Docs
Back to app

Recording Sources

Elly's Notetaker can capture meetings from several places. This page explains each source and how recordings flow into your Meetings list.

Recall bot for video calls#

For video calls on Zoom, Google Meet, or Microsoft Teams, Elly sends a recording bot ("Recall") into the call.

  • The bot joins as a participant with a visible name (e.g., "Elly Notetaker").
  • Hosts may need to admit it from a waiting room.
  • The bot records video and audio, which Elly processes into a transcript and summary after the call ends.

You can attach the notetaker to meetings you schedule through Elly (via the Invite notetaker toggle), and — depending on your calendar integration settings — to externally scheduled meetings on your connected calendar.

Twilio for phone calls#

For outbound phone calls placed from Elly, Twilio provides the call infrastructure.

  • Calls are initiated from a candidate's profile (Call button).
  • Both sides of the call are recorded, with a recording disclosure played at the start.
  • When the call ends, the recording is uploaded, transcribed, and summarized automatically.

Phone-call support is subject to Twilio's per-country availability.

Vapi for AI interviews#

When candidates complete an AI voice interview (see the Interviewer section), the interview audio is captured and processed through the same pipeline as other meetings.

AI interview recordings appear in the Meetings list alongside other calls, but they're also linked directly to the candidate's pipeline card for quick access.

Manual file upload#

You can upload meeting recordings you captured elsewhere — for example, a phone call recorded on your device, or a video export from another tool.

Supported formats:

  • Audio: MP3
  • Video: MP4

Upload from the Meetings page using New meeting → Upload file, and choose the candidate and job the recording is associated with.

Processing pipeline#

Every recording — regardless of source — follows the same chain:

  1. Store — The raw recording is saved.
  2. Transcribe — Speech is converted to text, with speaker labels where available.
  3. Summarize — An AI-generated summary is produced, structured by the scorecard template attached to the meeting's stage (see Scorecard Templates).

Typical turnaround is a few minutes after the call ends; longer for long recordings or during peak load.

Credit cost#

Each meeting summary costs 5 credits. AI interviews cost 10 credits total (the interview itself plus the summary).

If your workspace doesn't have enough credits when processing runs, the recording and transcript are still stored, but the summary is marked Completed (withheld). You can regenerate the summary after your credits reset or after upgrading your plan. See Transcripts and Summaries for how regeneration works.

Next steps#

Last updated