Skip to main content
Forwardcycle
Menu

AI call triage for repairs and emergencies in property

By Keith Hamilton · · 4 min read

AI call triage for repairs sorts every call to your repairs line by urgency, from a dripping tap to a gas smell. If you manage property or run a maintenance service, the hard part is not answering the call but working out what is an emergency, what is urgent and what can wait for a booked visit. This article explains how AI call triage for repairs and emergencies works, why the rules must come from you, and how to keep residents safe when something serious is reported.

What does triage mean for repair calls?

Triage means deciding how quickly a reported problem needs a response, and who should deal with it. In property and maintenance businesses, calls fall into three groups:

  • Emergency: a risk to someone's safety or serious damage to the property, needing action now
  • Urgent: something that seriously affects the home, needing attention within a short, defined time
  • Routine: repairs that can be booked in the normal way

An AI agent asks the questions, sorts the call into the right group, records the details and passes it on. What it must not do is invent the definitions. Those come from your repair policy, your contracts and, for some landlords, the law.

How do you set the priority definitions for AI triage?

Write down your categories and the response time for each, with examples, before any AI is involved. Be specific. "No heating" may be urgent in winter and routine in summer under your policy, or urgent all year for an older or disabled resident.

Social landlords and some other organisations have legal or regulatory timescales for certain hazards. Check your obligations and make sure the triage rules reflect them.

If your current definitions live only in an experienced colleague's head, writing them down is valuable in its own right.

What does an AI repairs reporting line call look like?

An AI repairs reporting line follows a set path on every call:

  1. The agent says it is an AI assistant and asks what the problem is
  2. It checks first for immediate danger, such as fire, a gas smell, flooding near electrics or someone hurt
  3. It confirms who is calling and which property
  4. It asks a short set of questions for the type of problem
  5. It tells the caller what happens next and when
  6. It sends a text confirming the reference, and a link to upload a photo where one would help

How should AI handle emergency repair calls in property management?

For anything that suggests immediate danger, the agent follows a fixed script you have approved, never an improvised answer. For example, if someone reports a gas smell, it should tell them to leave the property and ring the National Gas Emergency Service on 0800 111 999. If someone is in danger, it should tell them to call 999.

For emergencies that fall to you, such as a major leak, the agent contacts your on-call person directly rather than only logging a job. Agree how that works: a phone call, a text, or both, and what happens if the first person does not respond within a few minutes.

This matters most for out-of-hours repairs triage, when nobody in the office is there to spot a serious call.

When in doubt, escalate

Uncertainty always moves a call up, never down. If the agent cannot tell whether a leak is contained or whether electrics are affected, it treats the call as the higher priority. A wasted call-out costs far less than a missed emergency.

How should AI treat vulnerable residents?

The agent should recognise signs of vulnerability, raise the priority where your policy says so, and pass the caller to a person when that is the right thing to do. Callers who need more care include older residents, people with disabilities or health conditions, families with young children, and anyone who sounds distressed or confused.

Your records may already flag some of them. Use those flags in the triage rules.

What record should each triaged call leave?

Every triaged call leaves a record your team can trust: who called, about which property, what they reported, the answers to each question, the category given and what the agent told them. If your repairs system accepts jobs automatically, the record goes straight in, without anyone re-keying it.

These records include personal data, so UK GDPR applies. Our UK GDPR and AI checklist covers the basics.

How do you test AI for maintenance call handling before going live?

Run realistic scenarios through the agent and have someone who knows your repairs policy sign off every result. Triage is one area where testing matters more than usual. Before the agent takes real calls:

  • Write out realistic scenarios for each category, including awkward ones
  • Run them through the agent and check every categorisation
  • Include calls where the caller is vague, upset or describes two problems at once
  • Have someone who knows your repairs policy sign off the results

Once live, review a sample of calls every week, and every call categorised as an emergency.

How do you measure AI call triage for repairs?

Track these five measures:

  • The share of repair calls answered, day and night
  • Time from an emergency call to contact with your on-call person
  • Calls where your team changed the category after review, and why
  • Repeat calls about the same repair
  • Complaints about how a repair report was handled

Where Forwardcycle fits

Forwardcycle helps property and maintenance businesses turn their repair policies into clear triage rules and design AI call handling that escalates safely to people. For how AI and your team share the work, see blending AI and people in a contact centre. Find out more about our customer experience services or take the free AI Readiness Assessment.

All insights

Choose your next step

  • Explore on your own

    About ten minutes. See where AI could release time in your business.

  • Discuss a specific problem

    A free 30-minute video call with Keith about your situation.

  • Quick question

    Ask about our services or the right next step, at any time.