Problem Statement
Loom Meeting Recordings repeatedly misspell participant names, company names, product names, and other domain-specific terminology in transcripts, AI summaries, and Confluence meeting notes. While users can use Correct all to fix a term within a single transcript, those corrections do not persist to future videos. This makes transcript cleanup repetitive and unsustainable for recurring meetings. Additionally, Loom does not appear to reliably use available identity context from Microsoft Teams participant lists or Atlassian account data to improve name accuracy in transcripts and downstream AI outputs.

 

Suggestion
Add support for persistent transcript corrections through a personal and/or workspace-level custom dictionary, so saved names and terms can automatically carry forward into future transcripts and AI-generated outputs. Also improve participant name resolution by using available meeting context — such as Microsoft Teams attendee lists, invitee rosters, and Atlassian identity/email matching — to map spoken names to the correct participant spelling in transcripts, summaries, action items, and Confluence meeting notes.