We built a Claude routine that reads our Zoom webinar transcripts and generates a complete post-webinar marketing report in 90 seconds. Follow-up email, 5 takeaways, audience questions, 10 quotes, 3 LinkedIn posts, content opportunities, audience insights. The report saves to Google Drive automatically. Saves us roughly 3 hours per webinar and makes sure no good moment gets lost.
We host a webinar every Friday through our Meetup community. The webinar itself takes one hour. The follow-up work used to take three.
There was the email to the Meetup group. The recap doc we kept meaning to write. The LinkedIn posts pulled from the best moments. The handful of quotes that would have made great graphics. The list of audience questions we wanted to answer in next week's session.
Most of it never got done. Or it got done badly two days later when we had already forgotten the best parts.
So we built a routine.
What the routine actually does
Every time we wrap a webinar and the Zoom transcript becomes available, the routine runs. It reads the full transcript and generates a single polished report covering:
- A short webinar summary, 1 to 2 paragraphs
- A ready-to-send follow-up email for the Meetup group
- 5 strategic takeaways from the session
- The 5 most valuable audience questions discussed
- 5 short-form content ideas with platform suggestions
- 3 LinkedIn posts in our voice, each focused on a different insight
- 10 quotes and soundbites pulled directly from the transcript, ready for graphics
- A content opportunities section identifying what could become a reel, carousel, blog, or lead magnet
- An audience insights section on what attendees seemed most interested in, recurring pain points, and skill gaps that came up
The report saves to a folder in Google Drive named /Webinars/[Webinar Title] - [Date] and we get a notification when it is ready.
End-to-end, the whole thing runs in about 90 seconds.
What changed in practice
The follow-up email goes out same day instead of three days late. The LinkedIn posts get scheduled by Sunday. The quotes get pulled before we forget which moments mattered.
More importantly, we now treat every webinar as a content source instead of a one-time event. The routine surfaces things we would never catch manually. Lines we thought were throwaway moments turn out to be the most quotable thing in the hour. Patterns in the audience questions reveal what to teach next week.
One webinar produces around 15 to 25 usable content assets. Multiply that across 50 webinars a year and you have a content engine that runs on autopilot off live work you were already doing.
How the routine works, in plain terms
Three pieces wired together:
- A trigger. The Zoom transcript becomes available after the meeting ends.
- Claude does the analysis. We send the transcript, our brand voice guidelines, and a structured prompt. Claude reads it and writes the full report following our format.
- Google Drive stores the output. The report saves to the right folder, named correctly, and we get a Slack ping when it lands.
The whole thing is one workflow. No human intervention between "webinar ends" and "report is in Google Drive."
We use n8n as the automation backbone that connects the pieces. Claude does the actual reading and writing. The same pattern works in Make.com or Zapier if you do not want to self-host.
Why this matters for any team running content events
Webinars, podcast episodes, panel recordings, conference talks, internal all-hands. Anything with a transcript.
The work that follows a content event is almost always the same:
- Pull the best moments
- Repurpose into social
- Send a follow-up to attendees
- Identify what your audience cared about
- File it somewhere you will actually look at later
That work is repetitive, requires reading the whole thing carefully, and never gets done well when you are busy. Which is exactly the kind of work AI is genuinely good at.
We see teams hosting weekly webinars who never do any of this consistently. Not because they do not want to, but because the math does not work. 3 hours of follow-up per week is 150 hours per year. Nobody has that.
A routine collapses that to 90 seconds.
The honest limitations
A few things to know before you build something similar:
- Transcript quality matters. Bad Zoom transcripts produce mediocre reports. We use Zoom's auto-transcript plus light cleanup before processing.
- You still need to edit. The LinkedIn posts and the email get tightened by a human before publishing. Claude's draft is 80% there. The last 20% is voice.
- Brand voice is the real moat. A generic prompt gives generic output. The reason this works for us is that our system prompt has been refined over dozens of runs to match how we actually talk.
- It does not replace strategy. The routine generates assets. It does not decide what to publish, when, or what your content strategy should be. That stays with your team.
The take-home
One Claude routine. One hour of webinar input. A complete marketing report out the other end, every Friday, automatically.
This is the kind of work AI is actually for in 2026. Not replacing the thinking. Replacing the repetitive packaging that surrounds the thinking.
If you host any kind of recorded content, the same pattern works. The question is not whether to build something like this. It is why you have not yet.