Dispatch
    Human-confirmed handoff

    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.

    The 2 AM test

    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.

    01

    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.

    02

    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.

    03

    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.

    04

    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.

    05

    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.

    A real moment

    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.

    See it, don't take our word

    The chain, drawn out

    Acknowledgment or escalation at every step — and the customer only ever hears confirmed facts.

    The escalation chain
    Emergency triaged
    Burst pipe, 2:04 AM — matched against your emergency rules, details captured.
    On-call tech: Mike
    iMessage → SMS → then a phone call. Explicit acknowledgment required — silence is not consent.
    Next tech in the chain: Sarah
    Same ladder, same requirement: a real ‘I’ve got it,’ not a delivered notification.
    Fallback human number
    The chain always ends at a phone that rings until a person answers — a guaranteed human backstop.
    The customer hears the truth
    "Sarah’s on her way — about 40 minutes." Said only because Sarah actually confirmed.

    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

    Never fakes an ETA

    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.

    Never marks an emergency handled on its own

    Only a human acknowledgment closes the loop. The agent cannot decide an emergency is covered.

    Never leaves the customer in the dark

    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.

    Never starts with full autonomy

    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.

    FAQ

    Common questions about Dispatch.

    What happens if no tech responds?+
    The escalation chain keeps moving — next tech, then the next, and finally a fallback human number that rings until answered. And if everything somehow exhausts, the customer is told the truth and given your emergency number, never a fabricated ETA.
    Who decides what counts as an emergency?+
    You do, during onboarding: the keywords, the symptoms ("active leak", "no heat in winter"), and who gets woken for what. Dispatch executes your rules — it doesn’t invent its own.
    Does it blast every tech at once?+
    No. It consults your on-call schedule — who’s on, their skills, their territory — and notifies the right person first. Escalation is sequential and timed, not a group page.
    What does the customer experience?+
    A calm, immediate answer at 2 AM, honest expectations while the team is being reached, and a callback with a real name and a real ETA once a human has confirmed. The difference shows up in your reviews.

    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.