Best AI Prompts for Podcasts

AI summary

Seven AI prompts for solo, interview, and panel podcasters: show notes drafting, title generation, audiogram clip extraction, guest pitch letters, pre-interview prep briefs, sponsor read drafting, and engagement pattern audits. Built to scaffold the back-office so your time goes to the conversation.

Podcasting is the medium where the conversation is the work and everything else is back-office. The seven prompts below take the back-office side (show notes, titles, clips, pitches, prep, sponsor reads, analytics) and structure them so the conversation gets your full attention. This is the podcast slice of the AI Prompt Library. For broader content prompts see Best AI Prompts for Creators.

Why do most AI podcast-AI workflows produce show notes nobody reads and titles that lose the click?

The default podcast-AI risk is letting AI write show notes that read like every other AI-written show note. Listeners scroll past; SEO does not work; the back-office becomes wasted effort. The prompts below force the structure to come from actual episode content.

Use AI for the structure and drafting; you bring the episode-specific substance. Run any guest-facing or audience-facing content through How to Edit AI Out of Your Writing. When a prompt becomes weekly, graduate it using the Prompt-to-Workflow Ladder.

What are the seven for podcasts prompts?

Prompt 1

Show Notes Drafter

Most show notes are dumped transcripts that nobody reads. This prompt drafts notes that actually drive listens and SEO.

Episode just recorded:

EPISODE TITLE: [WORKING TITLE]
GUEST (if any): [NAME, ROLE]
KEY TOPICS covered: [BULLETS]
MOST QUOTABLE MOMENTS: [2-3 QUOTES OR PARAPHRASES]
TOOLS, BOOKS, RESOURCES mentioned: [LIST]
MY PODCAST AUDIENCE: [WHO]
EPISODE LENGTH: [MINUTES]

Draft show notes:

1. OPENING HOOK: 2-3 sentences that earn the click. Names the most surprising or useful thing in the episode.
2. ABOUT THE GUEST (if any): 2 sentences, factual.
3. WHAT YOU WILL LEARN: 3-5 specific takeaways, not generic.
4. TIMESTAMPS: 5-8 markers tied to the most useful moments.
5. RESOURCES MENTIONED: with links.
6. QUOTE PULL: 1-2 quotable lines listeners can share.
7. SUBSCRIBE / FOLLOW CTA at the end.

Keep under 500 words. Use specific verbs, not "discusses," "explores," "dives into." SEO-relevant terms appear naturally from real episode content.

When to use: Within 24 hours of recording. · Best model: Claude. Discipline about avoiding podcast-show-note cliches matters.

Prompt 2

Title Generator (Podcast)

Podcast titles get one click each. This prompt produces 10 variants so you pick the strongest.

Episode context:

KEY POINT or argument: [ONE SENTENCE]
GUEST AND THEIR ANGLE (if any): [BRIEF]
MOST SURPRISING MOMENT: [BRIEF]
MY AUDIENCE: [WHO]
PLATFORM TITLE-LENGTH limit: [BRIEF: Apple cuts at ~60 chars]

Produce 10 title variants:

1. SPECIFIC PROMISE: names the takeaway.
2. CONTRARIAN: challenges a common belief.
3. STORY ANGLE: leads with the guest's specific moment.
4. QUESTION FORMAT: poses the question the episode answers.
5. NUMBER OR DATA: leads with a specific stat.

For each:
- Under 60 characters when possible.
- Avoid "How I," "What I learned," "The truth about," "You won't believe."
- Specific enough that only this episode could have this title.

Name your top 3 and the one most likely to underperform.

When to use: After episode is mixed. · Best model: Claude. No-clickbait discipline matters.

Prompt 3

Audiogram / Clip Extractor

Short clips drive subscribers. Most pods extract the wrong clips. This prompt finds the 3 strongest.

Episode transcript or summary:

[PASTE TRANSCRIPT OR DESCRIBE STRUCTURE]

EPISODE LENGTH: [BRIEF]
MY PLATFORMS for clips: [INSTAGRAM REELS / TIKTOK / SHORTS / X]
GOAL: [SUBSCRIBE / ENGAGEMENT / SHARE]

Identify the 3 strongest clips:

1. THE HOOK CLIP: 30-60 seconds. Self-contained. Earns attention in the first 3 seconds.
2. THE INSIGHT CLIP: 60-90 seconds. Delivers a specific, quotable takeaway.
3. THE STORY CLIP: 90-120 seconds. A specific moment or anecdote.

For each:
- Quote the exact line or describe the moment with timestamp.
- Why it works as a clip.
- The caption for the clip (under 30 words).
- The platform-specific adjustment (TikTok needs different opening than LinkedIn).

