Key takeaways
- Lettings disputes resolved by The Property Ombudsman rose 65% to 3,119 in 2025, and poor or delayed communication tops the list of issues[1].
- AI is most useful today for drafting, summarising and sorting: applicant replies, inspection write-ups, document chasing and repair logging.
- Keep notices, deposit decisions, tenant selection and anything that could count as legal communication with a person; even a WhatsApp message can be a notice to quit[2].
- A short daily review of the queue is what makes AI admin safe, so build that habit before you switch anything on.
Where lettings time actually goes
A lettings team's day is mostly messages. Propertymark letting branches averaged 121 new applicant registrations in August 2026, with around ten applicants for every available property, while each branch agreed 9.92 new tenancies[3]. Most of those applicants will never rent through you, yet each one expects a reply, a viewing slot or a clear no. Add landlord updates, repair reports, document chasing and renewal paperwork, and the inbox fills faster than people can empty it.
Much of this work also runs to legal deadlines. A deposit must be protected within 30 days of receipt[4], and existing tenants must receive a copy of the electrical safety report within 28 days of the inspection[5]. The work is not intellectually hard; it is relentless and easy to drop when the office is busy. That combination, high volume and predictable shape, is exactly where current AI tools help.
Communication is the complaint
The Property Ombudsman resolved 3,119 lettings disputes in 2025, up 65% on the year, compared with 2,243 sales disputes[1]. Poor, unclear or delayed communication was listed first among the top issues in lettings cases, and the average lettings award was £497[1]. Many complaints start not with a bad decision but with silence: a tenant who reported a leak and heard nothing, or a landlord who learned about a void from the rent statement.
Source: The Property Ombudsman, 2025[1]
Show the data
| Item | Value |
|---|---|
| Lettings | 3,119 disputes |
| Sales | 2,243 disputes |
That is the strongest argument for AI in lettings admin. It does not replace judgement, but it can make sure every message is acknowledged, logged and routed, and that nobody waits days for a first reply. The ONS found that businesses using machine learning for data processing most often reported impacts on administrative or clerical roles, at 41%[6], which matches what agencies describe: the tools take the typing and sorting, not the decisions.
Tasks that work well today
| Task | What the tool does | What a person does | Risk if unchecked |
|---|---|---|---|
| Applicant enquiries | Replies with the advertised details and asks the standard qualifying questions | Chooses who to invite to view | Low, if replies use approved facts only |
| Viewing confirmations | Sends confirmations, directions and reminders, and logs cancellations | Handles clashes and special access | Low |
| Inspection write-ups | Turns voice notes and photos into a structured report draft | Checks findings and actions before sending | Medium: wrong condition notes reach landlords |
| Document chasing | Reminds applicants and landlords about missing items on a schedule | Decides when to escalate | Low |
| Repair logging | Captures the fault, photos and access details, and suggests urgency | Confirms urgency and instructs the contractor | Medium: urgency misjudged |
| Landlord updates | Drafts a weekly summary from the file | Reviews and adds advice | Medium: tone and accuracy |
| Compliance reminders | Flags certificates and deadlines due in the next 60 days | Books the work and records it | Low, if dates are right in the CRM |
The common thread is that the tool drafts, sorts or reminds, and a person approves anything that commits the agency or the landlord. Where a task is purely informational, such as confirming a viewing time, it can run with spot checks. Where it carries advice or a judgement, it needs a human read every time.
Tasks to keep with people
Some lettings communications have legal effect, and the line is not always obvious. Government guidance on the Renters' Rights Act notes that a notice to quit can include texts or WhatsApp messages, so agents' messaging needs care[2]. It also says remarketing includes posting an online advert or telling someone the property is or may be available to let[2]. An automated reply that cheerfully tells an applicant a home will be free soon could, in the wrong circumstances, cause real trouble.
- Any notice, including possession notices, rent increase notices and anything that could be read as ending a tenancy.
- Decisions on deposit deductions and the wording of proposals to tenants.
- Choosing between applicants, or explaining to an applicant why they were not chosen.
- Responses to complaints, threats of legal action or anything mentioning the redress scheme.
- Advice to landlords on rent levels, possession grounds or compliance choices.
- Messages to tenants who say they are vulnerable, unwell or in financial difficulty.
A daily review routine
First thing: the overnight queue
Read every message, call summary and repair report that arrived out of hours. Confirm urgency flags and reassign anything the tool routed wrongly.
Mid-morning: drafts awaiting approval
Approve or edit landlord updates, inspection reports and any reply the tool marked as needing a person.
