SVENNISAIArtificial intelligence (AI) for estate agents: how to improve your residential sales processes
See how AI can be applied to the real processes of a UK estate agency: portal enquiries, viewings, valuations, particulars, sales progression and AML.
Book a 30-minute call · with a Claude Certified Architect from our team.
- Portal enquiries: answered while the buyer is still looking, not when the branch reopensAn enquiry about a house for sale seldom arrives while the branch is open. Buyers scroll Rightmove from the sofa in the…
- Viewings that happen: booked, confirmed and coordinated around the people who live thereA viewing is where an online shortlist becomes a genuine buyer, and getting the right people through the door quickly is…
- Buyer qualification: capturing a buyer's position to proceed in the first conversationAgreeing a sale is not the same as completing one. Industry trackers put the share of agreed UK sales that collapsed bef…
9 processes across 5 areas of the business, each with an example below.
The processes that are specific to an estate agency
Here we get into the detail: the real day-to-day workflows of a UK estate agency, from the Rightmove enquiry landing overnight to the AML checks before a sale is agreed.
UK homes changed hands at roughly 100,000 completions a month into 2026, and the agent who replies first is usually the one who gets the viewing. Each process comes with examples and an honest view of the technology.
Portal enquiries: answered while the buyer is still looking, not when the branch reopens
An enquiry about a house for sale seldom arrives while the branch is open. Buyers scroll Rightmove from the sofa in the evening, line up their shortlists at the weekend, and message or ring long after the office has locked up. One newly listed property can throw off a burst of near-identical questions within hours, is it still available, when can I view, has it had an offer, and most of them land out of hours.
Where those enquiries arrive is not really in doubt. Rightmove held a record 89% share of all time spent on UK property portals at the end of 2025, with consumers spending 16.8 billion minutes researching the market across the year. For a UK estate agent, enquiry handling in practice means Rightmove lead handling first, then the agency's own website and the phone. Whatever answers your enquiries has to plug into that portal lead flow, not a generic contact form.
The load behind it is steady and high. UK residential transactions ran at roughly 100,000 a month into 2026, with a provisional 98,450 completions in May 2026, up 17% on the year. That means a continuous stream of buyers enquiring across evenings and weekends, when nobody is at the desk. The same buyer is usually enquiring on several listings at once, so the agent who replies first, even out of hours, is generally the one who gets the viewing.
An assistant built for your agency sits exactly in that overnight gap. It reads the Rightmove enquiry, checks the actual listing, answers availability and the obvious questions from live data, and proposes viewing slots, turning a burst of overnight messages into a clean shortlist rather than a Monday inbox backlog. Not a canned autoresponder, but a natural conversation on your rules: what it may confirm, when it hands to a negotiator, what it records.
First reply to Rightmove enquiries, in minutes rather than the next working day
The assistant is connected to your Rightmove lead flow and your live listings. When an enquiry lands against a property, it replies at once: it confirms the listing is still available and on the market, answers questions on tenure, council tax band and chain from the listing data, and proposes the next step, a viewing. On anything it cannot verify, it says so and passes the conversation to a negotiator.
At half past nine on a Sunday evening a buyer asks about a three-bed semi in York: still for sale, and could they view at the weekend? The assistant confirms availability from the live listing, answers the tenure and council tax band, and offers two Saturday slots. On Monday the negotiator opens a booked viewing, not a two-day-old email.
No enquiry cools overnight or over a weekend. The buyer gets a solid answer while the property is still fresh in their mind, and the branch starts the week with viewings in the diary instead of a queue in the inbox.
Website and WhatsApp enquiries treated as first-class channels
Through your website chat and WhatsApp, the assistant runs the same conversation it runs on portal leads: natural language, the key facts from the listing, follow-up questions on budget, timescale and what the buyer is looking for. It recognises which property is meant even when someone only sends a link, and it files every exchange as a structured enquiry rather than leaving it buried in one negotiator's phone.
A buyer sends the link to a terraced house and asks whether the garden faces south and if there is a chain. The assistant answers both from the listing, notes that they are first-time buyers, and proposes a viewing. By the time a negotiator picks it up, the buyer's position is already on the record.
Enquiries that used to end as an unread notification become documented leads with the buyer's position attached, whichever channel they arrived on.
An out-of-hours phone assistant for the calls nobody can take
An AI phone assistant answers when the team cannot, after closing, at weekends, at busy moments in the branch. It introduces itself as an AI assistant, establishes which property and what the caller wants, takes callback details and can offer a viewing slot on the spot. Every call becomes a structured note against the enquiry, not a silent entry in the missed-call list.
On a Saturday afternoon a buyer rings about a York listing while both negotiators are out on viewings. The assistant confirms the property is available, notes the wish for a weekday evening viewing and a callback, and books it in. The negotiator returns the call to a named, prepared prospect.
A published number that goes unanswered is a buyer moving on to the next agent. An answered call with a captured callback keeps that buyer in your pipeline.
Structured capture, so negotiators open a prepared conversation
While it chats, the assistant gathers what a good first call would gather: whether the buyer has anything to sell, the rough budget, the timescale, whether funding is in place. It writes a short structured summary against the enquiry and flags the conversations where a person should take over, so negotiator time goes where it earns most.
A buyer asking about the York semi mentions they are renting, have an agreement in principle and want to move within three months. The negotiator sees that position before dialling and opens with a plan rather than with questions.
The portal sends enquiries in bursts. Structured capture turns that volume into booked viewings and sends selling hours to the buyers who are actually ready to proceed.
The technology is ready, and the UK's own market data shows where it earns its keep.
- The integration point is settled. With Rightmove holding an 89% share of portal time and 16.8 billion minutes of research across 2025, a UK agency's enquiry handling is Rightmove lead handling first, then its own website and phone. An assistant that reads the enquiry, checks the live listing and proposes a viewing can be built today on your listing data and your rules.Rightmove plc
- The load is real and it never stops. UK transactions ran at around 100,000 a month into 2026, a provisional 98,450 in May alone, and those buyers enquire across evenings and weekends when the branch is shut. An assistant that answers in minutes from live data, at any hour, is a buildable system now, not a promise.GOV.UK / HMRC
- This sits at the very top of the funnel. A home took 60 days on average to find a buyer in June 2026, and more than a third of new listings did not go on to sell at all, so whether the first wave of enquiries is answered fast, from live data, and turned into booked viewings is often what separates a listing that sells from one that stalls.Rightmove
- One number deserves scepticism before you build. We found no neutral UK statistic for how many enquiries arrive out of hours, nor for agent response times, so treat any borrowed figure as a compass, not a promise. The figures that settle it are ones you already own: time to first reply, viewings booked per enquiry, and how many weekend enquiries are still live on Monday, each measured before and after.
Three points belong in the build from the first day.
- Honesty rules are technology-neutral. There is no UK AI Act and no blanket duty to disclose AI use, but the ASA is explicit that its codes apply however content is generated, so the assistant introduces itself as an AI assistant and answers availability and price only from the live listing, never a guess.ASA
- Replying to an enquiry is solicited; following up with marketing is not. Under PECR regulation 22, enforced by the ICO, unsolicited direct marketing by electronic mail needs prior consent or the soft opt-in, with a working opt-out in every message, so the assistant keeps each contact's consent status attached.legislation.gov.uk
- Names, numbers and property circumstances in an enquiry are personal data under the UK GDPR, as amended by the Data (Use and Access) Act 2025. An external chat or AI provider processes them on your behalf, so a processor contract and a clear retention schedule are part of the build, not an afterthought.legislation.gov.uk
Viewings that happen: booked, confirmed and coordinated around the people who live there
A viewing is where an online shortlist becomes a genuine buyer, and getting the right people through the door quickly is largely a scheduling problem. How fast a request is booked, how reliably it is confirmed, whether a reminder goes out, and how easily a slot can be moved all decide whether the diary holds up on a busy Saturday.
