Repurpose one podcast or webinar into a week of assets

One recording, turned into a long-form article, clip scripts, a carousel, a newsletter section, and a handful of social snippets, all in your voice, from a single transcript.

What you'll have when you're done

A workflow that takes one recording you already made (a podcast appearance, a webinar, a long internal talk) and multiplies it into a week or more of content across formats. You feed the transcript in once, get back drafts of every asset type, edit, and distribute. The hard, expensive part (the actual thinking, said out loud for 45 minutes) is already done; this is how you stop wasting it.

You spent 45 minutes being smart and published one thing

Every founder does this. You go on a podcast, run a webinar, give a great internal talk, and then the recording goes up (maybe) and that is the end of it. One asset from an hour of your best thinking. Meanwhile the same hour contains a long-form article, several short clips, a carousel, a newsletter, and a dozen quotable lines, you just never extract them, because extracting them by hand is a half-day job. I have done podcasts I was genuinely proud of, said things sharper than anything I would have managed at a keyboard, and then let the whole episode evaporate into a single "great chatting with..." post. The thinking was the expensive part, and I did it, and then I threw away nine-tenths of the output because mining it felt like more work than it was worth.

AI collapses that half-day into minutes. Hand it the transcript and it pulls every reusable piece out, in your voice. The one thing to guard against: it will sometimes "improve" a quote into something you did not actually say, so the rule is every quote and clip-line must match what is really in the transcript.

What you need first

• A recording with a transcript. Most podcast/webinar tools provide one; if not, Granola or any transcription tool works.
• A Claude Project loaded with your voice (the same one from your content engine).
• Light editing tools for the visual pieces: a clip tool (Descript, Opus Clip) for video, your scheduler for text.
• Rights to the content, especially if it was a guest interview (covered below).

Step-by-step

Step 1Get the transcript in front of the voice Project

Drop the full transcript into your content Project. Because the Project already knows your voice, the outputs come back sounding like you, not like a generic summarizer. If the recording was a two-way conversation, tell it which speaker is you.

Step 2Ask for the full asset pack

Prompt it to extract every format at once:

From this transcript, draft a content pack in my voice:
1. A long-form article built around the strongest idea in the conversation.
2. 3-4 short-video clip picks: for each, the timestamp/quote and why it lands.
3. A carousel outline (5-7 slides) on one theme from the talk.
4. A newsletter section summarizing the best insight.
5. Five standalone social snippets, pulled or tightly adapted from what I said.
Use only things I actually said. Do not invent quotes. Flag anything you adapted.

Here is the shape of the pack, illustrative and abbreviated, from a 40-minute podcast on hiring:

Article angle: "Why your first sales hire keeps failing," built around the strongest six minutes.

Clip picks:

• 12:30 "If you can't close it, you can't delegate it." Punchy, stands alone, strong hook.
• 24:10 the story about the rep who quit in 60 days. Narrative, better as a longer cut.
• 31:45 the contrarian line on quotas. Likely to spark debate in the comments.

Carousel: "5 signs you're hiring sales too early," from the minute 18-26 stretch.

Snippets:

• "Founder is salesperson #1. Always." (adapted: tightened from a longer aside)
• "Your first AE inherits your clarity or your chaos."

Notice the clip picks come with timestamps and a reason for each, so you are not re-watching 40 minutes to find the good 30 seconds. And the one adapted snippet is flagged as adapted, so you know which lines to check against what you actually said before they go out under your name.

Step 3Verify the quotes and clips against the transcript

Before anything ships, check the pulled quotes and clip-lines against the transcript. The failure mode here is subtle: the AI tightens a rambling real quote into a crisp one you never quite said. Crisp is good; fabricated is not. Keep the meaning true to what you actually said.

Step 4Cut the clips and edit the text

Take the clip picks to your video tool and cut them, the AI told you which moments and why, a human still makes the cut land. Edit the text pieces for your final voice. This is the same 15-minute editorial pass as the content engine; the AI got you 80% there.

The thing the AI cannot judge is energy. It picks clips by what the words say, but a clip lives or dies on delivery, the pause before the punchline, whether you were animated or flat, whether your hands were moving. So treat its picks as candidates and gut-check each against the actual footage: a brilliant line delivered in a monotone makes a worse clip than a decent line delivered with conviction. Often the best clip is fifteen seconds the transcript made look ordinary, because the magic was in how you said it. Use the AI to narrow 40 minutes to six candidates; use your eyes and ears to pick the two that actually move.

Step 5Stagger the distribution

Do not dump all ten assets in one day. Space them over a week or two through your scheduler. One recording can quietly feed your feed for half a month, which is the whole point of repurposing: maximum mileage from one input. A workable cadence from a single recording: the long-form article anchors week one, two short clips go out midweek and the following week, the carousel lands when engagement usually dips for you, the newsletter section ships in your next send, and the standalone snippets fill the gaps between the bigger pieces. The recording stops being an event you did once and becomes a fortnight of presence, which is what consistency actually looks like for someone who does not have time to create from scratch every day.

How you'll know it's working

Your content output per unit of effort jumps, because you are mining recordings you were already making instead of generating everything fresh. One podcast appearance now visibly powers a week-plus of posts. And the pieces sound like you, because they are literally your words, extracted rather than invented.

When it breaks

• A quote is wrong. It "improved" your words past what you said. This is why Step 3 exists; verify against the transcript, always.
• Everything sounds flat. The Project does not have your voice loaded, or the recording was a Q&A where you said little. Feed it your voice samples and pick a recording where you actually talked.
• The clips do not land. The AI picked moments by transcript, not by energy. Use its picks as candidates; your eye makes the final cut.
• Guest-content rights. If it was someone else's podcast or a guest interview, confirm you can repurpose it before publishing their words.
• Every asset says the same thing. The conversation had one strong idea and the pack just reshaped it five ways. That is fine for a tight talk, but for a wide-ranging one, tell the model to build each asset around a different theme so you get range, not five echoes.
• The article reads like a transcript, not an essay. It summarized chronologically instead of arguing a point. Tell it to pick the single strongest claim and build a real argument around it, using the talk as raw material rather than a script to condense.

Make it yours. A high-production podcast appearance is worth the full pack; a rambling internal all-hands might only yield two clean snippets and a newsletter note, and that is still found money. Match the ambition to the source. If video matters most for your audience, prioritize the clip picks and treat the text as secondary; if you live on LinkedIn or a newsletter, invert it. The transcript supports all of it, so point the effort where your audience actually is.

Where this fits in your harness

This is the multiplier on your content engine, same voice Project, bigger raw input. The transcripts can come from Granola, the same capture layer behind case studies and meeting workflows. Repurposing is how a CEO with no time produces like a media company: do the thinking once, distribute it ten ways.