1. Where AI can help your team
Much of arrears work comes back every month. AI can prepare it or carry out the repetitive parts, so your team has more time for tenants who need a conversation.
- Drafting messagesAI writes a reminder or reply in your tone, adjusted to the amount, the period and anything agreed.
- Choosing the moment and channelAI picks the time and channel, such as email or text message, from when the tenant replied before.
- Handling repliesAI answers routine questions, such as the balance or how to pay, and passes anything else to a person.
- Spotting early signsAI can notice that a tenant pays a little later each month, or that a Direct Debit failed for a tenant who always paid on time.
- Summarising a caseAI sets out what happened, what was agreed and what is still open, so a colleague can pick up the case quickly.
Logging every message, reply and agreement is ordinary software and needs no AI. An early sign is a reason to get in touch, not a judgement about the tenant.
Read more: Rent arrears: from first missed payment to resolution
2. What a person decides, and why
Some decisions have serious consequences for a tenant and need judgement that data cannot capture. AI can gather the information; a member of your team decides.
- Agreeing a payment planThe system can propose a standard plan within limits your team sets. Outside those limits, or when the tenant's situation calls for it, a person decides.
- Possession or eviction proceedingsThese can end with the tenant losing their home, so a person weighs all the circumstances.
- Support for a vulnerable tenantIllness, bereavement, a disability, debt problems or caring responsibilities may call for a different approach.
- Stepping off the usual routePausing reminders, skipping a step or referring someone to debt advice are choices a person makes and can explain.
Read more: Payment plans with tenants
3. How AI and your team work together
The system does the work that is the same for every tenant. Your team steps in as soon as there is something to decide.
- The payment does not arriveThe software sees that the rent has not come in and opens a case.
- A reminder goes outAI drafts it from wording your team approved.
- The tenant repliesAI summarises what the tenant asks or promises.
- A person decides when neededA plan outside your limits, a difficult situation or a request to speak to someone goes to your team.
ExampleExample: one reply, two outcomes
A tenant pays £1,150 rent a month and misses one payment. Your team's rule lets the system propose a plan of up to six monthly instalments. The tenant offers to pay £200 a month on top of the rent.
- Amount owed
- £1,150
- Five instalments of £200
- £1,000
- Sixth and final instalment
- £150
- Plan length
- 6 months, within the team's limit
The system may propose this plan. Had the tenant mentioned a hospital stay, or offered £50 a month, the case would go to a person. The figures are an illustration only.
4. What UK GDPR says about automated decisions
UK GDPR applies across the UK. The Data (Use and Access) Act 2025 replaced the old Article 22 with new Articles 22A to 22D, in force since 5 February 2026.
The rules cover significant decisions made solely by automated processing, which means without meaningful human involvement. A decision is significant when it has a legal or similarly significant effect on the tenant.
Such decisions are now allowed in more situations than before, but only with safeguards. Article 22C lists four safeguards:
- The tenant is told about significant decisions made about them.
- The tenant can give their side of the story.
- The tenant can ask for a person to look at the decision.
- The tenant can challenge the decision.
Stricter rules apply when a decision is based on special category data, such as health information. Such a decision may only be solely automated in limited cases, for example with the tenant's explicit consent.
The ICO, the UK's data protection regulator, published draft guidance on these rules on 31 March 2026; at the time of checking it was still a draft. It says human intervention cannot be tokenistic: the reviewer must be able to change the outcome.
Profiling means using data to assess a person, for example predicting that a tenant who pays later each month will fall behind. Under Article 35, systematic and extensive profiling that leads to significant decisions needs a data protection impact assessment (DPIA), a written check of the risks to people's rights.
Whether a decision in your process counts as significant depends on what it does. Ask your data protection lead or a lawyer to look at your own set-up.
5. The Equality Act and vulnerable tenants
The Equality Act 2010 applies in England, Wales and Scotland; Northern Ireland has its own equality laws. Those who let or manage homes must make reasonable adjustments for disabled tenants when asked, such as large print, phone contact instead of text, or more time to respond.
Section 19 covers indirect discrimination: a rule that applies to everyone but puts people with a protected characteristic, such as a disability, at a particular disadvantage. That is unlawful unless the rule can be justified, so check how a fixed reminder route works for different tenants.
In its Spotlight report on attitudes, respect and rights (January 2024), the Housing Ombudsman for social housing in England found that residents' vulnerabilities are too often missed. It asked landlords to recognise, respond to and record vulnerabilities.
AI can flag words in a reply that point to illness, job loss or bereavement. The flag sends the case to a person.
In England and Wales, a tenant can enter Breathing Space, a legal pause on debt recovery. Creditors must then not ask for payment of the arrears it covers or take enforcement or possession action over them. In a standard breathing space the current rent stays due, so your system has to keep the two balances apart. Scotland and Northern Ireland have their own rules.
Read more: Breathing Space and rent arrears
Reesure sends the reminders, logs every step in the case record and passes the cases that need a decision to your team.
6. What you tell your tenants
If you make solely automated significant decisions, Articles 13 and 14 of UK GDPR say you must tell tenants in your privacy information. You must also give meaningful information about the logic involved and the likely consequences. The Article 22C safeguards mean tenants also need a way to ask for a human review.
Good practice is that you also say where you use AI when a person makes every decision, and that automated replies show how to reach your team.
7. The risks: tone, unequal treatment and wrong data
AI makes mistakes in a different way from people: one mistake repeats for every tenant who fits the same pattern.
- The wrong toneA message sounds threatening to someone who has just lost their job.
- Unequal treatment (bias)A model learns from past data and may treat some groups more harshly, for example by area or language.
- Wrong dataA payment not yet matched to the right rent charge can show up as arrears.
- An incorrect summaryAI may get an agreement wrong or mention something that is not in the record.
- Too much trustA person who sees good proposals every day looks less closely.
Read more: Rent reconciliation at month end
8. Using AI responsibly: a checklist
These are rules you set yourself, on top of what the law asks. Write them into your procedures and review them regularly, for example every quarter.
- Your team approves the wording AI uses as its starting point.
- Decisions with serious consequences always go to a person.
- The person can change or reject what AI proposes.
- The case record shows what AI did and what a person decided.
- The arrears balance matches your accounts before a message goes out.
- A colleague reads a sample of messages and summaries.
- You check whether different groups of tenants are treated the same way.
- Recorded adjustments apply to every message, and every message says how to reach a person.
9. Choosing a system: what to ask a supplier
These questions show how a system handles cases that go wrong as well as right.
- 1Which decisions does the system make on its own?Ask what happens automatically and what always goes to a person.
- 2Can your team set the line itself?For example, the limits for a standard plan and the signals that send a case to a person.
- 3Where does the data come from?Ask how it connects to your PMS, the property management system that holds your tenancies and rent.
- 4What does the system record?Ask whether you can see per case what AI drafted, what was sent and who decided.
- 5How does it handle adjustments and Breathing Space?Ask how these change or stop the messages.
- 6How are bias and tenant data handled?Ask how the supplier tests for bias, where data is kept, whether it trains models and what it provides for your DPIA.
Read more: Arrears management on top of your PMS
10. Starting small
Start where a mistake does little harm and shows up quickly, such as a reminder before the due date.
- Pick one taskFor example, reminders before the rent due date.
- Check everything at firstFor the first few weeks, a person checks every message before it goes out.
- Adjust and reduce the checksFix wording and settings, then move to sample checks once the messages are right.
- Add the next taskThen start again at the first step.
Read more: Rent arrears management
Checked on 20 November 2026 against multiple sources.
This is a general explanation, not advice for your situation.