AI in Estate Agency: Why Experience Still Counts

Over the past year, artificial intelligence has become part of everyday working life at Noble Estates. It has changed how we research, how we prepare valuations and how we market property. I have worked in property for 36 years and have seen a great deal of new technology come and go. This is different. It is a genuine step change, and we are making full use of it.

It does not, however, sell or let properties. People do. This article explains how we use AI, where it is heading, and why experience remains at the centre of what we do.

Which AI tools do we use?

We use Claude, made by Anthropic, for most of our work. ChatGPT and Google’s Gemini sit alongside it.

Claude does the heavy lifting. We use it to read long documents, analyse figures and prepare reports. In a typical week that might include reading residential and commercial leases to establish exactly what is let, what the freeholder retains and what obligations a buyer or tenant will take on, working through a prospective tenant’s published accounts before we recommend them to a landlord, comparing builders’ quotes line by line for a client, or pulling together sold prices, rental evidence and EPC floor areas for a portfolio valuation.

ChatGPT and Gemini are useful for quick drafting and editorial work and, in Gemini’s case, for anything connected to Google’s own services. We also use them to cross-check one another. Where two tools reach different answers, that is usually a sign that a person needs to look more closely.

Every piece of work is checked by a member of our team before it reaches a client. The tools are fast, but they can be wrong, and they do not carry our professional responsibility. We do.

Where AI is making a difference

Valuations. Gathering comparable evidence used to take hours. We can now review every recorded sale on a street, current asking prices and rental evidence, and test the figures on a price-per-square-foot basis, in a fraction of the time. The analysis is broader, but the final figure is still a matter of judgement. Recently, checking EPC floor areas on a portfolio showed that several units were smaller than their descriptions suggested. The tools found the data; it took experience to see what it meant for the price.

Property details and marketing. Our process is straightforward. We write the original copy ourselves, having been round the property. AI then helps us refine it, and we make the final amendments before anything is published. The facts always come from us: room sizes, tenure, lease length, service charges and the condition of the property.

This matters. Some agents, whether unscrupulous or simply lazy, use AI to enhance photographs and generate descriptions without checking the details. One well-known franchise agency was recently criticised in the national press for this kind of practice. A buyer who arrives to find the property does not match the photographs or the particulars is unlikely to proceed, and the agent has wasted everyone’s time. We will not market a property that way.

Our website, SEO and GEO. Buyers, sellers and landlords increasingly ask AI assistants for recommendations rather than searching Google. Generative engine optimisation (GEO) is the practice of making sure we appear in those answers as well as in search results. We have brought this work in-house, and AI tools help us keep our website current, accurate and visible.

Research. Planning history, building regulations, lease terms, company accounts and market data can now be reviewed in hours rather than days. For clients, that means clearer advice, sooner.

Property management: the next big change

Some of the most promising work is happening in property management. A number of new companies are building their services around AI, and I expect them to change that industry considerably.

The need is real. Around a quarter of my sales are delayed while we wait for managing agents to provide service charge information and details of section 20 consultations for major works. Those replies can take weeks, and sometimes months. During that time buyers lose patience, mortgage offers approach expiry and chains come under strain. Managing agents using AI well should be able to answer the same enquiries in days. For leaseholders, and for everyone in a chain, that would be a significant improvement.

Software is only part of the answer

AI is not the be-all and end-all. Our agency software, Eurolink’s Veco Plus, now integrates AI, and that is welcome. But the real benefit comes from the workflows we build around it.

A workflow sets out each step in a process, who is responsible for it and when it should happen. In Veco Plus we use workflows to manage a new instruction from valuation to launch, to make sure every viewing receives feedback, to record and confirm offers, and to track each sale from memorandum of sale to completion. Tasks are prompted automatically, so nothing waits on someone’s memory. On a leasehold sale, for example, the workflow prompts us to request the managing agent’s information on the day the sale is agreed rather than weeks later.

Software and AI support those processes. They do not replace them, and the results still depend on the people using the system and how well they understand it. Good software used within a poor process simply produces mistakes more quickly.

What AI cannot do

Buying or selling a home is still a very personal experience, and the parts that matter most depend on people. AI cannot negotiate across a table, or judge when to hold firm and when to bring two parties together. It has not stood in hundreds of Victorian terraces and conversions, and it cannot tell a client from experience which survey findings matter and which do not. It cannot guide a family selling a home of forty years through a legal process that is rarely straightforward, and it cannot take responsibility for the advice it gives.

Our team has more than 100 years of combined experience in the property industry, and we are one of the most experienced agents south of the river. That experience is what our clients pay for.

A lesson from Purplebricks

Software has been used before to replace the agent and reduce costs. Purplebricks set out to replace the traditional agent with a low-cost model that was mainly online and largely agent-free. Despite heavy advertising and significant investment, the model did not deliver, and in 2023 the business was sold to its rival Strike for £1. Technology can make a good agent better. It does not make the agent unnecessary.

The figure that matters most

Nationally, around one in four agreed sales fails before completion. TwentyEA put the figure at 23.7% in the first quarter of 2026, and at 27% in Inner London. At Noble Estates, our fall-through rate is less than 8%.

That difference does not come from software. It comes from preparing sellers properly, checking buyers carefully, spotting lease, survey and building control issues before they surface, and staying in close contact with both sets of solicitors until completion.

In summary

AI is an excellent set of tools, and we intend to make full use of everything it offers. It helps us work faster and more thoroughly, and it frees our team to spend more time with clients.

For most people, their home is their biggest asset. The one factor that can maximise the return on it is an experienced agent. Negotiation, building knowledge, the legal process and the personal side of moving home still need people who have done it many times before. AI is a valuable tool, but it does not replace a decent agent.

If you are thinking of selling or letting in Clapham, Battersea, Balham or the surrounding areas, please contact either of our offices:

• Clapham Old Town: 020 7498 8600
• Northcote Road: 020 7228 9265
• Email: info@noble-estates.co.uk

Nick Goble MNAEA MARLA AssocRICS
Founder, Noble Estates

Sources: TwentyEA via PropertyWire; The Negotiator.