Thousands of tickets a month poured in by app, email and phone, and they all landed in the same queue. A dead light bulb sat ahead of a flood in a server room, because a person had to read every one, work out the building, guess the urgency and phone a technician by hand. Tenants waited, and the wrong trade kept showing up.
A single undifferentiated queue meant urgency was whatever an agent happened to notice, so a burst pipe could sit behind a burnt-out bulb until someone read far enough down the list, and the contractual response times promised to tenants slipped.
The routing was manual and error-prone. An agent had to read or listen to each request, work out which of dozens of buildings it came from, decide the trade, and phone around for whoever was free, which is exactly how an electrician ends up dispatched to an air-conditioning fault.
And tenants were left in the dark. With no status visibility, the only way to find out what was happening was to call the help desk again, which piled more inbound calls onto the same overloaded team and made the bottleneck worse.
Classify every request the moment it lands, dispatch by location and skill, and keep the tenant informed automatically.
The engine reads each ticket in plain language, "there's a huge puddle under the kitchenette sink on floor 4", identifies the fault, and assigns the right category instantly, so nothing waits for a person to interpret it first.
It sets urgency from the content, so a water leak or a server-room emergency is flagged high-priority and surfaces immediately, while the dead bulb takes its place further down instead of blocking the line.
Once classified, the algorithm checks technician and subcontractor schedules by geolocation and skill and sends the job straight to the most suitable available person's mobile app, so an air-conditioning fault never goes to an electrician again.
An autonomous status channel keeps the tenant updated in real time, with the technician's ETA and an automatic SMS to confirm the job is done and rate the service, so people stop calling the desk just to ask what's happening.
Because the right trade receives the exact fault description up front, the technician turns up with the correct parts and context, which is what lifts the share of jobs fixed on the first visit instead of a wasted call-out and a repeat.
Ambiguous or high-risk tickets are routed to a human dispatcher rather than forced through, so the automation runs the routine flood at speed while a person keeps judgement over the exceptions that need it.
Faster assignment, fewer wasted call-outs, and a help desk that stopped ringing off the hook.
Initial response and assignment fell from an average of 45 minutes to under 3 minutes with no human touch, so the emergencies get moving straight away and the SLAs promised to tenants hold instead of slipping in a manual queue.
Sending the right trade with the exact description lifted the first-time fix rate 35%, cutting wasted call-outs and maintenance cost, and because tenants now get full status visibility, inbound calls to the help desk fell about 60%.
Reading a ticket is the easy half; the value is turning that into the right trade, already nearby, arriving with the exact fault in hand, and a tenant who never has to chase. It is the same intent-classification-and-routing discipline behind our ticket-triage work, aimed at building maintenance and field crews rather than a support inbox.
A large property manager had 8 people re-keying data by hand. We automated it end to end around a 15-year-old ERP. Average handling time fell from 4 hours to 30 seconds.
Read the case → Fleet routing · LogisticsHundreds of trucks were dispatched by phone and gut feel. Unified live data plus AI routing and backhaul matching cut queue time 30% and moved 15% more loads a day.
Read the case →
Reef TRH