Do not pick clips because they are general overviews. Pick clips that earn the click on their own.

When to use: Same day as publish. · Best model: Claude. Discipline about clip quality matters.

Prompt 4

Guest Pitch Letter

Most podcast guest pitches read like every other pitch. This prompt drafts one that earns the yes.

Potential guest:

GUEST NAME, ROLE: [BRIEF]
WHY THIS GUEST specifically: [SPECIFIC REASON, not just title]
WHAT I HAVE READ / WATCHED / HEARD of theirs recently: [SPECIFIC]
MY PODCAST'S AUDIENCE and reach: [BRIEF]
WHAT MAKES MY POD different from others they might be on: [SPECIFIC]
WHAT WE WOULD COVER on the episode: [ANGLE]
LOGISTICS: [REMOTE / IN-PERSON, LENGTH, FORMAT]

Draft a 150-word pitch:

1. OPENING: references something specific they did recently (not their bio).
2. THE ANGLE: what we would cover that they probably have not covered elsewhere.
3. THE AUDIENCE: who would be hearing this episode and why they would care.
4. THE LOGISTICS: brief, low-friction.
5. THE CLOSE: warm, professional, no "hoping to hear from you" desperation.

Do NOT use: "Huge fan," "Would love to have you," "Honored to invite," "Quick favor." Avoid flattery; lead with specificity.

When to use: After you have done 15 minutes of research on the guest. · Best model: Claude. No-flattery discipline matters.

Prompt 5

Pre-Interview Prep Brief

Most podcast hosts walk into interviews without doing the work. This prompt structures the prep so the conversation is real.

Upcoming interview:

GUEST: [NAME, ROLE]
THEIR PUBLIC WORK: [BOOKS, ARTICLES, INTERVIEWS, COMPANY]
WHAT THEY ARE MOST KNOWN FOR: [BRIEF]
WHAT I HAVE READ / LISTENED TO recently: [BRIEF]
MY ANGLE for this episode: [WHAT MAKES IT DIFFERENT]
MY GOAL: [SPECIFIC OUTCOME]
MY EPISODE LENGTH target: [MINUTES]

Draft a prep brief:

1. THE 3 THINGS the guest has said many times in prior interviews. Avoid these on my show.
2. THE 3 THINGS the guest probably has not been asked, based on their work.
3. THE OPENING QUESTION that signals I did the work.
4. THE 5 CORE QUESTIONS in priority order.
5. THE FOLLOW-UPS for each core question.
6. THE MOMENT TO HOLD: the place where my instinct will be to interrupt; do not.
7. THE CLOSE: the final question or sentence I want to land.

Do not give me canned interview templates. Build from the guest's actual work.

When to use: Night before the interview. · Best model: Claude. Discipline about going beyond canned questions matters.

Prompt 6

Sponsor Read Drafter

Sponsor reads either flow with the show or sound like they were jammed in. This prompt drafts the read so it feels like part of the show.

Sponsor context:

SPONSOR: [BRAND]
WHAT THEY MAKE: [BRIEF]
WHO THEIR PRODUCT IS FOR: [AUDIENCE]
KEY BENEFIT or differentiator: [SPECIFIC]
OFFER for my listeners: [PROMO CODE / DISCOUNT / LINK]
MY SHOW'S VOICE: [DRY / WARM / ANALYTICAL / CONVERSATIONAL]
MY ACTUAL USE of the product: [BRIEF: have I tried it, what did I think]
RUN LENGTH: [60 SECONDS / 90 SECONDS]

Draft 2 sponsor read variants:

1. THE STORY READ: tied to my personal experience with the product (only if I actually have one).
2. THE PROBLEM-SOLUTION READ: opens with a problem my audience has, transitions to the offer.

For each:
- Time-correct for the run length.
- Match my show's voice.
- Include the offer and code at the right moment (not buried, not jammed first).
- Avoid "Today's episode is brought to you by" if I have other patterns I prefer.

Flag if my use of the product does not match the script. Listeners detect sponsored authenticity gaps.

When to use: Before recording the episode that runs the sponsor. · Best model: Claude. Voice and authenticity discipline matters.

Prompt 7

Engagement Pattern Audit

Most podcasters look at downloads and stop. This prompt audits the patterns that actually predict show health.

Engagement data (past 12 episodes):

FOR EACH EPISODE: title, length, format (solo / guest / interview), downloads, completion rate, share count if available, review/rating signal.

MY SHOW GOAL: [GROWTH / DEPTH / MONETIZATION / AUTHORITY]
MY POSTING CADENCE: [WEEKLY / etc.]

