The Conversation Intelligence mobile app lets field reps record in-person conversations — on-site visits, walk-ins, and other settings where a video conferencing link isn't practical. Once uploaded, in-person recordings go through the same processing as virtual calls: transcription, speaker identification, and CRM sync.
In this article:
- Navigation may look different if your organization uses standalone access to Conversation Intelligence. This article reflects the Salesloft platform experience.
- Who this is for: field-based AEs and SDRs whose prospect conversations happen in person — on-site visits, walk-ins, and similar settings where a video conferencing link isn't practical.
- Availability: iOS only. Update the app from the App Store to get this feature.
- Set up your voice sample (a short recording made once in your profile) before your first in-person recording — see seRecording Quality Tips below for why this matters more here than on virtual calls.
Select Record from either a scheduled meeting on your Home screen, or start an ad-hoc recording with no meeting attached.
Scheduled meetings have a toggle controlling whether that meeting gets recorded. You can override this either way right before you start, even if it was previously set to no.
Before recording starts, confirm that everyone in the meeting has agreed to be recorded. An audio chime plays to mark this. If consent isn't confirmed, the recording won't start.
The recording keeps running if you lock your phone or switch apps.
- On iOS, an active recording shows in the Dynamic Island with pause and stop controls, so you don't need to reopen the app to manage it.
- If a phone call comes in, the recording pauses automatically and resumes once the call ends.
- You'll be notified if you're running low on storage, both before and during recording.
Select Stop — the recording saves locally right away; nothing else is required before you leave the meeting.
From there, you can update the meeting title and add contacts who were there. You can only attach contacts that already exist in your CRM; adding a brand-new contact from the app isn't supported yet.
- If you have a solid connection, Upload now.
- Save as a draft and fill in the details later. Partial entries, including optional fields, are saved automatically if you exit before finishing. The draft stays in your queue until you finish or delete it.
Completed and draft recordings:
- Completed Recordings: recordings that have uploaded and finished processing appear here. A recording won't show as completed until fully processed — uploading adds it to Conversation Intelligence's servers first, then processes it into a transcript.
- Draft Recordings: a recording saved as a draft stays here until fully processed. Select a draft to reopen the meeting details screen, update fields, and upload.
Recording Without a Scheduled Meeting
You don't need an existing meeting to use this feature. For an unplanned conversation, or if a prospect declines to be recorded and you want to record your own recap afterward instead, start a recording and attribute it to an account later, the same way.
Things to Note: these currently log as a standard meeting event. A way to separate "notes to self" from actual meetings is being explored for a future release.
- Your recording appears under Completed Recordings once processing finishes.
- The call is tagged as an in-person call and distinguishable from video-platform recordings in My Calls.
- Speaker labels appear correctly in the transcript (if a voice sample was set up in advance).
Once uploaded, a recording goes through standard processing: transcription, speaker identification, and Smart Topics. Currently, the mobile app shows the transcript and lets you play back the audio; the full AI summary, next steps, and Smart Topics currently display in the desktop app, with mobile display planned for a future update.
Every in-person recording is automatically tagged as an in-person call, distinguishing it from Zoom or Teams recordings in My Calls and in reporting.
Recording Quality Tips
- Place your phone in the middle of the table for the clearest audio.
- Set up your voice sample if you haven't already. It matters more here than on virtual calls, since in-person recordings are single-channel audio with multiple speakers — a voice sample helps the transcription service tell people apart.
Storage and Offline Behavior
- Recordings are written to a secure, sandboxed area of your phone's storage — not your regular Photos or Files app — and cache locally before anything uploads. This means you can keep recording with no signal; the app uploads automatically once you're back on stable Wi-Fi or cellular.
- You'll get a low-storage warning before and during recording if you don't have enough free space for a meeting of typical length.
- Once you run out of storage, you can't record anything new until you upload or delete existing drafts.
- Uploading a recording removes it from your phone's storage. Drafts are the only thing that stay stored locally, and you can hold onto multiple drafts as long as you have space.
- If you uninstall the app while you have unattributed drafts, those recordings are gone for good — upload or attribute anything important before removing the app.
- As a rough reference, a 45-minute call took under a minute to upload on a normal connection (audio only, so file sizes are small). More specific storage-per-minute guidance is still being finalized.
|
Problem
|
Likely cause
|
Fix
|
|---|
|
Upload stuck at "Saved Locally"
|
Weak or unstable Wi-Fi/cellular connection
|
Uploads are intentionally skipped on weak connections and resume automatically once the connection improves
|
|
Low storage warning
|
Insufficient free space on your phone
|
Free up space, or upload/delete existing drafts to clear local storage used by the app
|
|
A call interrupted my recording
|
Expected behavior
|
The recording pauses automatically for incoming calls and resumes once the call ends
|
|
Speaker labels are missing or wrong
|
Voice sample not set up
|
Without a voice sample, the transcript comes through as single-speaker output even with multiple people talking — set up your voice sample in your profile
|
|
Processing is taking longer than expected
|
Normal range
|
Upload plus processing typically takes 10–15 minutes
|