Speed matters because the market rewards it. A home took 60 days on average to find a buyer in June 2026, and Rightmove found that homes which did not need a price reduction found a buyer far faster, around 36 days, than those that did, around 127. Getting qualified buyers in front of a property early is what compresses that average, and viewings are the mechanical layer where time is either saved or quietly lost.
In a busy branch a negotiator juggles several viewings a day, coordinated around sellers who still live in the property, applicants who work office hours, and the diary of whoever accompanies the viewing. With transactions running near 100,000 a month through the spring and summer selling season, the double bookings, gaps and no-shows that eat field time multiply exactly when the team can least afford them.
An AI assistant, built to your agency's rules, can own that stretch. It takes the request at any hour, checks the negotiator diary and the occupier's stated availability, confirms in writing, sends the reminder, offers a one-tap reschedule and refills a cancelled slot from a waiting list. Your negotiators take over where it counts, standing in the property with the buyer.
Viewing requests taken around the clock, Sunday evening included
The assistant is connected to your listings and diaries: it knows which property is available, which negotiator can accompany the viewing and how long to allow. When a buyer asks for a slot on your website or over WhatsApp, it checks availability, offers two or three concrete times within the windows the occupier has agreed, books the chosen one and confirms in writing. Anything ambiguous goes to a negotiator with the full chat attached.
A buyer messages on Sunday evening about a house on Cornwall Road in Harrogate and asks for a Saturday morning viewing. The assistant offers 10am or 11am, both inside the times the sellers have agreed, books 10am and notes the buyer is in a chain. Monday morning the negotiator finds a ready appointment, not an unread message.
No viewing request waits for the next working day, and no buyer is offered a time the occupier never agreed to. The first impression of your agency is a prompt, firm answer.
Confirmations and reminders that keep the diary honest
Straight after booking, the buyer receives a written confirmation with date, time, property and who they will meet. The day before, a short reminder follows with the option to confirm or move the viewing by replying. If the buyer cancels, the assistant offers alternatives at once; if a reminder goes unanswered, it sends a single follow-up and then frees the time for someone else. Every message concerns the viewing the buyer asked for; nothing promotional is added unless that contact's marketing consent is on record.
The reminder for a Saturday 10am viewing goes out on Thursday evening. The buyer replies that 11.30am suits better. The assistant moves the booking, checks the new time against the sellers' agreed window, and offers the freed 10am slot to the next person waiting on the same house.
A viewing that would have been a silent no-show turns into a rebooking made in good time, and the Saturday diary reflects what will actually happen on Saturday.
The occupier kept in the loop, not booked over
Because most viewings are at a home someone still lives in, the workflow keeps the occupier's confirmation inside it. The assistant only offers times the seller or sitting occupier has agreed, gives reasonable notice, and where a request falls outside the agreed windows it asks the occupier before promising anything. The occupier can see and confirm each viewing rather than discovering strangers on the doorstep.
A buyer asks to view at short notice on a weekday evening. The time is outside the sellers' agreed windows, so the assistant tells the buyer it will check, messages the sellers, and only confirms once they agree. Nobody is booked in over the occupier's head.
Access is arranged with the people who live there, which keeps sellers on side and viewings running smoothly, exactly where a rushed booking usually goes wrong.
Freed Saturday slots handed to your most proceedable buyers
For in-demand homes the assistant runs a priority queue rather than a plain list. When a viewing is cancelled or lapses after the check-in message, it offers the open slot first to the buyers most likely to proceed, cash buyers and chain-free buyers, then those with their own sale already agreed, before working down to the rest, all inside the occupier's agreed windows. On open-house or block-viewing days it fills the remaining times the same way. You decide which homes use a priority queue and at what point a negotiator steps in.
A chain-free buyer drops out of Saturday's 11am viewing on a sought-after Harrogate townhouse. The assistant offers the time straight to a cash buyer who enquired on Thursday and is ready to move, confirms it within twenty minutes, and keeps the sellers posted. The branch's best Saturday window goes to the buyer most likely to turn it into an offer.
The most valuable window in the week is filled by the buyer likeliest to complete, not simply the next name on a list. A recovered slot becomes a proceedable viewing, and a home that sells sooner frees the chain sooner.
Solid ground for viewing scheduling today, with one figure to hold lightly.
- An assistant that reads a naturally phrased request, checks the negotiator diary and the occupier's agreed windows and writes the viewing straight into your systems is within reach on your current systems. Speed is the lever it pulls: a home took 60 days on average to find a buyer in June 2026, so getting qualified buyers through the door early is what compresses that average, and confirmations, reminders and a one-tap reschedule are the mechanical layer software owns well.Rightmove
- The season concentrates the load. With UK completions running near 100,000 a month through spring and summer, a negotiator juggles many viewings a day around sellers, applicants and whoever accompanies the viewing. Coordinated scheduling with confirmations and reminders reduces the double bookings and gaps that bleed hours out of the field day, and that coordination is straightforward to build on your diaries.GOV.UK / HMRC
- A light qualification step before a slot is offered protects the most expensive resource, time on the road. More than a third of new listings in June 2026 did not go on to sell, so a viewing on a buyer with no chance is a direct cost, and the assistant proposes times only after it has captured the basics.Rightmove
- One honest reservation on the numbers. Rightmove's finding that priced-right homes sold in around 36 days against 127 for those that needed a reduction is its own correlation on its own platform, useful as a direction rather than a guarantee for your branch. Prove it on your own figures: time from request to viewing, viewings booked per enquiry, and your show-up rate, each before and after.Rightmove
Build these limits in from the outset.
- Viewings at an occupied home rest on the occupier's agreement. The assistant can propose and hold slots, but access must be arranged with the seller or sitting occupier, with reasonable notice, and never booked over their head, so the occupier confirmation step stays inside the workflow.
- Booking confirmations and reminders about a viewing the buyer asked for are service messages. Layer anything promotional on top, other properties or referrals, and PECR regulation 22 applies, so keep each contact's consent status attached and a working opt-out on every marketing message.legislation.gov.uk
- Be honest where the data is thin. There is no neutral UK statistic for viewing no-show rates worth quoting, so promise the mechanism, confirmation, reminder and one-tap reschedule, and measure your own show-up rate rather than borrowing a percentage from a vendor.
Buyer qualification: capturing a buyer's position to proceed in the first conversation
Agreeing a sale is not the same as completing one. Industry trackers put the share of agreed UK sales that collapsed before completion in 2025 at around a quarter, with mortgage finance, survey findings and broken chains the leading causes. Most of those causes are visible right at the offer stage, if someone captures the buyer's position.
That position is a short, specific set of facts: whether the buyer has a mortgage agreement in principle, their funding route, whether they are in a chain and how long it is, whether they are a cash buyer or a first-time buyer with nothing to sell, and their timescale. A busy negotiator taking a dozen enquiries a day often cannot gather this consistently. An assistant can, politely, every time.
Qualification is not a knockout quiz, and it must never become one. A home took 60 days on average to find a buyer in June 2026, so a proceedable buyer is worth qualifying for, not turning away. The point is to capture the facts that decide whether a viewing becomes a sale that completes, then route a proceedable buyer straight to a viewing and hand the negotiator a clear picture.
An AI assistant, built to your agency's rules, does exactly that and no more. It gathers the position in a natural conversation, passes structured notes to a person, and stays firmly on the right side of the law: it never advises on a mortgage and never filters an offer or a buyer. It prepares; your negotiator and the seller decide.
