Key takeaways
- An AI employee completes a defined task end to end, such as qualifying a caller and booking a valuation, rather than just taking a message.
- The work is real and unevenly timed: Propertymark sales branches registered 69 new buyers each in August 2026 while staff were out on viewings[1].
- Answering questions and arranging viewings count as estate agency activity, so your AML, redress and consumer law duties cover what the AI says[2].
- Keep pricing advice, offers, AML checks, tenant selection, notices and complaints with people, and make the handoff to a person easy at any point.
What an AI employee is, and is not
The phrase covers software that does a piece of agency work from start to finish within rules you set. A caller asks about a house; the AI employee answers from the approved property details, asks the buyer's position, takes a viewing request, writes the notes into the CRM and tells the right negotiator. A basic answering service or receptionist bot, by contrast, mostly takes a name and number and leaves the work for later.
It is not a negotiator, valuer or property manager. It has no judgement about value, no feel for a seller's nerves and no authority to commit the agency. The useful question is not whether AI can replace a member of staff but which complete tasks, with a clear start, end and handoff, it can take off their desks.
The workload these tools are aimed at
In August 2026 a typical Propertymark sales branch registered 69 new prospective buyers, had 51 properties available, placed about 10.8 homes on the market and agreed 9.5 sales[1]. With an average of 1.7 viewings per available property[1], negotiators spend much of the day away from their desks, which is exactly when new buyers and sellers ring.
Staffing is stretched in the same way. Estate agents typically work 35 to 40 hours a week and may cover weekends on a rota[3], and about 22% of employees in Great Britain's real estate agencies worked part time in 2024[4]. Calls at lunchtime, early evening and on Sunday mornings land on whoever is free, or on nobody.
Tasks that can be handed over end to end
| Task | What the AI employee completes | When it hands back to a person |
|---|---|---|
| Property enquiry call | Answers from the approved property sheet and records the buyer's position | Questions not on the sheet, or anything about price flexibility |
| Viewing request | Takes the property, preferred times and buyer details as a request | When a negotiator needs to confirm the slot with the seller or tenant |
| Valuation enquiry | Takes the address, reason for moving and timescale, and books the appointment | Seller wants advice on price before the visit |
| Rental enquiry | Gives the advertised rent and asks every applicant the same questions | Applicant asks about a decision, an offer or a special arrangement |
| Repair report | Logs the fault and access details and flags urgent issues | Anything urgent goes straight to the out-of-hours contact |
| Call routing | Routes by branch, department or property with a warm transfer | Whenever the caller asks for a person |
| Call record | Recording, transcript, summary and a contact record for the CRM | Staff review summaries and correct errors |
Tasks that should stay with people
- Pricing advice, valuations and marketing strategy, which depend on judgement and evidence the caller cannot see.
- Negotiating offers or passing on information about other buyers' offers.
- Customer due diligence and anti-money laundering checks, including judging whether a source of funds makes sense.
- Choosing tenants or deciding whether an application is accepted.
- Serving notices or sending anything that could change the legal position of a tenancy or sale.
- Complaints, threats of legal action and conversations with clients who say they are vulnerable or distressed.
These limits are not just caution. They are where the legal duties and the commercial value of an agency sit. A tool that answers a hundred enquiries well but mishandles one offer can undo the benefit of all the others.
It works under your registrations, not its own
HMRC counts arranging viewings and answering questions from potential sellers, buyers, landlords or tenants as estate agency activity for money laundering registration[2], and trading as an estate agency business without that registration is a criminal offence[2]. When an AI employee does that work on your behalf, it is your business doing it. What it says is covered by your redress scheme membership, your consumer protection duties and your data protection obligations, exactly as if a member of staff had said it.
That has practical consequences. The approved facts it uses must be accurate and complete, the questions it asks must be the same for everyone, the recordings and notes it creates are records you control, and a person must be accountable for how it is configured.
Being open with callers
Most people are happy to get a fast answer from software, but they want a person behind it. In the government's public attitudes survey, 91% of UK adults said they were concerned about important decisions being made by computers without human input[5]. Tell callers early that they are speaking to an automated assistant, explain what happens next and offer a transfer to a person whenever they ask.
Script: 'Hello, you are through to the lettings team. I am the agency's automated assistant. I can answer questions about our properties, take viewing requests and log repairs, and I can put you through to a member of the team at any point. How can I help?'
