Property operations automation
Maintenance requests triaged, dispatched and closed without a coordinator
Six thousand tenants sent maintenance requests by email, phone, WhatsApp and a portal nobody liked. Three coordinators spent their days asking for photos, calling tradespeople and chasing invoices. We built an AI agent that triages every request, dispatches the right contractor, and keeps the tenant and owner informed until the job is closed.
- Client
- Residential property management company
- Region
- Australia
- Scale
- 40 to 70 employees, 6,000 units managed
- Timeline
- 6 weeks
Representative engagement. Client details are anonymised and figures reflect typical outcomes from comparable deployments.
What the client got
0%
Requests dispatched without a coordinator
Emergencies and ambiguous cases still go to a person.
0 hrs
Median time to contractor assignment
Down from more than a day.
0%
Rise in tenant satisfaction score
Driven by status updates and faster first response.
The bottlenecks
Bottleneck 1
Every request started with a clarification
Tenants wrote “the tap is broken”. Coordinators replied asking which tap, how bad, and for a photo. A day passed before anyone knew whether it was an emergency.
Bottleneck 2
Dispatch was manual and slow
Finding an available plumber meant calling three. Owner approval for anything above the threshold was another email thread.
Bottleneck 3
Nobody closed the loop
Jobs were marked done when the invoice arrived. Tenants were not told; owners found out at the monthly statement.
What we built
Step 1
Multichannel intake with AI triage
Requests from any channel land in one queue. An LLM classifies the trade, severity and urgency, asks the tenant for a photo or clarification if needed, and flags emergencies (gas, flooding, lockouts) for immediate human attention.
Step 2
Rule-based dispatch
Non-emergency jobs go to approved contractors by trade, area and rating, with the request, photos and access details attached. Contractors accept or decline from their phone; declines re-route automatically.
Step 3
Owner approvals inline
Quotes above the owner’s delegated limit are sent for one-tap approval with the AI’s summary of the issue and the recommended option.
Step 4
Status to everyone, invoice to ledger
Tenants and owners receive updates at each stage. Completed jobs trigger an invoice match against the quote and a posting to the trust accounting system.
The automation workflow
Maintenance request lifecycle
Tenant message to closed job
Questions clients ask about this engagement
What counts as an emergency?
A list agreed with the client: gas smell, active flooding, no power, lockouts, security damage. These bypass automation and ring the on-call manager.
Can contractors refuse jobs?
Yes. A decline re-routes to the next contractor by rating and availability. Repeated declines lower a contractor’s ranking.
More in Real Estate & PropTech
Industry overview
AI enquiry handling
Every property enquiry answered and qualified in under two minutes
Enquiries from portals, the website and WhatsApp were answered hours later by whichever agent was free. We built an AI enquiry agent that replies instantly in the buyer's language, qualifies budget and timeline, and books viewings into the right agent's calendar.
2 min
Median first response
Design automation
From client brief to 2D plan and 3D render without a designer touching it
Designers spent the first days of every project redrawing plans and producing concept renders before the client had even committed. We automated the path from brief and measurements to 2D floor plan and first 3D concept, so the studio could pitch in hours and reserve designers for the work clients pay for.
90%
Less pre-sale design time
Commercial real estate outbound
Pre-leasing 68 percent of a commercial tower before completion
A developer with 1.2 million square feet coming to market relied on brokers who brought the same tenants to every building. We built a direct outbound engine to corporate real estate heads, GCC expansion leaders and facilities directors, and the leasing team pre-let most of the tower before the occupancy certificate.
68%
Pre-leased before completion
Revenue operations for brokerages
A brokerage CRM that updates itself, and agents who finally use it
A fast-growing brokerage had 120 agents, a CRM nobody updated, and no idea which marketing spend produced closings. We rebuilt the revenue operations layer so the CRM fills itself from calls, WhatsApp and portals, and management sees a pipeline they can trust.
94%
Of agent activity now in the CRM
Have a similar bottleneck?
Tell us how the work happens today. We will show you the workflow we would build and prototype it on your own process.