Technology AI

Automating routine agency admin

Which estate and letting agency admin tasks to automate first, which to keep human, and the checks that stop automation from quietly lowering quality.

Ava research teamPublished 6 min read

On this page
  1. What automation means in an agency
  2. What the evidence says about jobs
  3. How to choose what to automate first
  4. Tasks to automate first
  5. What to keep human
  6. Checks that keep quality up
  7. Where AI-assisted tools fit
  8. Measuring the time saved
  9. Getting past the usual barriers
  10. Example: viewing feedback
  11. Automation and a part-time team
  12. Common mistakes
  13. Frequently asked questions
  14. References

Key takeaways

  • Start with frequent, rule-based tasks that already follow a clear process, such as confirmations, reminders and status updates.
  • Keep humans on decisions, advice, negotiations, notices and anything involving a vulnerable client.
  • Most UK businesses using AI to improve operations reported no change in headcount; automation mostly reshapes admin work[1][2].
  • Pilot each automation, sample its output every week and keep an error log.

What automation means in an agency

For most estate and letting agencies, automation is less about artificial intelligence and more about rules: when this happens, do that. When a viewing is booked, send a confirmation. When a gas safety record is 60 days from expiry, create a task and email the landlord. When an offer is accepted, generate the memorandum of sale. Many of these features already exist in agency CRMs and are simply not switched on.

Newer tools add judgement-like steps: summarising a call, drafting a property description, reading a certificate to pull out its expiry date. They are useful, but they need more checking. A good automation plan uses both, and is clear about which is which.

What the evidence says about jobs

Headcount change at UK businesses using AI to improve operationsONS, Artificial intelligence in UK businesses: 2023 to 2026.
Headcount change at UK businesses using AI to improve operationsBar chart. No change: 63%; Decrease: 6%; Increase: 1%.No changeNo change: 63%63%DecreaseDecrease: 6%6%IncreaseIncrease: 1%1%Headcount change at UK businesses using AI to improve operationsBar chart. No change: 63%; Decrease: 6%; Increase: 1%.No changeNo change: 63%63%DecreaseDecrease: 6%6%IncreaseIncrease: 1%1%

Source: Office for National Statistics, 2026[1]

Show the data
ItemValue
No change63%
Decrease6%
Increase1%

Of UK businesses using AI to improve their operations, 63% reported no change in headcount, 6% a decrease and 1% an increase[1]. Where businesses use machine learning for data processing, they most often report impacts on administrative or clerical roles, at 41%[2]. The pattern is consistent: automation changes what administrators spend their time on more than it removes them.

Adoption is growing but uneven. Self-reported AI use among UK businesses with 10 or more employees rose from around 12% to around 35% between late 2023 and 2026[3]. Yet 45% of businesses in the 2025/26 cyber security survey said AI was not relevant to their organisation, while 21% reported adopting some AI tools[4]. For many agencies, simple rules-based automation is the right starting point.

How to choose what to automate first

Score each candidate task against four questions. Automate the ones that score well on all four, and leave the rest for later.

  1. Frequency: does it happen many times a week? A task done twice a month rarely repays the set-up time.
  2. Rules: can you write down exactly what should happen, with no judgement calls?
  3. Risk: if the automation gets it wrong, is the damage small and easy to fix?
  4. Data: is the information it relies on already accurate and in the right fields?

The fourth question catches most failed projects. Automating a reminder from a compliance date that is wrong in half the records produces wrong reminders, faster.

Tasks to automate first

Example first wave of automations for a sales and lettings agency, each with a human check.
TaskTriggerWhat is automatedHuman check
Viewing confirmationViewing bookedEmail and text to applicant and seller with time and addressNegotiator checks diary clashes
Viewing feedback requestViewing time passesTask created for the negotiator; feedback form to applicantNegotiator calls the seller
Applicant matching alertsNew listing goes liveMatched applicants notified by their chosen channelConsent checked by the system
Compliance remindersCertificate within 60 days of expiryTask to property manager; email to landlordManager confirms booking
Deposit protection trackingDeposit receivedCountdown task to protection deadlineAdministrator confirms certificate
Rent receipt and arrears flagsRent due date passesFlag on the tenancy and task to callProperty manager makes contact
Memorandum of saleOffer acceptedDocument generated from the recordNegotiator checks and sends
Progression nudgesNo progress note in 7 daysTask to progressorProgressor updates parties
Call summariesCall endsSummary and notes added to the recordSpot-check of a sample weekly

What to keep human

  • Price advice, valuations and any opinion on what a property is worth.
  • Offer decisions and negotiations between buyer and seller.
  • Anything that could be read as a notice about a tenancy. Government guidance notes that a notice to quit can include texts or WhatsApp messages[5], so automated messages about ending or changing a tenancy are a risk.
  • Responses to complaints, and first contact with tenants in financial difficulty or other vulnerable clients.
  • Marketing to people who have not consented. Unsolicited marketing calls must not be made to numbers on the Telephone Preference Service register or to people who have said no[6], and automated campaigns must respect each contact's preferences.

