How to Turn One Podcast Episode Into a Week of Content With Claude: 7 Pieces From One Recording

You sat down, hit record, and talked for forty-five minutes about something you actually know. Then you published the episode, posted the link once, and moved on.

That is the part that hurts. You made the hardest thing already. Nobody wants to sit down and write a blog post from scratch on a Tuesday, but you just said 6,000 words of good material out loud and let almost all of it disappear into a podcast app.

One recording is not one piece of content. It is a week of it. The blog post is in there. The email is in there. So are the social posts, the clips, the carousel, and probably next month's freebie. The only reason it does not come out is that pulling it apart by hand takes longer than recording it did.

That is the job Claude is genuinely good at. Not inventing things you never said, which is where AI content goes wrong. Taking what you already said and reshaping it for the places your people actually hang out.

I have trained more than 7,500 people at this point, and this workflow is the one that makes creators exhale. Here is how to run it.

Step 1: Get a clean transcript and give Claude the context

Export the transcript from wherever you record. Riverside, Descript, Zoom, even the auto-captions off YouTube will do. It does not need to be pretty. Timestamps are a bonus because they make the clip step much faster later.

Before you paste it in, spend two minutes on setup. Open a Claude Project called something like "Podcast repurposing" and put a short note in the instructions: who you are, who listens, what you sell, and how you sound. Something as plain as this works:

"I run a bookkeeping practice for creative freelancers. My listeners are solo business owners who hate numbers. I sell a monthly bookkeeping service. Write the way I talk: short sentences, no jargon, warm, a little blunt. Never use corporate language like leverage or unlock."

That note is the difference between content that sounds like you and content that sounds like a press release. You write it once. Every episode after this one inherits it.

Then paste the transcript in and ask for one thing before anything else: a summary of the ten strongest ideas in the episode, in your own words, with the best quote for each. Read that list. It takes ninety seconds and it tells you whether Claude actually understood the conversation. If the list is off, fix it now, not after you have generated fifteen pieces of content built on a wrong reading.

Step 2: The blog post that is not a transcript dump

The mistake almost everyone makes is publishing the transcript with a headline on top. Nobody reads that. It rambles, because talking rambles, and search engines have gotten very good at knowing the difference.

Ask for a proper article instead. Pick the single strongest idea from your list of ten, not all ten, and have Claude build a 1,200 word post around it using your own examples and stories from the episode. Tell it to keep your phrasing where the phrasing is good and cut the tangents.

Give it the structure you want: a real headline, sections with subheads, short paragraphs instead of walls of text, and a frequently asked questions section at the bottom. That last part is what gets pulled into AI search results when someone asks a question you already answered.

One episode can carry two or three blog posts if it was a good conversation. Do not write them all in the same sitting. Publish one this week, keep the other in a drafts note, and you have next week half done.

Step 3: The email to your list

Your email list is the audience you own, and it is the one most podcasters neglect because writing an email on top of everything else feels like too much.

Ask for a short one. Two hundred and fifty words, one idea, one link to the episode. Tell Claude to open with the story or the moment from the recording, not with "in this week's episode we discussed." Nobody has ever gotten excited about that sentence.

The version that works sounds like you forwarding something to a friend. "My guest said something on Tuesday that I have not stopped thinking about." Then the idea, then why it matters for the person reading, then the link. That is the whole email.

Step 4: Five to seven social posts, written differently for each platform

This is where the volume comes from. Ask for the ideas first, then the posts, in that order. Have Claude pull the seven most quotable or most useful moments from the transcript, then write each one up for the platform you actually post on.

Be specific about the platform, because the same idea has to wear different clothes. LinkedIn wants a story with a lesson and room to breathe. Instagram wants a hook in the first line before the caption gets cut off. TikTok and Reels want a spoken hook, which means you are writing a script, not a caption.

Ask for three hook options per post so you get to pick. Reading three and choosing one is fast. Staring at a blank box is not.

Step 5: Three to five short video clips

If you recorded video, or even just audio with a camera on, the clips are the highest-value thing in the whole episode. They travel further than anything else you will publish.

Ask Claude to go through the transcript and find the moments that work standing alone: a strong opinion, a surprising number, a story with a beginning and an end, a myth being corrected. Ask for the start and end timestamps and a one-line reason each clip works. If your transcript has timestamps, this turns a forty-minute scrubbing job into a five-minute cut.