The buyer's position captured, consistently, in the first conversation
As it answers an enquiry, the assistant asks the questions that decide whether a sale will hold: funding route, agreement in principle, chain position and length, anything to sell, and timescale. It records the answers as structured fields against the enquiry, in the same format every time, so the negotiator reads a clear position rather than piecing it together from a phone call they half remember.
A buyer enquiring about a terraced house in Norwich mentions they are first-time buyers with an agreement in principle and nothing to sell, hoping to move within three months. The assistant files that as a proceedable position and books a viewing, and the negotiator opens the file already knowing this is a clean buyer.
The facts that predict a fall-through are on the record from the first exchange, so viewings and offers can be prioritised by who can actually proceed, not by who happened to ring loudest.
Proceedable buyers routed straight to a viewing
When the captured position is strong, the assistant does not sit on it. It proposes viewing times at once and flags the buyer to the negotiator as proceedable, so a good buyer is not left waiting behind an inbox backlog. Where the position is weaker or unclear, it still books politely and notes what is outstanding, without ever refusing anyone.
A cash buyer with no chain asks about a property on a Sunday night. The assistant captures the position, offers two viewing slots and marks the enquiry as a priority. The negotiator starts Monday with a strong buyer already in the diary.
A proceedable buyer reaches a viewing faster, which shortens the whole journey in a market where the average home already takes 60 days to find a buyer.
Every offer and buyer passed to a person, never filtered
The assistant captures position, but it does not judge offers or screen buyers out. Anything that looks like an offer is passed straight to the negotiator, in writing, with the buyer's details, because the law requires every offer to reach the seller promptly. The assistant's role is to gather and route, and the human stays in the loop on anything that touches price or the seller.
A buyer says they would offer a figure below the asking price, subject to survey. The assistant does not accept, decline or comment. It records the wording exactly, marks it as an offer and notifies the negotiator the same evening, so it reaches the seller promptly and in writing.
You get consistent capture and fast routing without any risk of an offer being quietly missed or a buyer being screened out, which keeps the agency on the right side of its legal duties.
A prepared handover, not a cold call
From the conversation the assistant builds a short brief for the negotiator: the property, the captured position, the questions already asked and anything outstanding. Sensitive details the buyer shares, funding and deposit, are handled as financial personal data under a clear retention rule, and the negotiator picks up a prepared, documented conversation.
Before returning a call, the negotiator reads that the buyer has an agreement in principle, a property to sell that is not yet on the market, and a three-month timescale. The call opens with a plan for the chain rather than with a blank appraisal of the buyer.
Negotiator time goes into advising and progressing the buyers who can proceed, instead of re-collecting a position a conversation already captured.
What can be built for buyer qualification today, and the one figure to treat with care.
- An assistant that captures a buyer's position to proceed, funding route, agreement in principle, chain position, anything to sell and timescale, and records it as structured data is ordinary engineering today. Most of the causes behind a collapsed sale, mortgage, survey and chain problems, are visible at the offer stage, so gathering that position consistently in the first conversation is exactly the discipline a busy negotiator often cannot keep.ReadySteadySell
- It shortens the journey rather than lengthening it. With the average home taking 60 days to find a buyer in June 2026, a proceedable buyer is worth qualifying for, and an assistant that gathers the facts politely and routes a clean buyer straight to a viewing moves the whole transaction along faster.Rightmove
- The build has a firm boundary drawn into it. Capturing whether a buyer has an agreement in principle is fine; advising on the mortgage or judging an offer is not, so the assistant gathers and routes while a person decides. Under the Estate Agents Act 1979 every offer must be communicated to the seller promptly, and that duty is a feature of the design, not a limitation to work around.legislation.gov.uk
- Hold the headline figure loosely. The roughly 24% to 25% fall-through rate for 2025 is an industry estimate from property trackers, not an official national statistic, and it varies by region. Use it as a reason to capture position well, then measure your own numbers, your fall-through rate and how many qualified buyers reach a viewing, before and after.ReadySteadySell
Three hard limits shape this build.
- The firm line for AI: arranging or advising on mortgages is an FCA-regulated activity. The assistant may capture whether a buyer has an agreement in principle and their funding route, but it must not advise on, recommend or quote a mortgage, which would be a financial promotion for the authorised firm to make, not a chatbot.FCA
- The Estate Agents Act 1979 requires every offer to be communicated to the seller promptly and in writing, with no discrimination between buyers. A qualification tool must therefore never suppress an offer or screen a buyer out; every offer it captures has to reach a person and the seller.legislation.gov.uk
- Funding position, deposit and chain details gathered in chat are personal and financially sensitive data. An external AI provider processing them needs a written processor contract under Article 28 of the UK GDPR and a defined retention schedule, set up as part of the build.legislation.gov.uk
Valuations and instructions: a defensible asking price, evidenced from the first appraisal
The market appraisal is where a fee is won or lost, and the decision that carries it is the asking price. Price to sell on day one and the home moves: Rightmove found that properties needing no reduction found a buyer in around 36 days, against around 127 days for those that had to be cut. Win the instruction on a flattering figure that later fails, and you have tied up marketing, disappointed a vendor and added a stale listing to the branch numbers.
The raw material for a defensible price is already public and free. The UK House Price Index is official monthly data from HM Land Registry, built from completed sale prices across England, Wales, Scotland and Northern Ireland. What most agencies lack is not the data but the time to turn it, together with current portal competition, into a per-property recommendation at every appraisal.
The risk of getting it wrong runs one way in this market. More than a third of new listings coming to market in June 2026 were not going on to sell, so an over-optimistic instruction is rarely a harmless punt: it costs months of marketing and erodes the vendor relationship before the inevitable reduction.
An AI assistant, built for your agency, steps into exactly that gap. It assembles the comparable evidence, drafts an asking-price range with the sold prices and portal competition behind it, and answers the 'what is my house worth' enquiry the moment it lands. It never commits a firm figure unseen: the range is explicitly subject to a physical appraisal, and the valuer decides.
Comparable evidence assembled before the appraisal
The assistant pulls completed sale prices for the street and the immediate area from the UK House Price Index and Land Registry records, filters them to the closest matches by property type, size and date, and lays the current portal competition alongside. The valuer walks into the appointment with the evidence already sorted, rather than assembling it by hand the night before.
A three-bed semi comes in for appraisal in Meanwood, Leeds. Before the visit the assistant has the six closest sold comparables from the last twelve months, the three similar homes currently listed nearby, and the average time each has stood. The valuer arrives with a picture of the local market, not a blank sheet.
Every appraisal starts from the same evidenced baseline, so the recommendation rests on completed sale prices and live competition rather than gut feel. The valuer spends the appointment advising the vendor, not gathering data.
A defensible asking-price range, with the comparables behind it
From the assembled evidence the assistant drafts a price range and the short rationale behind it: which comparables, how they differ, what the current competition is asking. Nothing is presented as a fixed figure, and the valuer signs off the range before it reaches the vendor. The output is a recommendation the agency can stand behind, not a single number pulled from the air.
For the Meanwood semi the assistant proposes a range with the extended kitchen weighed against two comparables that also had extensions, and flags one nearby listing that has sat for eleven weeks at an optimistic figure. The valuer adjusts, approves, and presents the vendor a price with the reasoning attached.
Pricing to sell on day one becomes a disciplined, evidenced conversation. When a vendor pushes for more, the comparables are on the table, and the agency wins the right instruction rather than an unsellable one.
The valuation enquiry answered as a lead, round the clock
The 'what is my house worth' enquiry rarely arrives in office hours. On the website or over WhatsApp the assistant takes it at any time: it captures the address, property type and the vendor's rough timeframe, gives an indicative range drawn from sold comparables, states plainly that a physical appraisal follows, and books the market appraisal. Every enquiry is filed as a qualified valuation lead.
