Split session transcripts into tagged chunks
Run it once to turn every unprocessed session transcript in Airtable into tagged, searchable strategy and workflow chunks
Version 1
What it does
You start this workflow by hand. It reads the chunks you already have so it can skip sessions that were processed before, then finds every session in Airtable that has a transcript.
For each new session it sends the title, speaker, categories and transcript (cut at about 45,000 characters) to Claude through OpenRouter, which pulls out each distinct strategy, workflow, framework or prompt. Every chunk is saved as an Airtable record linked to the session, with its title, type, summary, category tags, who it applies to, platforms, format, difficulty and the matching transcript excerpt. Responses that cannot be read are logged and skipped. The newsletter matching workflow searches these chunks.