Produce an audit:

1. THE TOP 3 EPISODES and what they have in common.
2. THE BOTTOM 3 and why they likely underperformed.
3. THE FORMAT PATTERN: do solo, guest, or interview episodes perform better.
4. THE LENGTH PATTERN: does length correlate with completion.
5. THE TITLE PATTERN: which title style opened best.
6. THE NEXT 4 EPISODES I should plan to test what is working.
7. THE METRIC I should add to track that I am not currently tracking.

Do not let me confirm what I want to believe. The data is the data.

When to use: Monthly, on every active show. · Best model: Claude or Grok. Both push back on confirmation bias.

These work across Claude, ChatGPT, Gemini, and Grok. Claude is the strongest default for voice work. Grok is sharpest for the engagement audit. For transcription and clip-finding specifically, dedicated tools (Descript, Riverside, Castmagic) are better than general AI.

What is the worst thing you can do with AI for podcasts?

Three patterns will burn podcasters fastest.

  • Letting AI dump the transcript as show notes. Listeners scroll past; the show notes do no SEO work. Use AI for structure; you write the show notes.
  • Using AI to write sponsor reads that pretend authentic use. Listeners detect fake-authentic sponsor reads. Either use the product or pivot to problem-solution framing.
  • Trusting AI to write guest pitches with flattery. Flattery pitches get ignored. The Guest Pitch Letter prompt is built to fight flattery; never paste an AI-written pitch unedited.

What if you want to take this further?

Each prompt above takes inputs you paste in. The next move is connecting AI to your podcast platform.

Connectors are now standard

Claude, ChatGPT, and Grok all support connectors that let your AI read live data from your work tools (Gmail, Notion, GitHub, Asana, HubSpot, Stripe, and many more) instead of relying on you to paste context. For podcasters this means the AI can read your Descript transcripts, your Riverside recordings, your Spotify or Apple analytics, your Notion content calendar, or your Gmail thread with guests.

For podcasters, the connectors worth pairing with these prompts:

  • Descript connector — reads transcripts for show notes and clip extraction.
  • Riverside connector — if your recording lives in Riverside, AI references session metadata.
  • Notion / Google Drive connector — reads your content calendar and prior episode notes.
  • Gmail connector — references guest correspondence for pitch and prep prompts.
  • Buzzsprout / Transistor / Captivate — if your hosting platform exposes data, reads analytics for the engagement audit.

What are common questions about AI for podcasts?

Should I use AI to transcribe my podcast?

Yes via dedicated tools (Descript, Otter, Riverside). General-purpose AI is not optimized for audio transcription at scale; use the dedicated tools for transcription, then feed the transcript into the prompts above.

Will AI replace podcast hosts?

No. AI cannot read the moment, the guest’s body language, the unexpected turn in a real conversation. AI is changing back-office work; use it there and free time for the actual interviews.

Which AI tool is best for podcasters?

Claude Pro for the writing-heavy prompts (show notes, pitches, prep briefs). Descript for transcription and editing. Castmagic or Cast Magic for batch show-notes work specifically for podcasters.

Should I tell guests I use AI?

If AI is used in the back-office (notes, transcription, scheduling), most guests do not require disclosure. If AI is used in the editing of their voice (audio cleanup, gap-removal), increasing transparency is the trend. Check your specific situation.

Is guest content safe in AI tools?

Paid plans with no-training and no-retention terms are appropriate. For sensitive guest content (off-record portions, executive coaching session content), de-identify before any data flows.

Can AI predict which episodes will perform?

AI can find patterns in your past data. AI cannot predict which guest will resonate with your audience the week of release. Use AI for pattern surfacing; you bring the audience read.

How long does it take to build the podcast-AI loop?

Three weeks. Start with the Show Notes Drafter and the Title Generator. Add the Pre-Interview Prep Brief on your next guest. Most podcasters settle into 4-5 of the seven prompts within a month.

🎯

The AI Prompt Library · $39

Podcast workflows, prompt-paved.

Soon to be 1000+ prompts in Notion organized by use case. The full podcast section includes everything above plus prompts for season planning, mid-roll ad placement strategy, listener Q&A integration, podcast network application, and live-event podcast prep. Plus prompts for every other field. Lifetime access.

Get the Library →

Get Smarter About AI Every Morning

Free daily newsletter. Built for people who want to use AI well, not chase every model.

Free forever. Unsubscribe anytime.

Sources to read next?

You might also like

Two ways to go further

The AI Prompt Library

1,000+ ready-to-use prompts for Claude, ChatGPT, and Gemini. Stop staring at a blank box.

Get it for $39 →

2-Hour Live AI Crash Course

A private, beginner-friendly session across Claude, ChatGPT, Gemini, and the wider landscape.

Book for $125 →

Discover more from Beginners in AI

Subscribe now to keep reading and get access to the full archive.

Continue reading