Late on a Sunday evening a homeowner asks what their semi might fetch, mentioning a recent kitchen extension. The assistant gives an evidenced indicative range, explains a valuer confirms it in person, and offers two appraisal slots for the week. Monday morning opens with a booked appraisal, not a missed enquiry.
The biggest self-service question in residential sales gets answered on your own website rather than someone else's. A valuation request becomes a booked appraisal with the vendor's timeframe already on record.
Stale instructions flagged for a reduction conversation
The assistant keeps a running view of every live instruction: days on the market, enquiry and viewing count, and how the asking price now sits against fresh comparables and current competition. When a listing drifts, it prepares the reduction conversation for the negotiator, with the evidence for a specific new figure, rather than leaving it to whoever next notices.
A listing in the same branch passes eight weeks with viewings tailing off. The assistant shows two comparable homes that have since sold below the asking price and proposes a reduction to a range that matches them. The negotiator calls the vendor with evidence, not just a suggestion to drop the price.
No instruction quietly ages into a stale listing. Reductions are proposed early and backed by comparables, which protects the relationship with the vendor and keeps the pipeline moving.
The ground here is solid, and the UK's own market data shows where the discipline pays.
- An assistant that reads completed sale prices from the UK House Price Index and Land Registry records, filters them to the closest comparables and drafts an asking-price range with the reasoning attached is well within reach today. The data is official, monthly and free; what gets built for you is the layer that turns it, plus live portal competition, into a per-property recommendation at every appraisal.GOV.UK
- Pricing to sell on day one is not a slogan, it is measurable: Rightmove found homes that needed no reduction took around 36 days to find a buyer, against around 127 days for those that had to be cut. An evidenced range at instruction, presented with its comparables, is exactly what protects that first-day position.Rightmove
- The same discipline answers the market's biggest self-service question. With more than a third of new June 2026 listings not going on to sell, winning the right instruction matters more than winning every instruction, and an assistant that captures the valuation enquiry as a qualified lead and books the appraisal is a build you can start today on your own website.Rightmove
- Keep a level head about the vendor numbers. Accuracy and conversion claims made for instant online valuation tools tend to come from other markets and from the firms selling them, so treat them as a direction of travel rather than a promise for your branch. Judge this on three figures of your own: time to find a buyer, the share of your instructions that sell without a reduction, and instruction to completion, each measured before and after.Rightmove
A valuation steers what a seller does next, so three limits are fixed before the assistant offers a figure.
- An automated valuation is an estimate that cannot see condition, recent improvements or local demand, so it must present a range with its comparables, state that a physical appraisal follows, and never commit a firm figure on a home nobody from the agency has seen. The valuer, not the model, sets the price.
- Any valuation or pricing figure shown to a consumer must not mislead. Under the Digital Markets, Competition and Consumers Act 2024, in force since 6 April 2025, misleading actions and omissions are unfair commercial practices the CMA can act on directly, so an indicative range has to be honest about what it is and what still needs checking.GOV.UK
- A property address tied to an owner and their circumstances is personal data under the UK GDPR. An external valuation or AI service that reads it is a processor, which means a written contract, a lawful basis and a retention schedule are part of the build, not an afterthought.legislation.gov.uk
Listing particulars: complete, material-information-compliant, ready for the portal
Complete particulars are now a legal duty, not a finishing touch. Since 6 April 2025 the Digital Markets, Competition and Consumers Act 2024 has replaced the Consumer Protection from Unfair Trading Regulations 2008, and omitting material information from a listing is automatically an unfair commercial practice. The CMA enforces it directly, with fines up to 10% of worldwide turnover or £300,000 on an individual.
Material information still falls into the three areas the trade knows as Parts A, B and C: costs and tenure, physical characteristics, and the legal or environmental constraints that affect the specific property. The old National Trading Standards guidance that set out those areas was withdrawn from official channels when the DMCC Act took over, but the duty to disclose did not go with it. The legal source moved; the obligation stayed. Treating the withdrawn guidance as live law is a mistake to avoid.
The stakes sit where the buyers are. Rightmove held 89% of all time spent on UK property portals at the end of 2025, so a thin or inconsistent listing loses views and enquiries in the one place that matters, and a factual error is now visible to every buyer and carries direct CMA liability.
An AI assistant we build for your agency drafts particulars quickly from the property record and treats the material-information fields as required inputs. It flags a missing council tax band, an unstated tenure or an unrecorded lease length before the listing can go live, and a person approves every set of particulars before publication.
Particulars drafted from the property record, in the agency's tone
The assistant pulls the recorded facts of the property, room dimensions, tenure, council tax band and key features, and drafts clear, consistent particulars in the agency's house style. It writes strictly from the record and invents nothing; where a detail is missing it asks rather than fills the gap. Every draft goes to a member of staff for sign-off before it reaches the portal.
A two-bed flat in Bishopston, Bristol comes in for listing. The assistant drafts the description from the property record within minutes and returns it with two queries: the lease length is not recorded and the parking arrangement is blank. Staff supply both, read the copy through and publish.
Listings go live faster and to a consistent standard, even at the busiest times, and because the copy is drawn only from the record and approved by a person, it stays accurate and defensible.
The material-information checklist enforced before go-live
The assistant checks every listing against the Parts A, B and C fields the DMCC framework requires: tenure, council tax band, price and lease length; property type, construction, rooms, utilities and parking; and any legal or environmental constraint specific to the property. Missing fields are flagged and the listing is held back from publication until they are supplied or explicitly marked as being confirmed.
A period conversion is ready to list, but the assistant blocks go-live: the tenure shows leasehold with no lease length, and there is a known shared-access arrangement that is not mentioned. The branch chases the seller's solicitor, records both, and only then does the listing publish.
Material information is captured as a matter of routine rather than remembered case by case, which is exactly what the DMCC duty requires and what keeps the agency clear of a direct CMA problem.
Consistent, honest presentation across the portal
The assistant orders the photographs sensibly and keeps the description consistent with them and with the recorded facts. Image handling stays honest: tidying the light or a stray bin is fine, but nothing that hides a defect, fakes a view or misrepresents the property. The description has to match the actual home and its spec, and the assistant will not write a claim the record does not support.
For the Bishopston flat the assistant leads with the reception room and the kitchen, keeps the north-facing aspect accurately described, and declines to add a 'recently renovated' line that nothing in the record backs up. Staff approve the set as an honest, complete listing.
The listing performs where buyers actually spend their time, and it does so without drifting into the kind of claim that now carries direct liability. Presentation improves; the facts stay straight.
Listing questions answered from the record, at any hour
Buyers ask the same material questions before they enquire: tenure, council tax band, lease length, parking. On the website and over WhatsApp the assistant answers them straight from the property record, round the clock, introduces itself as an AI assistant, and says honestly when a detail is not yet confirmed. Anything beyond the recorded facts, a judgement or a negotiation, goes to a person with the conversation attached.
On a weekday evening a buyer asks whether the flat is leasehold and how the parking works. The assistant confirms the tenure and the allocated space from the record, notes that the exact lease length is being confirmed with the solicitor, and offers a viewing. The negotiator picks up a warm, informed enquiry the next morning.
Buyers get reliable answers to the questions that decide whether they enquire, outside office hours included, and the routine questions stop landing on negotiators who are busy with viewings.
The build rests on firm ground, and the duty behind it is now unambiguous.
