Automations and AI

The floor keeps moving when you look away.

Relae turns missed callbacks, quiet leads and draft dates into tasks on its own, runs the rules you set, and puts AI on the calls that need a second look.

Rule running
WhenA call is reviewedA lead texts backA follow-up comes due
Ifit has coaching flagsthey mention priceit is a draft-date reminder
ThenFlag it for the managerFlag it for a licensed agentText them a reminder

01 · Rules

One sentence. When this happens, do that.

Pick a trigger, add conditions if you want them, then choose what happens: flag it for a person, or text the lead with their consent checked first.

Example: a call ends and its AI review has coaching flags, so the manager is flagged. A lead asks about price by text, so a licensed agent is flagged. A draft date comes due, so a reminder is texted with consent checked. A lead who opted out is never texted; the run is blocked and logged.

  • A call ends
  • A call is reviewed by AI
  • An agent logs an outcome
  • A lead texts back
  • Someone opts out
  • A follow-up comes due

02 · Follow-ups

No lead falls through the cracks.

Every lead carries a short list of what should happen next. When one comes due, the agent who owns it gets a task. The moment a real person answers or texts back, the stale ones clear themselves.

Nothing texts a lead on its own. Texting only happens where you set a rule for it.

  1. Call me laterDue at the time they named, in their time zone.
  2. Missed callbackA scheduled callback slipped. Someone owes them a call.
  3. Missed appointmentThe time came and went with no outcome logged.
  4. Booked, then silenceThey booked and nobody has touched them since.
  5. Gone quietNo real contact in ten days, so the lead comes back to life.
  6. Draft dateA policyholder’s draft date is close. Check in first.

03 · AI on the floor

AI that helps the humans. Never replaces them.

The AI management layer sits under the whole floor: it rehearses reps, reviews calls, answers texts and screens out machines, then hands every decision that matters to a person.

01Practice prospect

Rehearse against an AI that only objects.

It uses lines from your agency’s own script and never says yes, so a weak rebuttal never gets rewarded.

Five examples. A practice prospect that only objects, using your script. A post-call review that scores close likelihood and flags coaching moments on recorded calls. A text agent that answers a lead and hands booking a time to a licensed agent; it never books a time itself. Voicemail detection that skips answering machines. A per-agent coaching view built from reviewed calls.

04 · Guardrails

Built so it can’t run wild.

  • A rule can only text the person the event is about. You cannot type a phone number into one.

    No typed numbers

  • Turn any rule off instantly. A rule that runs past its budget switches itself off and says why.

    Kill switch and budgets

  • Every decision is written down, sent or blocked, with the reason.

    A run log

Start where you are.