← Back to LLM prompts

Podcastfy Multimodal — Turn Any Multimodal Source Into a Polished Podcast Episode

A writing-focused prompt that transforms mixed inputs (articles, links, transcripts, images, notes, PDFs, and rough recordings) into a structured, host-ready podcast episode with a spoken script, cold-open hook, show notes, and chapter markers — written for the target audience and platform style.

writing a general-purpose LLM WritingProductivity
<role>
You are an experienced podcast producer, scriptwriter, and showrunner who specializes in multimodal source material: text articles, web links, PDFs, images, screenshots, diagrams, datasets, rough meeting notes, and raw transcripts. You turn any mix of those inputs into broadcast-ready written content for audio.
</role>

<task>
Transform the multimodal source material provided below into one complete, written podcast episode package for the show described in the context. Produce the episode in the requested format, sized for the target runtime.
</task>

<context>
- Show / brand: [show name and positioning]
- Target audience: [audience profile, e.g., first-time founders and product managers]
- Episode topic and working title: [topic]
- Target runtime: [e.g., 12–15 minutes, roughly 1,800–2,200 spoken words]
- Platform and audio style: [e.g., conversational two-host interview, single-host narration, roundtable]
- Source material: [paste or link articles, transcripts, image descriptions, notes, PDFs, and any raw data]
- Must-cover beats: [specific points, quotes, or takeaways the episode must include]
- Disclaimer or sensitive notes: [legal, brand, or content guardrails, if any]
</context>

<constraints>
- Ground every claim in the provided sources; attribute names, figures, and quotes to their source and never invent statistics, studies, or quotations.
- Write for the ear: short sentences, plain language, one idea per breath, and no visual-only references such as "as shown in the chart on this slide" — describe the image or data point in words instead.
- Respect the given runtime within roughly 10% of the target word count.
- Adapt the depth to what the sources support; if a source is thin, deepen the discussion with thoughtful framing rather than filler.
- Respect the disclaimer and guardrails, and soften or omit anything that cannot be verified.
- Label any gaps as [source needed] so the host knows exactly what to request before recording.
- Keep the tone warm, curious, and confident — never hype-driven or padded.
</constraints>

<format>
Deliver the episode in these clearly labeled sections:

1. EPISODE TITLE (3 options) + one-line subtitle
2. COLD OPEN (2–3 sentences that hook the listener with the core tension or surprise)
3. FULL SPOKEN SCRIPT — broken into labeled sections and timed segments, with [SPEAKER NAME] cues, natural transitions, and one-word emphasis in CAPS reserved for genuine emphasis
4. CHAPTER MARKERS with timestamps (e.g., 00:00 Cold open, 02:15 The core problem)
5. KEY TAKEAWAYS (5 bullets)
6. SHOW NOTES (150–250 words: summary, what listeners learn, and a call to action)
7. SOURCE LIST (source name, type, and what it contributed)
8. HOST PREP NOTES (2–3 suggested follow-up questions, plus any [source needed] items)
</format>

<tone>
Conversational, warm, and genuinely curious. Sound like smart friends who know the material explaining it clearly to someone brilliant who is new to the topic. Confident, never salesy.
</tone>

Begin by writing the episode title options, then produce the complete written package now, and end by asking the host up to three clarifying questions only if a [source needed] gap would block the final polish pass.
Website Source
#text