- An assistant that drafts a full set of particulars from the property record and checks it against the material-information fields before anything publishes is buildable on the data you already hold. It captures the Parts A, B and C areas as required inputs, tenure, council tax band, lease length, physical characteristics and any known constraint, and holds a listing back when one is missing.GOV.UK
- The duty this serves is settled and enforced. Since 6 April 2025 the DMCC Act 2024 has made omitting material information automatically an unfair commercial practice, and the CMA enforces directly, with fines up to 10% of worldwide turnover or £300,000 on an individual. Complete, accurate particulars are compliance, not polish.GOV.UK
- It also pays where the buyers are. Rightmove held 89% of all time spent on UK property portals at the end of 2025, so a complete, well-ordered listing wins views and enquiries in the one place that decides them, while a thin one quietly underperforms.Rightmove plc
- Be sceptical of the time-saving headlines. The 'minutes saved per listing' figures attached to AI copy tools come from the companies that sell them, so treat them as direction and measure what matters to you: enquiries per listing and the share that go live complete first time. And remember that no efficiency gain counts for anything if a fact is wrong, because on this platform an error is visible to every buyer and now carries direct CMA liability.GOV.UK
Three rules keep AI-assisted particulars on the right side of the line.
- Honest particulars come first, whoever or whatever drafts them. Under the DMCC Act 2024 omitting material information is automatically an unfair commercial practice, and the ASA's rules are technology-neutral, so an AI description must match the actual property and its facts. Disclosing that AI was used does not cure a misleading listing.GOV.UK
- Image tools may tidy, not deceive. Removing clutter or lifting the light is acceptable; hiding a defect, faking a view or straightening a subsidence crack is a misleading action. Both the ASA and the CMA test the message the buyer receives, not the tool that produced it.ASA
- Do not treat the withdrawn NTSELAT Parts A, B and C guidance as the live legal source. Since 6 April 2025 the operative law is the DMCC Act 2024, enforced by the CMA. Keep templates mapped to that framework and re-check as CMA guidance develops.GOV.UK
Sales progression: every milestone chased, the stalling chain caught early
Agreeing a sale is not the same as completing one, and the gap between the two is where fees are lost. Industry trackers put the share of agreed UK sales that collapsed before completion in 2025 at around a quarter, and many of those failures now happen deep into the transaction, after weeks of time and money have gone in. There is no official government figure here, so this is an industry estimate rather than a national statistic, but the operational point is not in doubt.
The clock a vendor feels is long. It took 60 days on average simply to find a buyer in June 2026, and that is before the conveyancing, searches, mortgage offers, surveys and chain coordination that follow the sale being agreed. Progression is the long tail of the job, and it runs across parties who each keep their own timetable.
The scale underneath it is relentless. UK completions ran at roughly 100,000 a month into 2026, and every one is a stack of dependencies, memorandum of sale, searches, enquiries, mortgage offer, exchange and completion, with dates that have to line up across the whole chain.
An AI assistant, built to your agency's rules, is well suited to exactly this rule-bound work. It timestamps every milestone, chases the next party the moment something falls due, and surfaces a stalling chain before it becomes a collapse. It does not give legal advice: anything with legal effect stays with the solicitors and a person, and legal questions are routed to the conveyancer, not answered in a chat.
Every milestone tracked across the chain
The assistant keeps a structured record of each sale from memorandum of sale through searches, enquiries, mortgage offer and exchange to completion, and knows which milestone each party owes next. It timestamps every step and every chase, so the branch always sees where a transaction actually stands rather than relying on memory or a scribbled note.
A sale on Elm Road in Nottingham is four weeks in. The assistant shows searches applied for but not yet returned, the mortgage valuation booked, and three links below in the chain, one of which has not yet instructed a solicitor. The negotiator sees the real bottleneck at a glance.
The status of every sale is visible and current, so nothing drifts unnoticed and the branch spends its time on the transactions that need pushing, not on finding out where they are.
The next party chased, politely and on the record
When a milestone falls due, the assistant contacts the party responsible, buyer, seller or solicitor, with a factual, non-pressuring reminder, and logs the message against the sale. It knows whose turn it is and follows up on a schedule you set, so the chasing that normally eats negotiator hours happens consistently and is always documented.
Searches on the Elm Road sale pass the point where they were expected. The assistant sends the conveyancer a polite, logged chase for an update and lets the branch know it has done so. The negotiator does not have to remember to ring, and there is a clear record that the agency stayed on it.
Multi-party follow-up stops depending on who has time, and every chase is factual, timely and recorded. Sales keep moving without a negotiator's day disappearing into the phone.
A stalling chain caught early
The assistant watches the pattern of a transaction, milestones met against milestones slipping, and flags a sale that is going quiet before it dies. It surfaces the specific link that is holding things up and the deadline that is at risk, so the branch can intervene while there is still time to save the sale rather than after it has fallen through.
On a linked purchase two places down the chain, no progress is logged for a fortnight while everything above waits. The assistant raises it as an early warning with the exact link and the dates at risk, and the negotiator gets the whole chain talking again before anyone pulls out.
The failures that happen deep in the transaction, once time and money are already spent, are exactly the ones an early warning can prevent. Catching a stall early is the difference between a rescued sale and a collapsed one.
Factual status updates, with legal questions routed on
The assistant gives vendors and buyers a clear, factual update on where their sale stands whenever they ask, drawn from the tracked milestones. It introduces itself as an AI assistant, keeps every message non-pressuring, and hands anything with legal effect, or any question of conveyancing or legal advice, straight to the conveyancer or a named person, never answering it itself.
A vendor messages on a Sunday for an update. The assistant confirms searches are back, the buyer's mortgage offer is issued and exchange is the next step, then, asked whether a clause in the contract is normal, it declines to advise and offers to pass the question to the conveyancer first thing Monday.
Clients get prompt, accurate updates without waiting for office hours, while every legal or conveyancing question reaches the right qualified person. The agency stays responsive and stays within its lane.
This is rule-bound, deadline-driven work, which is where an assistant earns its place fastest.
- An assistant that keeps a structured record of every sale, timestamps each milestone from memorandum of sale to completion, chases the responsible party when a step falls due and surfaces a slipping date early sits well within what can be built today. UK completions ran at roughly 100,000 a month into 2026, each one a stack of dependencies whose dates must line up, and structured tracking that never forgets whose turn it is handles precisely that kind of work well.GOV.UK / HMRC
- The problem it addresses is real and expensive. Industry fall-through trackers put around a quarter of agreed UK sales collapsing before completion in 2025, and many failures now come deep in the transaction rather than early, once weeks of conveyancing have already gone in. A progression assistant is the operational insurance against a sale quietly dying.ReadySteadySell
- The timeline gives the work its weight. It took 60 days on average just to find a buyer in June 2026, and progression is the long tail that follows: solicitors, lenders, surveyors and every party in the chain, each on their own schedule. Automated status tracking and logged chasing across all of them is exactly the multi-party follow-up that otherwise consumes negotiator hours.Rightmove
- Stay honest about the headline number. There is no official government fall-through statistic; the roughly one-in-four figure comes from property-industry trackers and varies by region, higher in the South East and much lower in Scotland, so treat it as direction, not a fixed benchmark. Measure your own fall-through rate and average time from agreed to completed, before and after, and let those decide whether the assistant is earning its keep.ReadySteadySell
Three boundaries keep progression support useful and safe.
- Sales progression runs alongside conveyancing and legal process. The assistant can track milestones, chase parties and summarise status, but it must not give legal or conveyancing advice; anything with legal effect stays with the solicitors, and legal questions go to the conveyancer rather than the chatbot.
- A sale file is full of personal data: memorandums of sale, searches, identity documents and mortgage details. An AI service that reads them is a processor under the UK GDPR, so a written contract, data minimisation and a defined retention schedule are part of the build.legislation.gov.uk
- Chasing messages to buyers, sellers and solicitors must stay factual and free of pressure. Keep a human sign-off on anything that changes a party's legal position or commits the agency, and never let automation close or escalate a sale on its own.