Checks that keep quality up

  1. Pilot with one team

    Switch the automation on for one branch or one negotiator for two weeks before rolling it out.

  2. Sample the output

    Every week, read a sample of what the automation sent or wrote: ten confirmations, five summaries, every compliance reminder for one landlord.

  3. Keep an error log

    Record every mistake with its cause: bad data, a wrong rule or a tool fault. Fix the cause, not just the message.

  4. Give clients an easy way out

    Every automated message should make it easy to reply to a person or to stop messages.

  5. Name an owner

    Each automation has a named person who checks it monthly and can switch it off.

  6. Review quarterly

    Ask whether each automation still saves time. Retire those that do not.

Where AI-assisted tools fit

Once the rules-based basics are running, AI-assisted tools can take on tasks that involve reading or writing. Typical examples in agencies are call summaries added to the record, first drafts of property descriptions from the valuer's notes, extracting dates from scanned certificates and sorting incoming emails by type. Each saves time, and each can be wrong in ways a rule cannot.

Treat their output as a draft. A negotiator reads the description before it goes live; a property manager checks the extracted gas safety date against the certificate the first few times; a manager reads a sample of call summaries each week. If the checking takes longer than doing the task by hand, the tool is not ready for that job yet. Be clear with clients where a tool is involved, and never let one make a decision about a person, such as whether an applicant is suitable.

Measuring the time saved

Before switching anything on, time the task as it is done today. How long does it take to send viewing confirmations for a typical Saturday? How many minutes does an administrator spend each week checking certificate dates? Record the figures, then measure again a month after the automation goes live. Count the time spent checking and fixing as well as the time saved.

Share the results with the team. People accept automation more readily when they can see that it removed the work they disliked, and when the time saved went into calls, valuations and visits rather than into a longer list of admin.

Getting past the usual barriers

Difficulty identifying use cases, cost and a lack of expertise are the most commonly reported barriers to AI adoption in UK firms[7], although 41% of businesses with 10 or more employees reported no barriers at all in mid-2026[8]. In agencies, the use-case barrier is the easiest to remove: sit with an administrator for a morning and list every task they repeat. The expertise barrier usually means starting with features already in your CRM, with your supplier's help, before buying anything new.

Example: viewing feedback

Example. A branch handles around 40 viewings a week. Feedback depends on negotiators remembering to call applicants and then sellers, and on busy weeks it slips. The branch sets up two automations: a short feedback form texted to each applicant an hour after the viewing, and a task for the negotiator the next morning with the applicant's response attached. The negotiator still calls the seller personally. After a month, the share of viewings with feedback logged by the next working day has risen sharply, and negotiators say the calls to sellers are easier because they have something concrete to share.

Automation and a part-time team

About 22% of employees in Great Britain's real estate agencies worked part time in 2024[9]. Part-time teams depend on handovers, and handovers are where tasks get lost. Automations that create tasks and summaries in the shared system, rather than reminders in one person's inbox, make the work visible to whoever is in the office. That is often the biggest practical benefit for smaller agencies: not saving time, but making sure nothing depends on one person's memory.

Common mistakes

  • Switching on every available automation at once, so nobody knows which one caused a problem.
  • Automating a broken process instead of fixing it first.
  • Sending messages that sound robotic or contradict what a negotiator has just said.
  • Forgetting to switch automations off for a property that has been withdrawn or a tenancy that has ended.
  • Measuring messages sent rather than outcomes, such as feedback logged or certificates renewed on time.

Frequently asked questions

What should an estate agency automate first?

Frequent, rule-based tasks with low risk and good data, such as viewing confirmations, feedback requests, compliance reminders, deposit protection countdowns and progression nudges.

What should never be automated?

Price advice, offer decisions, negotiations, complaint responses, first contact with vulnerable clients and any message that could be read as a notice about a tenancy.

Will automation reduce staff numbers?

Most UK businesses using AI to improve operations reported no change in headcount. In agencies, automation mostly changes how administrators spend their time.

How do we check automations are working?

Pilot with one team, sample the output weekly, keep an error log, name an owner for each automation and review them every quarter.

References

  1. Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑

  2. Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑

  3. Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑

  4. Department for Science, Innovation and Technology (GOV.UK) (2026). Cyber security breaches survey 2025/2026 (opens in a new tab) ↑

  5. Ministry of Housing, Communities and Local Government (GOV.UK) (2026). Enforcement measures for landlords: Renters' Rights Act 2025 (opens in a new tab) ↑

  6. legislation.gov.uk (The National Archives) (2019). The Privacy and Electronic Communications (EC Directive) Regulations 2003, regulation 21 (opens in a new tab) ↑

  7. Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑

  8. Office for National Statistics (2026). Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab) ↑

  9. Office for National Statistics (Nomis) (2024). Business Register and Employment Survey: open access (SIC 68310, 68320), 2024 (opens in a new tab) ↑

Book a demo

See Ava handle your calls.

Book a short demo. We’ll walk through your calls and workflow, and agree what Ava should take care of.

Book a demo