Never tells a customer "someone’s coming" until a human has actually said yes.
Anyone can answer the 2 AM call. Dispatch closes the loop after it: the right on-call tech notified, explicit acknowledgment required, escalation until a person confirms — and only then a callback the customer can trust.
Acknowledgment or escalation. Nothing in between.
Every other agent’s failure mode loses a lead. Dispatch’s failure mode is a flooded basement at 2 AM and a customer who trusted a machine’s promise. That’s why acknowledgment-or-escalate isn’t a feature here — it’s the entire design.
The emergency comes in
A burst pipe at 2 AM. The Receptionist answers in under a second, triages it as a true emergency against your rules, captures the details, and hands it to Dispatch.
It checks who’s actually on call
Your on-call schedule — rotations, skills, territories — decides who gets notified. The right tech for the right job, not a blast to everyone.
It notifies — and waits for a real yes
Text first, then a call. The tech has to explicitly acknowledge: "I’ve got it." Silence is not consent, and a delivered notification is not an answer.
No answer? It escalates — relentlessly
Timeout hits, it moves to the next tech in the chain. Then the next. The chain always ends at a human phone number that rings until answered — a guaranteed human backstop, by design.
Only then: the callback the customer can trust
"Mike’s on his way — about 40 minutes." Said only because Mike actually confirmed. The whole chain — who was notified, when, who acknowledged — is on the audit trail.
2:04 AM. A burst pipe and a flooded laundry room.
The Receptionist answers in under a second, triages it as a true emergency against your rules, and hands it to Dispatch. Mike is on call tonight. He gets a text — no answer in five minutes, so his phone rings. Still nothing: the chain moves to Sarah, who replies “I’ve got it.” Only now — with an explicit human acknowledgment on the record — does the customer get the callback: “Sarah’s on her way, about 40 minutes.” Not a comforting fiction, a confirmed fact. The whole chain — who was notified, when, who acknowledged — is on the audit trail you can read over breakfast.
The chain, drawn out
Acknowledgment or escalation at every step — and the customer only ever hears confirmed facts.
No acknowledgment, no promise. The customer is never told "someone's coming" until a human has explicitly said yes — and the whole chain is on the audit trail.
What you see
- Every dispatch chain on the audit trail: who was notified, when, who acknowledged
- After-hours emergency jobs captured — the calls that used to be voicemail roulette
- Escalations that exhausted the chain, flagged loudly so the gap gets fixed
- On-call schedule and escalation rules, editable in plain English
What it never does
No "someone will be right there" until a specific human has explicitly confirmed ownership. If it can’t back the promise with an acknowledgment, it doesn’t make the promise.
Only a human acknowledgment closes the loop. The agent cannot decide an emergency is covered.
If the entire chain somehow exhausts unacknowledged, the customer is told the truth — "we’re reaching our on-call team; if this is an emergency, call ___" — not a comforting fiction.
Dispatch launches at a conservative rung of the trust ladder regardless of how trusted the rest of your fleet is. Stakes set the starting point.
These guardrails are enforced by the system — checked in code before an action runs, not just written into a prompt.
Common questions about Dispatch.
What happens if no tech responds?+
Who decides what counts as an emergency?+
Does it blast every tech at once?+
What does the customer experience?+
The rest of the fleet
Hear it before you buy it.
The voice demo on this site is the same engine that would answer your phone. Try it, then book a demo and we'll map your leaks live.