Fee invoices and client money: the paperwork prepared, the ledger still yours
A residential sale keeps generating paper long after the keys change hands. The terms of business at instruction, the fee invoice with VAT shown separately, the completion statement, the supplier bills for photography, conveyancing referrals and portal listings. Much of that data gets typed by hand from one system into another, and that is precisely where the costly errors hide: a fee that disagrees with the signed terms, a supplier invoice that only surfaces at month end, a VAT figure keyed a digit out.
The legal frame here works in your favour, but it also draws a firm line. Where your agency holds client money in England, you have had to belong to an approved client money protection scheme since 1 April 2019, display the certificate, and face fines of up to £30,000 for trading without membership. Reconciling money in, agreed fees and payments out is regulated, rule-bound document flow, and the scheme assumes one thing above all: that the agent, not a piece of software, controls the ledger.
Commission is only ever invoiced on the terms the seller agreed up front. Under the Estate Agents Act 1979, you must give the seller clear written terms, including the fee and any tie-in, before they are committed, and disclose any personal interest. The fee, the trigger event and the sole-agency period all have to be captured at instruction and honoured at completion, not reconstructed from memory afterwards.
This is the gap an AI assistant built for your agency works inside. It drafts the terms of business and the completion invoice from the figures you agreed, reads incoming supplier invoices and checks them against the file, and prepares each entry for your accounting software. What it does not do is touch the client account or transfer your duties: nothing reaches the books, and no money moves, until a person in your office has approved it.
Terms of business and the completion invoice drafted from the agreed figures
At instruction the assistant drafts the written terms of business from what you agreed with the seller: the fee, the basis it is charged on, the trigger event and any sole-agency period. At completion it produces the fee invoice from those same recorded figures, with VAT shown separately for the seller's accountant. It writes strictly from the file and invents no number. A person in your office checks and sends every document, and the assistant flags, rather than assumes, that the personal-interest disclosure was made before the seller committed.
A vendor in Harrogate agrees a fee at instruction. When the sale completes, the assistant assembles the invoice from the signed terms, shows the VAT as a separate line and notes that the figure matches the agreed basis to the penny. The office manager reads it through in a minute and sends it, instead of rebuilding the fee from an email thread.
The invoice always matches the terms the seller actually agreed, so fee disputes at completion get rarer. The document is ready in minutes rather than assembled by hand, and because a person approves every one, it stays defensible.
Incoming supplier invoices read, checked and prepared for the books
The assistant picks up supplier invoices from the accounts inbox and extracts what matters: invoice number, date, line items, VAT, payment terms. It matches each one against the order or the listing it belongs to, checks the arithmetic, and prepares the entry for your accounting software. Anything that does not reconcile is parked in a query folder with a plain note on what is wrong. Nothing is posted until your bookkeeper approves it.
A portal invoice comes in billing for eight listings in the month. The assistant matches it against the properties that actually went live, finds seven, parks the invoice and drafts a query to the supplier quoting the reference. The bookkeeper resolves it the day it arrives, not three weeks later at the close.
Routine invoices land in the books already checked, and discrepancies surface on arrival rather than at month end. Because the data flows digitally instead of being typed twice, the process fits what Making Tax Digital already expects of your records.
Client money reconciliation prepared, never posted
Where you hold deposits or other client money, the assistant helps organise the repetitive part: it reads statements and receipts, matches money in against the file it belongs to, and lays the agreed fees and payments out in a prepared reconciliation for review. It changes nothing in the client account and moves no money. The extraction and matching save the hours; the agent keeps sole control of the ledger the scheme holds them responsible for.
At month end the assistant presents a prepared reconciliation for a Harrogate branch: deposits received, matched to files, with two receipts flagged as unallocated. The person responsible for the client account reviews it, allocates the two, and signs it off, working from a checked draft rather than a blank spreadsheet.
The reconciliation is prepared in a fraction of the time, and the exceptions are already isolated. The regulated control stays exactly where the client money protection scheme requires it, with the agent, not the automation.
Treat the 2029 e-invoicing mandate as a runway, not a deadline
Mandatory e-invoicing for all VAT invoices is announced from 2029, covering business to business and business to government, with the technical roadmap due at Budget 2026. There is nothing to buy yet. What the assistant usefully does now is make sure your invoice data already flows as structured digital data: it checks outgoing fee invoices for complete details before they are sent, and flags business customers whose records are missing information a structured invoice will need.
You ask the assistant to review the last twelve months of fee invoices. It returns a short working list: which are to business sellers and landlords who will sit inside the mandate, and where a customer record is missing a detail. A vague future obligation becomes a tidy-up you do at your own pace.
The 2029 mandate then arrives as a formality rather than a project. And the same structured data pays for itself immediately, because it is exactly what Making Tax Digital rewards today.
Fee invoicing and client-money paperwork sit on solid, buildable ground today, provided the responsibility lines are respected.
- The regulated part is document flow, and document flow is exactly what an assistant handles well. Where your agency holds client money in England, you have had to belong to an approved client money protection scheme since 1 April 2019, display the certificate, and face fines of up to £30,000 for trading without membership, so reconciling money in, agreed fees and payments out is rule-bound work. Extraction plus human sign-off saves the hours without touching the compliance perimeter or the ledger the agent must control, and it is a practical build today.GOV.UK
- Fee documents can be drafted from the agreed figures today. Under the Estate Agents Act 1979 an agent must give the seller clear written terms, including the fee and any tie-in, before the seller is committed, and disclose any personal interest. An assistant can draft those terms and the completion invoice from the recorded basis, so the fee, the trigger event and the sole-agency period are captured once and honoured at completion rather than reconstructed by hand.legislation.gov.uk
- The tax rules already reward this. Making Tax Digital for VAT requires digital records kept in compatible software, and mandatory e-invoicing for all VAT invoices has been announced for 2029. Extracting fee, invoice and expense data into the accounting system is the compliant path now, and structured invoice data means the 2029 mandate arrives as a formality rather than a scramble.GOV.UK
- Keep a level head on the numbers. The time-saved percentages quoted for document automation tend to come from other markets and from the firms selling the tooling, so read them as a direction of travel, not a promise. Judge the build on your own ledger: minutes per invoice, discrepancies caught on the day they arrive, and how many fee queries your accountant raises before and after.
Three responsibility lines belong in the build from day one.
- AI-extracted invoice and fee data still needs human sign-off before it touches client accounts. Client money protection assumes the agent controls the ledger, and automation does not transfer that responsibility, whoever or whatever did the typing. The working rule is simple: the assistant reads, matches and prepares; posting and any movement of money happen only after a person approves.GOV.UK
- The Estate Agents Act 1979 requires the fee terms and any personal interest to be disclosed in writing before the seller is committed. An assistant can draft the terms of business, but a person must ensure the disclosure actually happened before instruction, so the assistant flags that step rather than presuming it was done.legislation.gov.uk
- Do not buy 2029 e-invoicing compliance yet. The technical standards arrive with the Budget 2026 roadmap, and anyone selling certainty before that is selling ahead of the facts. What pays off now is structured digital records, which Making Tax Digital already rewards, so prepare calmly rather than under manufactured deadline pressure.GOV.UK
Anti-money-laundering checks: the evidence organised, the risk decision still a person's
Before an estate agency does any estate-agency work at all, it has to be registered with HMRC for anti-money-laundering supervision. Trading without that registration is a criminal offence, not a paperwork slip, and HMRC can publish the details of the businesses that fall short. Registration, customer due diligence and record-keeping are structured, checklist-driven duties, which is precisely why they can be organised well, provided a person stays responsible for the judgements.
