Water damage calls do not arrive during office hours. A burst supply line at 2am, a sump failure during a storm, a ceiling collapsing while the owner is asleep — the first crew that answers gets the job, and the second one gets a voicemail that the homeowner never returns. An AI receptionist answers every one of those calls, triages how bad the water is, and puts a booked inspection on the schedule before the caller tries the next number on the list.
“My basement is flooding right now” — the caller is asked when it started, whether the water is still running, and whether the source is a pipe, appliance, sewage backup, or groundwater. Severity and water category are recorded before a human picks up.
“Do you handle insurance claims?” — answered with your real policy: whether you bill the carrier directly, whether the homeowner must file first, and what documentation you provide. It never estimates a claim payout or promises coverage.
“How fast can someone get here?” — answered with your actual response window for the area the caller names, then a booking for the inspection slot.
“How much will this cost?” — answered only with the inspection fee and any published minimums you supply. Category-3 sewage, mould, and contents pack-out jobs are never quoted blind over the phone.
“Is it safe to stay in the house tonight?” — the agent gives the standing safety script you approve (power off at the breaker for wet electrical, do not run HVAC with standing water, evacuate for sewage) and escalates immediately.
“I have a tenant / I'm the property manager for 12 units” — captured with unit count and site address so commercial and multi-unit work is routed separately.
It does not give restoration advice, diagnose mould, interpret an insurance policy, or quote a mitigation total over the phone. Those calls are transferred to the on-call project manager with the intake notes attached.
It does not take card payments or hold claim numbers, policy numbers, or medical details from anyone in the household. Intake stops at name, callback number, address, and what is wet.
It does not promise an arrival time it cannot verify. It books the inspection window you configured and tells the caller that dispatch will confirm.
Mould colonises wet drywall and framing within 24 to 48 hours, so mitigation crews sell speed. The homeowner calling at 2am is often already on their phone to three companies at once, and a live answer with a triage question reads as competence in the way a voicemail never does.
A single category-2 or category-3 mitigation job commonly runs into the thousands in equipment days, demolition, and drying — one recovered job a quarter is worth more than the monthly fee, and the after-hours calls are the ones nobody in the trade wants to answer personally.
The agent does not replace your job management stack (DASH, Encircle, Xactimate workflows, Luxor, PSA, or a spreadsheet). It writes the intake into your inbox or CRM as a structured lead, books the inspection against the calendar you expose, and hands off to the on-call rotation. Nothing is migrated, and if you stop, the number returns to your carrier.
A number for the agent, or a forward from the existing after-hours line.
Your response window per service area, and the radius you actually cover.
Which jobs you refuse — sewage without PPE, hoarder cleanouts, jobs over a certain size, anything outside the county.
The on-call rotation and the hour at which it changes hands.
Inspection fees and any published minimums you are willing to quote out loud, plus your insurance paperwork policy.
Week 1: you send the service radius, refusal list, safety script, and insurance policy wording. We build the agent and test it with you on a private number, including a fake 2am sewage call.
Week 2: the after-hours line forwards to the agent. We watch every call for the first 48 hours and tighten the triage wording wherever a caller gave the wrong category.
Week 3 onward: live. You get a weekly email with call volume, jobs booked, category mix, and any call that ended without an inspection — the edge cases become rules in minutes.
Can it tell a category 1 from a category 3 call? It asks the questions that separate them — source, contamination, standing time, sewage involvement — and records the answer. It does not make the final call; the project manager confirms on arrival.
What happens when a caller is panicking? The agent reads a calm, fixed intake order: what is wet, whether water is still running, where the shut-off is, whether anyone is unsafe. Panic is handled by script structure, not by improvisation.
Does it work for commercial and multi-unit jobs? Yes. Unit count, site address, and building contact are captured separately so commercial dispatch is not mixed into the residential queue.
Can it text the homeowner after the call? Yes — a short confirmation with the booked window and what to do before the crew arrives, if your carrier supports A2P messaging.
How quickly can it go live? Two to three working days from the moment the scripts and calendar are supplied, and you hear it answer a test call before it touches a real customer.
It asks the questions that separate them — source, contamination, standing time, sewage involvement — and records the answer. It does not make the final call; the project manager confirms on arrival.
The agent reads a calm, fixed intake order: what is wet, whether water is still running, where the shut-off is, whether anyone is unsafe. Panic is handled by script structure, not by improvisation.
Yes. Unit count, site address, and building contact are captured separately so commercial dispatch is not mixed into the residential queue.
Yes — a short confirmation with the booked window and what to do before the crew arrives, if your carrier supports A2P messaging.
Two to three working days from the moment the scripts and calendar are supplied, and you hear it answer a test call before it touches a real customer.