Is there a better way to get Fireflies call transcripts into a system besides downloading them one by one?

Ben explained that transcripts are pushed directly into a database (Supabase) via an automated action, some of which run through GitHub daily actions. He demonstrated a table called call_transcripts with over 4,300 records covering every call that's been in Fireflies, storing participants, topics, action items, sentiment, a summary, and the raw transcript text (some entries exceeding 10,000 characters). Claude is connected to Supabase via MCP so the team can query this data directly, and it also feeds other tools they've built.