A tenant sends a message at ten to eight in the morning, on the way out of the door, saying the thing under the sink is leaking again and could someone come. There is no flat number and no building. Nobody can tell whether "again" means last week or last year, whether it is a drip into a bucket or water through the ceiling below, or whether anyone will be in to let a plumber through.
Property managers get messages like that all day. Each one triggers a reply asking for the details, which arrives in the tenant's inbox after they have gone to work, which gets answered that evening, which your team reads the following morning. Two days in, nobody has looked at the leak.
An AI receptionist for property management fixes the front of that chain. A website assistant can ask every tenant the same set of questions at the moment they report a problem, sort the request by urgency and hand your team something they can act on. It also needs one rule it never bends: emergencies go to a phone, not a chat window.
Intake questions that make a request usable
The goal is a request a coordinator can assign without replying. That means asking the right questions in a sensible order, and not asking twenty. A good sequence for most portfolios:
- Is anyone in danger right now? Gas smell, fire, sparking, flooding or someone unable to get out. If yes, stop and give emergency routes (see below).
- Property address and unit. Offer a format hint, such as "Flat 4, 12 Mill Road".
- Where in the property. Kitchen, bathroom, bedroom, communal area, outside.
- What is wrong, in their words. Let them describe it freely.
- When it started and whether it is getting worse.
- What they have already tried, such as turning off a stopcock or resetting a trip switch.
- Access. Can a contractor enter with a key, or do they need to be home? Any pets?
- Best contact number and email.
- Photos. Tell them whether photos help and how to send them through your usual route, such as replying to your team's follow-up email.
That is a lot of questions written down, but in a conversation it flows quickly, and the assistant can skip steps the tenant already answered in their first message.
Urgency tiers the assistant can apply
Your team already triages in their heads. Write it down so the assistant does the same thing, and so tenants see consistent answers. The response times below are placeholders; use the ones in your own management agreements.
| Tier | Examples | What the assistant says | Where the request goes |
|---|---|---|---|
| Emergency | Gas smell, fire, electrical danger, major water leak, no secure entry | Call the emergency line or national service now | Captured and flagged, but the phone call comes first |
| Urgent | No heating in winter, no hot water, only toilet not working, fridge failure | Logged as urgent, your stated urgent response window | Ticket, plus a webhook alert to the on-call coordinator |
| Routine | Dripping tap, sticking door, broken blind, faulty extractor | Logged, your stated routine window | Ticket in the normal queue |
| Not a repair | Noise complaint, bin days, parking | Answers from tenant information or routes to housing officer | Captured for the right person |
The assistant is applying your definitions, not making judgement calls. If a request does not fit a tier cleanly, it should log it as urgent and let a person downgrade it.
Emergencies never stop at the chat
This bears repeating because it is the part that matters most. A chat window is not a monitored emergency line. If a tenant describes a gas smell, the assistant's first words should be your instructions for gas (in the UK, leave the property and call the national gas emergency number), followed by your own out-of-hours emergency contact. Only after that should it offer to log the details.
Put that text in the knowledge base word for word, approved by whoever holds responsibility for health and safety. Test it with blunt and indirect wording: "smells like gas in the hall", "the boiler is making a weird smell", "water coming through the light fitting". Every one should trigger the emergency route.
The existing guide on handling tenant issues overnight covers the out-of-hours side in more depth.
Where the requests land
A good intake is wasted if it arrives as another unstructured email. Two options work well.
The first is numbered support tickets. Each request gets a reference the tenant can quote, and your team works through them from an inbox. Tenants chasing "any update on my leak?" can give the number, which saves a search.
The second is a webhook. The assistant sends each request as JSON to a URL you choose, which can feed an automation tool, a property management system that accepts incoming data or a Slack channel for the maintenance team. Our walkthrough on sending data with webhooks shows the pattern, and the integrations page lists common destinations.
Many teams use both: tickets for tracking, a webhook to alert the on-call person when something is tagged urgent.
Tenant questions that are not repairs
Once maintenance is sorted, the rest of the tenant inbox is surprisingly repetitive. These are easy to hand to the assistant if the information is written down:
- Bin collection days and recycling rules for each building.
- Rent payment methods and what to do if a payment will be late.
- Moving out: notice periods, check-out inspections, deposit return process.
- Who is responsible for what, such as light bulbs, garden maintenance or appliance repairs, according to the tenancy.
- Communal areas: bike storage, parking permits, noise rules.
Keep anything tenancy-specific or legal, such as disputes or eviction questions, routed to a person.
Your first step
Write your urgency tiers on one page this week, with examples and response windows your team is confident you can meet. Get it signed off internally. That document is useful even before an assistant exists, because it makes your own triage consistent.
Then draft the emergency text and test it hard. When both are ready, see how a maintenance intake flows on our property management page using a real request from last month's inbox.