01 the business
Residential letting and property management agency
2,400 managed homes, 10 property managers
Requests arrive by phone, text and email, often without the address, photos or access times. Urgent leaks sit in the same queue as dripping taps. Tenants call back repeatedly to ask what is happening.
02 before
The problem
Property managers get maintenance requests through calls, texts and emails, often without the address, photos or access details. Urgent leaks sit in the same queue as dripping taps. Tenants call again to ask what is happening.
- Requests are missing key details.
- Urgency is judged late.
- Tenants chase for updates.
03 the build
What the AI does
- Step 1: Answers tenant calls, texts and chats at any hour and asks the questions a property manager would.
- Step 2: Collects the address, problem, photos and access times and logs a work order.
- Step 3: Sorts the request by urgency using your rules and gives the tenant safety steps for emergencies.
- Step 4: Sends the job to the right approved contractor with photos and tenant contact details.
- Step 5: Updates the tenant when the job is booked, and again when the contractor marks it done.
paused: waiting for a person to approve
How it was put together
- Answered tenant calls, texts and chats at any hour with the property manager's questions.
- Collected address, problem, photos and access times, and logged a work order.
- Sorted urgency with the agency's rules and gave safety steps for emergencies.
- Sent jobs to approved contractors and updated tenants at each stage.
04 human in the loop
Where a person stays in charge
The property manager approves any job over the owner's spending limit and is phoned straight away for emergencies. Contractors are chosen only from your approved list.
- Emergencies trigger a phone call to the on-call manager straight away.
- Jobs over the owner's spending limit need the property manager's approval.
- Only approved contractors are used.
05 rollout
From kickoff to live in 5 weeks
- Weeks 1–2Agree urgency rules, spending limits and contractor lists.
- Week 3Pilot with one property manager's portfolio.
- Weeks 4–5Roll out across all portfolios.
06 after
What changes
Requests arrive complete and in the right order. Managers spend less time on the phone, and tenants know where their repair stands.
Assume 2,400 homes raising 0.5 requests each a month, and 15 minutes of manager time per request to collect details and chase.
2,400 × 0.5 = 1,200 requests. 1,200 × 15 minutes = 300 hours a month of collecting details and chasing.
What to measure
- Requests logged with complete details.
- Time to contractor assignment for urgent jobs.
- Tenant follow-up calls per job.
- Owner approvals turned around within a day.
07 evidence
What the research says
From August 2026, AI systems that talk directly to people in the EU must make clear they are AI, unless that is obvious.
EU AI Act, Article 50: transparency obligations (opens in a new tab) · 2024
Across 5,179 customer support agents, an AI assistant raised issues resolved per hour by 14% on average and by 34% for novice and lower-skilled agents.
NBER, Generative AI at Work (Brynjolfsson, Li and Raymond) (opens in a new tab) · 2023
08 lessons
Mistakes to avoid
- No clear emergency definition. Write it down.
- Silence after logging. Updates stop repeat calls.
- Unapproved contractors. Owners and insurers care.
09 questions
Common questions
- Will tenants know it is automated?
- Yes. It says so, and tenants can ask for a person.
- Which property systems does it use?
- AppFolio, Buildium and PropertyMe are common, plus SMS and WhatsApp.