, formal, professional-sounding letter in minutes.

I recorded an episode of This Is Surveying with Nick Rains, a senior contracts and wordings executive at Howden Insurance Brokers. The episode goes live tomorrow. Howden is a broker, and he was clear about that distinction throughout. What he described from the claims and wordings side is that the increase already visible is in complaints and letters of claim generated using free AI tools.

Read that with your own client list in mind. Nothing about it requires your firm to have used AI at all.

There is a second layer underneath. There is a difference between a complaint and a letter of claim. A complaint is an email saying the work looks wrong. A letter of claim is a formal step that starts a clock under the pre-action protocol. People asking a model how to complain about a professional are being handed the formal version without understanding what they have sent, and firms are receiving documents that trigger obligations they may not recognise.

Then look at what it costs to answer one.

A defence is prepared over weeks. Careful, considered, expensive. The claimant receives it, puts it through a free model and replies within hours. Every line of that reply still has to be read and assessed, because somewhere inside it one point might be correct and nobody can afford to assume otherwise.

So the volume of claims rises, and the work required to defend each one rises with it. Both sides of the equation are moving the same way.

The same pattern is being described in conveyancing and in parts of the legal profession, where claims volumes and the workload attached to them have climbed sharply.

That cost is being absorbed somewhere in the chain at the moment. It will not stay absorbed. It arrives eventually as premium, as excess, as a harder set of questions at renewal, and it will arrive for firms that have never used an AI tool in their lives.

The full conversation with Nick goes out tomorrow morning. It is worth a listen.

The chain above you

In parts of construction, larger clients now hold the data, run the analysis and produce the outputs themselves.

What they want from a surveyor is a signature.

The volume of work reduces. The accountability does not. Signing off somebody else’s analysis carries the same professional responsibility as producing it, and in some respects more, because the work you are attesting to was done in a system you did not build and cannot fully inspect.

That is a commercial shift and a liability shift at the same time.

Your competitors and the firms next to you

Adjacent businesses are using these tools to find, qualify and reach customers earlier in the process than you do.

Agents, brokers, conveyancers, developers. Some of them are now answering questions that used to arrive on your desk, and shaping the instruction before it reaches a surveyor.

Part two: what is happening inside firms

Most adoption is shallow

Where firms have started, they have mostly tried to make an existing process faster. Report writing above all.

That is a real saving and I understand why it is where people begin. It also bakes in every assumption already sitting inside the process.

The value that matters is further back. If a task that took two hours now takes twenty minutes, the question is not what to do with the ninety minutes. The question is whether the process around that task still makes sense, whether the service should be priced the same way, whether the roles in the firm should be arranged as they are.

Very few firms have started there. Almost everyone has started with the report.

Smaller firms can move faster

This is one of the few structural advantages the small firm has in this cycle, and it is being wasted.

A sole practitioner or a five-person firm can trial a tool, decide it does not work and abandon it within a fortnight. A large firm needs a business case, an IT review, a procurement cycle and a steering group.

Speed of iteration is worth more than budget at this stage. Firms that assume scale wins here have the shape of it backwards.

The vendor problem

New tools arrive weekly. Some are excellent. Many are a thin layer over somebody else’s model, built on rented infrastructure, with no clear answer on where data is held, who can access it, what it is used for and what happens to your records when the company folds.

I covered this in the last AI briefing, and the questions are not complicated. Where is our data stored? Is it used to train anything? Who has access? What happens on termination or insolvency? Can we get it all back in a usable form?

What is missing is not the questions. What is missing is anyone asking them in writing before signing.

The exposure sits with the firm. Vendors will tell you the tool is compliant. Compliance is a property of what your firm does, not a property of what you bought.

Shadow AI

Use is already inside firms that believe they have none.

Somebody is drafting with a free model at home. Somebody has a browser extension. Somebody is using the AI features that arrived inside software the firm has had for years without anyone deciding to switch them on.

A firm cannot govern what it has not looked for. Most firms have not looked.

Agents are the next wave

The tools most people have used so far are conversations. You ask, it answers, you check the answer, you decide what to do with it.

An agent is closer to an employee.

You give it a task rather than a question. It goes away, works out its own sequence of steps, gathers what it needs, makes decisions along the route and comes back with the job done. It is autonomous within the boundaries you set, and it finds its own solutions rather than waiting to be told the next move.

Surveying is barely at this stage. I am learning how to use agents at the moment, as this is the next big wave in AI evolution and use.

What I will say is that the risk profile changes completely. A tool that drafts a paragraph hands you something to check. A system that carries out ten steps and reports back hands you an outcome, with the reasoning distributed across steps nobody watched.

The oversight problem becomes considerably harder. The professional obligation to supervise does not soften to meet it.

Longer term, I think most people will have their own agents working alongside them, in the same way most people now have a phone. That is where this goes. The firms that will handle it well are the ones that started thinking about oversight while the tools were still only holding conversations.

Part three: the people

Graduates arrive AI literate

Students have used these tools throughout their degrees and their training. Not occasionally. Throughout.

They are joining firms with no AI tools, no policy and no appetite for either, and they are frustrated. Some will comply. Some will use their own tools without telling anyone, which returns us to shadow AI. Some will go elsewhere. I see this area being one way of firms needing to adopt AI sooner than they would like. Future generations are growing up with these tools, and they expect to use them. Watch how this shapes and influences recruitment.

Leadership is looking the other way

We have an ageing profession and an ageing leadership layer.

If retirement is five years away, the rational move is to change nothing. I understand the logic completely, and I do not think it is laziness. It is a reasonable calculation about time, energy and payback.

It is also the calculation that leaves a firm with no answer for the next generation, no position on the technology and no sale value to a buyer who will ask about all of this.

Many of the people making that calculation have not used any of these tools

APC assessors are being asked to judge competence in an area many have not used themselves.

The competency framework moves at the speed of committees. The technology does not. That gap is widening, and the people carrying the cost of it are the candidates.

Where this leaves us

Pull all of that back to where it started.

The public adopted these tools years before the profession did, and that single gap explains the client who arrives pre-briefed, the complaint that costs nothing to send, the search result you no longer appear in, and the graduate who cannot understand why the firm works this way.

Layered on top of that: vendors selling faster than firms can assess them, adoption that mostly makes old processes quicker, use already happening inside firms that have not looked for it, agents arriving next with a harder oversight problem than anything so far, and a leadership layer with a rational reason to wait it out.

The opportunities in this are genuine and large. So are the risks. What makes it difficult is the speed.

Professions are built to move slowly. That is a feature. Standards, competence, precedent and accountability all depend on deliberate pace, and surveying is slow to adopt new technology.

Nobody is keeping up.

Nobody is, and the honest position is that everyone is at a different point on this curve and working with a partial view.

None of that changes the obligation.

You do not have to adopt any of this. You do have to understand it. Understand what it is doing to your clients, your claims exposure, your visibility, your data, your staff and your pipeline, because every one of those is being changed by decisions other people are making.
That is the whole argument. The environment changed before the profession did, and it is not waiting.

If you want to keep up with other AI developments across surveying and free AI briefings, join the AI in Surveying newsletter here

Until next week, readers 🙂

Nina