The stakes stopped being abstract this year. Estate agencies were the worst-affected sector in HMRC's latest round of penalties: 170 penalties worth £835,842 in the six months to 30 September 2025, out of 369 across all supervised sectors, and 332 of those 369 were for trading without being registered. Compliance here is licence to trade, not back-office housekeeping, and the fines land hardest on process failures rather than on deliberate wrongdoing.
Alongside AML sits a second licence-to-trade condition. Every residential estate agent must belong to an approved redress scheme, either The Property Ombudsman or the Property Redress Scheme, mandatory since 1 October 2008 under the Estate Agents Act 1979 as amended by the Consumers, Estate Agents and Redress Act 2007. Both AML and redress generate exactly the kind of certificate, renewal date and evidence trail that automation keeps track of well.
So the design here is deliberate. Missing registration, weak due diligence and gaps in source-of-funds evidence are the failures the penalties punish, and a systematic, prompted onboarding flow prevents exactly those. An AI back office collects and organises the ID, the source-of-funds evidence and the beneficial-ownership picture, flags what is missing, and tracks what renews when. What it never does is own the risk assessment or file a suspicious activity report: those stay with your nominated officer.
A prompted due-diligence flow that will not let a gap through
The assistant runs the customer due diligence as a structured checklist tied to each transaction: identity evidence for buyer and seller, the beneficial owners behind any company or trust, and the standard checks your policy requires before a sale can progress. It marks each item complete, outstanding or queried, and it will not let a file read as ready while a required element is missing. The nominated officer makes every risk decision; the assistant makes sure the officer is deciding on a complete picture.
A buyer for a property in Nottingham comes through a company. The assistant confirms the director's ID is verified but flags that the beneficial-ownership evidence behind the company is still outstanding, and holds the file short of ready until it is supplied. Nobody has to remember the corporate structure check; the flow refuses to skip it.
The gaps that draw HMRC penalties, missing checks and half-finished onboarding, are caught before a sale progresses rather than at an inspection. The officer's attention goes to the judgement calls, not to chasing which box is unticked.
Source-of-funds evidence gathered, organised and chased
The assistant requests, receives and organises source-of-funds evidence in one place per buyer: bank statements, evidence of a deposit gift and its origin, sale proceeds, and any required explanation. It reads what comes in, notes what is still missing against your policy, and sends a polite follow-up for the gaps. It presents the assembled evidence to the officer for the risk decision; it never judges whether the funds are satisfactory itself.
A buyer's deposit includes a family gift. The assistant collects the bank statement showing the transfer, notes that the letter confirming the source of the gift has not yet arrived, and drafts the request for it. The officer reviews the complete bundle in one view and makes the call, instead of assembling it from three email threads.
Source-of-funds evidence stops living in scattered inboxes and becomes a consistent, complete file per buyer. The chasing that usually slips gets done reliably, and the officer decides on organised evidence rather than fragments.
Registration and redress renewals tracked before they lapse
The assistant keeps the register of your licence-to-trade obligations: HMRC AML supervision, redress-scheme membership, client money protection, with what is registered, when each renews and what has to be produced on request. It prompts ahead of every renewal and keeps the certificates and evidence to hand, so a lapse never happens by oversight and a request for proof is answered in seconds.
Six weeks before the redress-scheme membership is due, the assistant flags it with the certificate attached and the renewal steps listed. The office renews on time and files the new certificate, instead of discovering the lapse when a complaint or an audit asks for it.
The conditions that let you trade at all are tracked systematically, not held in one person's memory. Renewals happen on time, and the evidence trail is ready whenever HMRC, the redress scheme or a client asks for it.
Sanctions and PEP screening output routed for a human to clear
Where your policy requires sanctions and politically-exposed-person screening, the assistant runs it against the consolidated list at the right point in onboarding and presents the result. It never clears a match itself. A possible hit is flagged, held and routed to a person to review against the evidence, because both a false match wrongly acted on and a real match missed carry liability. The screening is a prompt for human judgement, not a verdict.
Screening a buyer returns a possible name match on the sanctions list. The assistant holds the file, flags the match with the detail behind it and routes it to a member of the team to clear or escalate. The sale is not marked as progressing until a person has resolved it.
Screening happens consistently and at the right moment, and every match gets a controlled, documented human decision. Neither a false positive nor a missed name slides through on an automated say-so.
Anti-money-laundering work is one of the strongest cases for a prompted assistant, because the failures that get punished are process failures.
- Registration, due diligence and record-keeping are structured, checklist-driven duties. It is a criminal offence to trade as an estate agency business without HMRC anti-money-laundering supervision, and the registration has to be in place before any estate-agency activity begins. Organising those duties, and keeping the human responsible for the risk decisions, is buildable now: the assistant assembles the evidence, the officer owns the call.GOV.UK
- The penalties show exactly where an assistant earns its place. Estate agencies took 170 of HMRC's latest AML penalties, worth £835,842 in the six months to 30 September 2025, and 332 of the 369 penalties across all sectors were for unregistered trading. The fines land on missing registration, weak customer due diligence and gaps in source-of-funds evidence, which a systematic, prompted onboarding flow prevents. That flow is deliverable today.GOV.UK / HMRC
- A second licence-to-trade obligation fits the same pattern. Every residential estate agent must also belong to an approved redress scheme, mandatory since 1 October 2008 under the Estate Agents Act 1979 as amended by the Consumers, Estate Agents and Redress Act 2007. Redress and AML both produce certificates, renewal dates and evidence trails, and tracking what is registered, when it renews and what must be produced on request is squarely what automation does well.legislation.gov.uk
- Stay sceptical of the sales pitch, though. The compliance-cost-reduction percentages quoted around screening and onboarding tools come from the companies selling them, in other markets, so treat them as a direction rather than a guarantee. Measure the build on your own audit readiness: checks completed before a sale progresses, source-of-funds bundles complete on file, and how few gaps an inspection would find.
The responsibility for the risk never moves, and the data is sensitive, so plan these in from the outset.
- Customer due diligence is a legal duty on the agent. An assistant can collect and organise ID, source-of-funds and beneficial-ownership evidence and flag what is missing, but the risk assessment and any suspicious activity report stay with the nominated officer, never with the automation. The assistant makes the picture complete; the officer decides on it.GOV.UK
- Sanctions and PEP screening became mandatory for all agents in 2025, and the consolidated list is updated regularly. Automated screening output has to be checked by a person, because both a false match acted on and a real match missed carry liability. The assistant runs the screen and holds any hit; a human clears or escalates it.
- ID documents, bank statements and source-of-funds evidence are high-sensitivity personal data. AI services that read them are processors under the UK GDPR, which means written contracts, data minimisation and deletion schedules that are actually followed, and it means the documents are never used to train external models without a lawful basis.legislation.gov.uk
Your pipeline in plain English: time to sell, stale instructions and where deals die in the chain
The numbers you need to run an agency already exist, and most of the market benchmark sits at national-statistics grade and free. HMRC publishes official monthly transaction counts, roughly 100,000 residential completions a month into 2026, and HM Land Registry publishes the official UK House Price Index. The gap in a typical agency is not data. It is that nobody translates any of it, monthly, into a per-branch answer to the only questions that matter.
Rightmove's monthly baselines give you the external line to measure against. In June 2026 the average time to find a buyer was 60 days, more than a third of new listings were not going on to sell, and homes that did not need a price reduction found a buyer in around 36 days against 127 for those that did. A sentence like "our average time to sell is 74 days against a 60-day market, and most of our over-60-day stock was priced above appraisal" changes behaviour, and producing it every month, per branch and per negotiator, is a reporting agent's job.