After lunch: spot checks
Open five automated replies sent the day before and check they used approved facts and the right tone.
End of day: deadlines
Review the compliance and document reminders due in the next week and make sure each has an owner.
Weekly: error log
Note every mistake the tool made and adjust the instructions or the approved facts so it does not happen again.
If nobody has time to review the queue, the tool is not saving time; it is moving risk. Start with fewer tasks and add more only when the review routine is holding.
Writing rules for AI drafts
- Use only facts from the CRM or the approved property sheet; never guess availability, rent or move-in dates.
- Give the advertised rent only, take no offers and ask every applicant the same questions in the same order.
- Never say a repair will be done by a date unless a contractor has confirmed it.
- Avoid legal conclusions such as 'the landlord is entitled to' or 'you must leave by'.
- Sign off with a named team, a phone number and office hours, so people know a person is behind the message.
These rules belong in the tool's instructions and in your staff guidance, so that a human editing a draft follows the same standard. Our page on an AI policy for staff shows how to write them in plain language.
Worked example: one lettings manager's week
Example: a fictional lettings manager looks after 140 managed homes. Before using AI tools she spent about nine hours a week answering applicant questions, four hours writing up inspections and three hours chasing documents and certificates. After a month with drafting, logging and reminders switched on, she records her time again.
Example, continued: applicant replies now take about three hours, mostly reviewing and handling exceptions; inspection write-ups take two, because she edits a draft instead of typing from scratch; chasing takes one. She also spends a new hour a day on the review routine. The net saving is around five hours a week, which she uses for landlord calls and rent reviews. Just as important, no message in her queue waits more than a working day for a first reply.
Questions to ask a lettings software supplier
Many property management systems now include AI features, and standalone tools promise to plug into them. Before you switch anything on, get clear answers to a short list of practical questions. Vague replies are a warning sign.
- Which facts does the tool use when it replies, and can we restrict it to the advert and our approved property sheet?
- Can we mark message types, such as notices or complaints, that must always go to a person?
- Where is tenant and landlord data stored and processed, for how long, and is any of it used to train the supplier's models?
- Does every automated message appear in the tenancy or applicant record, with a time and the text actually sent?
- How do we switch a feature off quickly if it starts making mistakes?
Telling tenants and landlords
People are more relaxed about automated help when they know it exists and know how to reach a person. Add a line to tenancy welcome packs, landlord terms and email footers explaining that the agency uses software to acknowledge and log messages, that a named team reviews them during office hours, and how to call for anything urgent. For landlords, explain which updates are drafted automatically and which advice always comes from their property manager. Clear expectations prevent the complaint that starts with 'I thought I was talking to someone'.
Getting started without disruption
- Pick one task with a visible backlog, such as applicant replies or inspection write-ups.
- Check the supplier's data protection terms and record your assessment before using real tenant data.
- Write the approved facts and writing rules the tool will follow.
- Run it for two weeks with every output reviewed, then move to spot checks only if the error log is short.
- Tell landlords and tenants how messages are handled and how to reach a person at any time.
For the data protection side in more detail, see our guide to data protection checks before using AI tools.
Frequently asked questions
Can AI reply to tenant enquiries on its own?
For factual replies such as viewing confirmations, availability already on the advert or the documents needed, yes, with regular spot checks. Anything involving advice, notices, complaints or vulnerable tenants should be reviewed by a person before it is sent.
Is it safe to put tenant data into AI tools?
Only into tools your agency has approved after checking the supplier's data protection terms, where data is stored and how long it is kept. Never paste identity documents or bank details into a consumer chat app.
Which lettings task should we automate first?
Choose the task with the most visible backlog and the lowest legal risk, usually applicant replies, viewing confirmations or document chasing.
Will AI reduce complaints?
It can reduce complaints caused by silence and delay, which are among the most common in lettings, but only if staff review the queue daily and act on what it shows.
References
The Property Ombudsman (2025). Annual Review 2025 (opens in a new tab) ↑
Ministry of Housing, Communities and Local Government (GOV.UK) (2026). Enforcement measures for landlords: Renters' Rights Act 2025 (opens in a new tab) ↑
Propertymark (2026). Housing Insight Report, August 2026 (opens in a new tab) ↑
GOV.UK (2026). Deposit protection schemes and landlords: Overview (opens in a new tab) ↑
Ministry of Housing, Communities and Local Government (GOV.UK) (2025). Electrical safety standards in the private and social rented sectors: guidance (opens in a new tab) ↑
Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