Resolved -
Between 11:45 and 12:16 UTC, our transcription system was unavailable. During this window, meeting transcripts could temporarily not be viewed, and transcripts for newly recorded meetings were not processed. As a result, some meetings recorded on the morning of Aug 7 received their transcript and AI features (summaries, follow-up emails, insights, scorecards) with a delay.
No data was lost. All meeting recordings were captured and stored normally throughout the incident. Affected meetings have since been processed, and all transcripts and AI follow-ups have been delivered.
Timeline (UTC)
- 11:45 A failed deployment left the transcription system in a faulty state; transcript viewing and processing began returning errors.
- 11:46 Incident identified, investigation started.
- 12:16 Fix deployed; transcript viewing and processing of new meetings restored.
- Afternoon Meetings affected during the window were identified and reprocessed. All delayed transcripts and AI follow-ups delivered.
What we're doing
We've identified why the failed deployment wasn't rolled back cleanly and why affected meetings weren't retried automatically. We're adding automatic retries and monitoring so a short disruption like this can no longer delay transcripts without our systems catching up on their own.
Aug 7, 13:31 UTC