Sales progression deserves its own line in the same report. Industry trackers put the share of agreed UK sales that collapsed before completion at around a quarter in 2025, roughly 24% to 25%, with mortgage, survey and chain problems the main causes. That figure is an industry estimate rather than an official statistic, and it varies by region, but the operational point is solid: agreeing a sale is not completing one, and the report has to show where deals are dying in the chain.
In most agencies the owner is the reporting department, also valuing, negotiating and progressing sales, so the job gets done last or not at all. An AI reporting agent reads the live pipeline, lays the market line alongside, and answers plain-language questions in a short, honest summary with the underlying figures named. Not another dashboard nobody opens, but a conversation with your own numbers that fits inside a full day.
Stale-instruction watchlist, flagged weekly
The assistant reads the live instruction list and checks it against the thresholds you set, typically 30, 60 and 90 days on the market. Once a week it flags every instruction past the line, notes when the enquiries and viewings stopped, and marks which have never had an offer. Nothing is maintained by hand; the list is built from the pipeline data you already hold.
Monday morning, a short summary for the Norwich and Ipswich branches: seven instructions over 60 days without an offer, the oldest a flat at 96 days with no viewing booked in a fortnight. Three sit above their appraisal figure. The vendor conversations happen that week, not at a quarterly stock review.
Slow instructions stop hiding in a long list, so price and marketing decisions happen weeks earlier. Chasing stale stock becomes a weekly routine instead of an occasional clear-out when someone finds the time.
Time to sell against the market baseline, per branch and per negotiator
The assistant computes your own average time to sell from the pipeline and lays it against Rightmove's published monthly baseline, broken down by branch and by negotiator. It presents the comparison as decision support with the source and the month stated, never as a verdict, and it names the figures behind every line so they can be checked.
The monthly summary shows the Ipswich branch selling in 74 days against a 60-day market baseline, while Norwich runs at 58, and it notes that most of Ipswich's slow stock was listed above the appraisal figure. The owner takes that into the branch meeting with the numbers agreed rather than disputed.
You see, in plain figures, where you are ahead of the market and where you are behind it, and why. The conversation with each branch happens over the same numbers instead of competing impressions.
Fall-through analysis: where deals die in the chain
The assistant tracks agreed sales from acceptance to completion and records where the ones that collapse fall over: mortgage, survey, a break lower in the chain, a buyer withdrawing. It reports your fall-through rate against the industry-tracker estimate, clearly labelled as an estimate, and highlights the stages where your deals most often fail so progression effort goes where it saves the most fees.
The quarterly view shows a fall-through rate broadly in line with the industry estimate, with chain breaks below the agreed sale the most common cause. The owner reprioritises chain-checking earlier in progression on affected deals, targeting the exact point the report identified.
Sales progression is steered by where fees are actually being lost, not by instinct. Because completing a sale is where the commission is earned, catching a common failure point early pays back directly.
Plain-English questions to your own pipeline
Between the fixed summaries the assistant stays available. You ask the question as it occurs to you: which instructions are stale, how does this month compare with the market, where are deals stalling? It reads the answer from the live pipeline, read-only and changing nothing, and names the figures and the month it relied on so you can verify them.
Evening, from the phone: how many sales did we agree this month, and how does our time to sell compare with the latest market figure? The answer arrives in under a minute, in three sentences, with the values and the source month named, instead of waiting for a report nobody has had time to build.
Decisions are made on current numbers rather than last quarter's picture. The reporting work stops being a dreaded chore and becomes a short routine conversation.
Owner-level pipeline reporting rests on data that already exists at national grade, so the AI layer on top is a realistic first build, not a research project.
- The benchmark data is public and free. HMRC publishes official monthly UK transaction counts, roughly 100,000 residential completions a month into 2026, and HM Land Registry publishes the official UK House Price Index. The bottleneck in a typical agency is nobody translating that, monthly, into per-branch answers: is our time to sell above or below the market, which instructions are stalling, where are we losing deals in the chain. That translation into a weekly decision list is exactly what the reporting agent provides, and it is ready to build on your systems.GOV.UK / HMRC
- There is a solid external line to lay your own numbers against every month. Rightmove's June 2026 baselines put the time to find a buyer at 60 days, more than a third of new listings not selling, and a 36 versus 127 day gap between correctly and over-priced homes. An owner can set their own listings, offers and fall-throughs beside those baselines automatically, per branch and per negotiator, and that comparison is a straightforward build on data you already hold.Rightmove
- The first step suits the shape of the trade. The UK estate agency business is thousands of small and mid-sized firms, and in most of them the person who should be reading the numbers is also valuing, negotiating and progressing sales. Plain-language questions answered from live pipeline data, which instructions are stale, where deals die in the chain, how this month compares with the market, are the realistic first AI step in the office, not another dashboard that goes unopened.GOV.UK / HMRC
- Hold the vendor claims lightly. The productivity gains quoted by firms selling reporting tools were earned in other markets, on their own products, so they are a compass, not a promise. The only benchmark worth trusting is your own branch: time to sell, stale-instruction count and fall-through rate, each measured before and after the reporting layer goes live.
Two data-law points and one honesty point govern how the reporting output should be built and read.
- Reports that slice by client or negotiator, who instructed, whose listings stall, who converts, are personal data processing under the UK GDPR. Purpose limitation applies, and any external analytics or AI tool acts as a processor, which needs a proper processing agreement. Pure pipeline metrics with no personal data sit outside that, but the line between the two should be drawn deliberately, not discovered later.legislation.gov.uk
- Market indices measure different things, and that matters when you compare them. HMRC counts completed transactions, HM Land Registry tracks completed sale prices, and Rightmove measures asking prices and time to find a buyer, so the numbers differ legitimately. Present the reporting output as decision support with the source and the month stated, never as an automatic verdict on a property or a negotiator.GOV.UK / HMRC
- Fall-through and time-to-sell benchmarks from industry trackers are estimates, not official statistics, and they vary by region. Label them as such in any report, so a soft market is not mistaken for a branch that is failing, and so a decision rests on what the figure can actually support.GOV.UK / HMRC
See what this looks like in your agency
Describe your agency in one sentence and Claude will show you live where AI would make the biggest difference, using your own numbers.
The pages that follow this one
Sources
- Rightmove plc - Full Year Results 2025 (RNS, 27 February 2026)
- GOV.UK / HMRC - Monthly property transactions commentary
- Rightmove - House Price Index (June 2026)
- ASA - Disclosure of AI in advertising: striking the balance between creativity and responsibility
- legislation.gov.uk - PECR 2003, regulation 22 (use of electronic mail for direct marketing)
- legislation.gov.uk - UK GDPR (Regulation (EU) 2016/679 as retained in UK law)
- ReadySteadySell - House sale fall-through rates in the UK: 2025 transparency report (industry data)
- legislation.gov.uk - Estate Agents Act 1979
- FCA - Financial promotions and adverts
- GOV.UK - UK House Price Index
- GOV.UK - Unfair commercial practices (CMA207)
- GOV.UK - Protecting clients' money if you're a property agent
- GOV.UK - Promoting electronic invoicing: consultation response
- GOV.UK - Money laundering supervision for estate agency businesses (registration guide)
- GOV.UK / HMRC - HMRC cracks down on unlawful estate agents (published penalties)
- legislation.gov.uk - Estate Agents Act 1979, redress schemes
Bring your own process to a 30-minute call.
You learn whether it is worth automating in your systems and what it costs, from the published price list. You talk to a Claude Certified Architect from our team, not a salesperson.
Prefer to call? +44 7588 772736 · Not ready to talk? See it on your business first.
- 1Time
- 2Details
- 3Booked