Dictate the handshake, not the follow-up email.
The two seconds that lose a sale live in a voice note nobody will ever transcribe. Put them in the CRM as notes — and never keep the audio.
Kriti Kapoor · Voice & field sales · 6 min read
The two seconds that lose the sale
Direct answer: the most valuable thing a rep hears all day is said twice, in passing, two seconds after the handshake: what the prospect actually worried about. It is never typed. It becomes a follow-up with no memory attached.
Typing it is the friction. Standing in a hall with a bag and a badge, nobody opens a form to record “she said the rollout timeline is the blocker, not the price.” So it stays in the rep's head, competing with forty other heads'-worth of noise on the ride home.
A dictation capture exists for exactly that moment: hold to speak, let go, done. The note is in the CRM under the right contact, tagged and attributed, while the handshake is still warm.
What happens to a voice note
Direct answer: the recording exists for exactly as long as it takes to become notes: transcribe, attach, attribute, delete. The CRM keeps the record; the voice never becomes data.
The two-second note becomes text as soon as it reaches RareMachine. With no signal in the hall, WhatsApp holds the message in its outbox and sends it when signal returns.
The transcript lands on the right contact — resolved by scan, QR, or the thread the rep captured from — with event and rep ownership already attached.
The note counts to the rep on the roster, in the event’s time zone, like any other capture. The leaderboard is blind to whether the input was voice or type.
The audio is discarded immediately after transcription. There is no voice library, no export of recordings, nothing to recover later.
In the CRM the transcript is an ordinary note on the contact, created through the CRM's own notes API (HubSpot Notes API (2026), Zoho CRM Notes API (2026)), so it can be read, edited and deleted there like any note a rep typed.
Why the audio is never stored
Direct answer: because the recording is a replayable copy of a person talking to a stranger in a crowd, and a transcript of the same moment is a business record. Storing the first to get the second is a privacy decision the product never needs to make.
A transcript is attributable — it says who said what, when, in which event, to which rep, and it can be edited and deleted like any other note. A WAV file is a souvenir of a conversation nobody consented to having played back. The moment a capture pipeline stores audio “just in case”, it has built a liability for a feature nobody asked for.
The privacy model
Direct answer: the same rule that keeps recordings out of the CRM keeps personal context out of it: personal tags live outside the system of record entirely.
Reps can tag a private note they carry with them — a reminder, a relationship nuance — and that tag is structurally prevented from ever reaching the CRM export. It cannot leak in a data cleanup, a sync, or an audit pull, because it is not in the path. This is how capture earns the reps who would otherwise censor their own notes.
Capture the sentence that closes the deal.
Dictate it at the booth. The CRM keeps the note. Nothing keeps the audio.
Questions, answered
Is the voice recording ever stored?
No. It is transcribed at the moment of capture and the audio is immediately discarded. The transcript becomes a note on the contact; the recording itself is never written to storage.
Why refuse to store audio if the transcript is kept?
The transcript is an attributable record of what a lead said; the audio is a replayable recording of a person you did not have a consent conversation with. Storing the replayable version is a liability the product has no reason to take.
Can a rep dictate without Wifi or cellular signal?
Yes. A voice note is a WhatsApp message, so WhatsApp holds it in its outbox and sends it when signal returns. RareMachine transcribes it when it arrives, attaches it to the contact and replies in the same chat.