Where Ava fits
Ava is an AI Employee designed for this work. The capabilities that are live today are answering every call, answering property questions from your approved property sheet, taking viewing requests, booking valuations into your calendar, logging tenant and landlord calls including repairs with urgent flags, routing to the right negotiator with a warm transfer, and a clean CRM handoff with recording, transcript and summary. Staff can listen live, whisper guidance or take over a call. Instant viewing booking and property feed sync are planned, not live, so viewing requests still go to a negotiator to confirm.
What it means for your team
Across UK businesses using AI to improve operations, 63% reported no change in headcount, 6% a decrease and 1% an increase[6]. In agencies the more common effect is that negotiators stop losing the first half hour of each day to voicemails and callbacks, and weekend cover becomes less painful. Tell staff plainly what the tool will and will not do, involve the people who answer the phones in writing its scripts, and make them the owners of its quality.
It also changes what good looks like for the people who used to answer every call. Their job shifts towards checking summaries, confirming viewing slots, calling back the buyers and sellers the AI has qualified, and spotting patterns in what callers ask. Those patterns are valuable: if a third of calls about a listing ask about parking, the listing is missing something. Treat the call records as a weekly source of insight about your market and your marketing, not just as admin.
Setting the limits: a five-step setup
Define the scope in writing
List the tasks it will complete, the tasks it must hand back and the phrases that trigger an immediate transfer, such as complaint, solicitor or offer.
Build the approved facts
Property sheets with tenure, council tax band, energy rating and viewing arrangements, plus office hours, fees pages and service descriptions.
Set the handoff rules
Who receives which calls by branch and department, who covers evenings and weekends, and what happens when nobody answers a transfer.
Test with real scenarios
Have staff ring in as awkward buyers, anxious tenants and pushy callers before go-live, and fix what goes wrong.
Review every week
Read a sample of transcripts, check CRM entries against the calls and track missed handoffs. Adjust the scripts, not just the settings.
Worked example: a Saturday at a two-branch agency
Example: a fictional agency opens both branches until one o'clock on Saturdays with two negotiators out on viewings. Between nine and five, 46 calls come in. The AI employee answers all of them: 19 property questions answered from the property sheets, 11 viewing requests taken, 4 valuations booked into the valuer's diary, 6 calls transferred to the negotiator on duty, 3 repair reports logged with one flagged urgent to the out-of-hours contractor line, and 3 wrong numbers.
Example, continued: on Monday morning the branch manager reads the summaries, confirms the viewing slots with sellers and calls back the two buyers who asked about price. Nobody listens to a voicemail, and the four valuation appointments are already in the diary with the seller's reason for moving noted.
Questions to ask before choosing one
- Which tasks are live today, and which are on a roadmap?
- Can we restrict answers to our approved facts and see exactly what was said on every call?
- How does a caller reach a person, and what happens if nobody picks up?
- Where are recordings and transcripts stored, for how long, and who can delete them?
- How do notes reach our CRM, and what happens when the CRM is unavailable?
For a side-by-side view of how this differs from a basic answering service, see AI receptionist vs AI employee.
Frequently asked questions
What is an AI employee in estate agency?
Software that completes defined agency tasks end to end within your rules, such as answering a property enquiry, qualifying the caller, taking a viewing request or booking a valuation, and recording it all in the CRM.
Can an AI employee negotiate offers?
It should not. Offers, pricing advice and anything about other buyers' positions need a negotiator's judgement and accountability. The AI should hand those conversations to a person.
Do we need to tell callers they are speaking to AI?
It is good practice to say so at the start and to offer a person at any time. Callers are more comfortable when they know what they are dealing with and how to reach the team.
Does an AI employee change our compliance obligations?
No new registration is needed, but its work counts as your agency's work. Your AML, redress, consumer protection and data protection duties apply to what it says and records.
References
Propertymark (2026). Housing Insight Report, August 2026 (opens in a new tab) ↑
HM Revenue & Customs (GOV.UK) (2021). Money laundering supervision for estate agency businesses (opens in a new tab) ↑
National Careers Service (GOV.UK) (2026). Estate agent job profile (opens in a new tab) ↑
Office for National Statistics (Nomis) (2024). Business Register and Employment Survey: open access (SIC 68310, 68320), 2024 (opens in a new tab) ↑
Department for Science, Innovation and Technology (GOV.UK) (2024). Public attitudes to data and AI: Tracker survey (Wave 4) report (opens in a new tab) ↑
Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