applaud
Sign upLog in

The platform · end to end

We reach every patient. You hear what they actually said.

applaud contacts every patient who visited, in your practice's name, asks the same check-in question, reads the answer, and puts all of it — the praise and the complaints — behind your login. Scroll for the whole loop. Or talk to the AI yourself, on the right.

Try Robin right nowBook a demo
  • 300msconfirmed turns on the phone
  • 5,000+simulated and real calls tested before first shipment
  • HIPAAcompliant on signed BAAs — not “eligible”

Robin

Invite only

Maya just left your practice.

Robin texts her. You answer as Maya, in your own words. Then she calls, already knowing what you said.

  1. Text
  2. Read
  3. Call
  4. Report

You'll need your invite phrase

The loop

Five things happen. Every time.

No step here is optional, and no step is decided by a model on its own. This is the whole product in the order it runs.

  1. 01we reach out

    Every patient who visited.

    Not a hand-picked list, not the ones the front desk liked. Each patient is enrolled and the first message goes out from your practice's name and number.

  2. 02we ask

    One question, identical for everyone.

    The same NPS-style check-in template for every patient — a question, not a pitch. Every message carries “Reply STOP to opt out.”

  3. 03we read it

    A model reads the answer, not a form.

    Sentiment, intent, risk and themes from a fixed twelve-term vocabulary. Quotes are only kept if they appear character-for-character in something the patient actually wrote.

  4. 04it forks

    Praise and complaints both go somewhere.

    Nobody is screened out of the ask on the strength of their answer. What changes is where the conversation lands: a review composer, or your feedback queue with your team told about it.

  5. 05it lands

    All of it, behind your login.

    Every patient, every piece of feedback, every posted review — on screens your team can open, plus a report every Friday.

The rails

The AI chooses the words. Deterministic code owns every action.

Robin never claims she sent a text until the engine has confirmed it landed, and no message reaches a patient without clearing the same gate.

  • The model proposes, the code disposes

    Every outbound message passes one deterministic gate before it can leave: consent, opt-out, quiet hours, frequency cap and a PHI lint. The AI writes the words. It does not decide whether they send.

  • Quiet hours on the practice's clock

    9am to 8pm in the practice's local time, checked at the instant of send. The practice's time zone is set when it joins and shown on its account page.

  • STOP is honoured deterministically

    A fixed word list plus the natural phrasings — “take me off”, “no more texts” — checked when the reply lands and checked again at the instant of send, never left to the model.

  • Zero taps wasted

    The review link opens directly in Google's review composer. Not your profile page, not a search result — the box with the stars, ready to type.

  • Follow-through that knows when to stop

    Gentle day-2 and day-5 reminders, then nothing ever again — and the moment the link is clicked or the review lands, the sequence stops. Nobody gets chased for a review they already left.

  • Angry replies go to humans

    A frustrated patient never gets an automated review ask, and an open complaint suppresses any that were already queued. Robin diverts them to service recovery and your team is told while it is still fixable.

The output

And this is what lands. Live in your own dashboard.

The week read back to you: how patients felt, what they kept talking about, who you caught before your listing did, and what actually got posted.

Your week

your practiceFridayprepared by applaud

Specimen

How your patients felt

72/100+4 vs last week

Derived from the conversation, not a survey. It is a reading of what patients said, and it is labelled that way on the report itself.

  • 26 positive
  • 6 neutral
  • 4 negative
  • 2 mixed

38 conversations read this week.

4unhappy patients came to you, not to your listing. We picked them up in the conversation and kept them private.

What they talked about

Mentioned warmly

  • 11the front deskWarm
  • 7how the team made them feelWarm
  • 5being kept informedWarm

Raised as a problem

  • 6waiting on the dayFriction
  • 3booking and reschedulingFriction
  • 2cost and billingFriction

Grouped from a fixed twelve-term vocabulary, so the same thing is called the same thing every week and a trend is a trend rather than a rewording.

Reviews

  • 9verified this week
  • 63total to date
  • 14in the review step — asked, not posted yet

Every review counted here was checked by a person against the patient we contacted. Auto-detected matches don't count until someone confirms them.

Patient names and their exact words stay behind your login, where your BAA covers them — never in email.

Specimen. Every figure, theme and line of feedback above was written for this page. Your report carries your own patients' numbers and your own patients' words, and nobody else's.

Ovation™ — how Robin talks

A voice engine we built, because rented ones lie.

Every Robin call runs on Ovation™, applaud's real-time conversation engine — built in-house, end to end. Not rented from a platform. The console at the top of this page runs the same speech models the phone agent does.

Sub-second replies

Typical answers in under a second. Confirmed turns land in as little as 300 milliseconds.

Thinks while you talk

Ovation drafts its reply while you're mid-sentence. Stop, and the answer is simply there. Keep going, and the draft is thrown away.

Never cuts you off

A breath is not a finished thought. False-interruption filtering gives trailing sentences patience and hesitation grace.

Handles talk-over

Robin listens while she speaks, tells her own echo from a real voice, and stops mid-sentence when you interrupt.

Adapts to your pace

Brisk callers get brisk. Unhurried callers get time. Robin measures each caller's rhythm and drifts toward it.

Filters the background TV

The television, the kids, the kitchen — heard, classified, and ignored instead of answered.

Zero dead air

When Robin needs a moment, she says so out loud — briefly, the way a person would. No bot-giveaway silence.

HIPAA-compliant

The whole pipeline runs on infrastructure covered by applaud's signed BAAs. Compliant — not “eligible”.

Typical voice AI softwareOvation
1.5–3 second responses; “800ms” at bestUnder a second typical — 300ms on confirmed turns
Interrupts callers on natural pauses1–3 interruptions in 20 calls, with false-interruption filtering
Repeated lines — the signature bot tellZero in validation
Dead air while “thinking”Bridged in under a second
Speech during agent turns is lostHeard, classified, and answered

Read the engineering story

Compliance and security

The BAA is signed before we text a single patient.

Signed first

Nothing goes out until the agreement and the BAA are executed. Their status is on your settings screen, in writing, where you can check it.

Minimum-necessary data

First name, phone, visit date. No charts, no diagnoses, no notes. Encrypted at rest, and every subprocessor that touches it signs a downstream BAA.

One gate, and it writes it down

Consent, opt-outs, quiet hours and a clinical-term lint are checked on every channel — and every block is written to an append-only audit log.

See what Robin does for your practice.

Twenty minutes with the team. Your real review numbers, your specialty's scripts, and Robin on the line — live.

Book a demoGet the free audit first