Then have it write the on-screen text and the caption for each clip while it is still holding the context. You are not going to want to come back and do that later.

Step 6: A carousel or a checklist

Somewhere in a good episode there is a list. The five things people get wrong, the four questions to ask before you hire someone, the three-step version of the thing you explained for ten minutes.

Ask Claude to find that list and turn it into eight slides: one idea per slide, a handful of words each, a cover slide with a promise and a last slide with one clear next step. That is a carousel, and carousels get saved, which is the thing the algorithms care about most.

The same list is also your lead magnet. A one-page checklist people can download is often better than a 30-page guide nobody finishes, and you just found it inside a recording you already made.

Step 7: Next week's episode, planned for you

Before you close the chat, ask one more question: based on this transcript, what did I open up and not finish?

Every episode leaves loose ends. Something you mentioned in passing that deserves its own conversation. A question you answered too quickly. A guest comment you wanted to push on and did not have time for. Claude reading back through the transcript will find five of those in a few seconds.

Now you have a blog post, an email, seven social posts, four clips, a carousel, a lead magnet, and next week's topic. From one recording. And you did not write a single thing from a blank page.

The rule that keeps this from sounding like AI

Everything above works because you are not asking for new ideas. You are asking for reshaping. The thinking already happened when you were talking.

So keep two guardrails. First, tell Claude explicitly that it can only use what is in the transcript, and to flag anything it is unsure about instead of filling the gap. That single instruction kills the invented-statistic problem, which is how repurposed content usually gets someone in trouble.

Second, read everything before it goes out. Not to rewrite it. To catch the two or three lines that are not how you would say it, fix those, and hit publish. That pass takes ten minutes and it is the entire difference between content that sounds like your show and content that sounds like everybody else's.

What a realistic week looks like

Record on Monday. Run the transcript through your Project on Tuesday morning while your coffee is still hot. Blog post goes up Tuesday, email goes out Wednesday, social posts get scheduled Thursday for the following week, clips get cut whenever you have thirty quiet minutes.

The whole pass is under an hour once you have done it twice. The first time takes longer because you are still writing the instructions and figuring out which phrases to ban. That hour is the setup you never have to do again.

Frequently asked questions

What if my podcast is just me talking, no guest?

It works exactly the same, and honestly a bit better. Solo episodes stay on one topic, which makes the transcript cleaner and the pulled quotes stronger. Interview episodes give you more variety but wander more, so you will lean harder on that list of ten ideas to find the good parts.

Will people notice the same content showing up everywhere?

Almost nobody follows you in more than one place, and the ones who do are your biggest fans and are not annoyed. The bigger risk is the opposite one, where you say something valuable once and 95 percent of your audience never sees it. Repeat yourself more than feels comfortable.

Do I need a paid transcription tool?

No. Most recording platforms include transcripts now, and YouTube will caption a video for free if you upload it unlisted. Accuracy matters less than you think here, because Claude reads around the typos. If it keeps mangling a name or a product, correct it once in the chat and move on.

Can I do this with old episodes, not just new ones?

Please do. This is the fastest content win available to anyone who has been publishing for a while. Go back to your ten best episodes, run each one through this, and you have a full quarter of content sitting in a folder. Nobody remembers what you said eighteen months ago, including you.

What if the drafts come back flat?

Tell it so, plainly. "Too formal, write it the way I would say it out loud." "This lost the story about the client who called me crying, put that back." One round of honest feedback fixes most of it. Accepting the first draft is the single most common reason people decide AI cannot do their voice.

Where do I actually keep all of this?

Anywhere you will look again. A folder per episode, a notes app, a spreadsheet with a row per piece. The system matters less than having one place, because the failure mode is not bad content, it is good content you generated once and never found again.

You already did the hard part

You showed up and talked. That is the part most people never manage. Everything after that is shaping, and shaping is the part you can hand off.

If you want to build this with someone walking you through it, that is exactly what we do inside Club Jam. There are 220+ on-demand modules, live calls where you can ask about your own show, and copy-paste prompts you can use the same day. It is $47 a month and the first 7 days are free, so you can try the whole workflow on your next episode before you decide anything.

Come record something, and let's stop letting it disappear. 🧡