Podcast show notes, chapters and social clips generation

Introduction

Creating podcast show notes is a necessary evil. Your episodes contain valuable information scattered across 30, 60, or 120 minutes of audio, but most listeners will never find that content unless you package it neatly: timestamps for key moments, a coherent summary, maybe even chapters for better navigation in podcast apps, and social media clips to drive traffic back to the full episode.

The traditional workflow is tedious. Someone listens to the episode (or worse, you do), takes notes manually, then stitches together timestamps, writes summaries, creates clips, and posts them across platforms. This takes hours per episode and scales poorly when you're producing multiple shows or publishing weekly.

What if you could automate this entire process? Three tools, combined properly, can handle the heavy lifting: transcript generation with timestamps, intelligent summarisation, and clip extraction. Wire them together correctly, and you go from raw audio file to polished show notes, chapters, and social content without touching a keyboard between upload and publication.

The Automated Workflow

This workflow captures your podcast audio, transcribes it with timing data, summarises key sections, extracts quotable moments, and generates both long-form show notes and short-form social clips. The entire process runs in your chosen orchestration tool with no manual intervention.

Tools Overview

Meetgeek records and transcribes meetings and podcasts, providing structured transcripts with timestamps. Resoomer AI summarises long documents and texts, distilling content into key takeaways. Clipwing extracts and edits video or audio clips from longer recordings, automatically generating social-ready formats. Together, they cover capture, analysis, and distribution.

Choosing Your Orchestration Tool

For this workflow, I'd recommend starting with Zapier if you want the simplest setup with minimal configuration. Use Make (Integromat) if you need more control over data transformation between steps. Use n8n if you're self-hosting and want complete ownership of the workflow. Claude Code isn't an orchestration tool per se, but you can use it to handle complex data transformations or generate formatted content between tool calls.

For this guide, I'll show you the Zapier approach first, then note where n8n diverges significantly.

Step 1: Capture and Transcribe with Meetgeek

Start by uploading your podcast episode to Meetgeek or recording it directly within their platform. Meetgeek generates a full transcript with speaker identification and timestamps. The API provides structured output that includes individual transcript blocks with timing data.

When you call the Meetgeek API, you'll receive something like this:

Copy
{
  "meeting_id": "mtg_12345",
  "title": "Episode 47: AI in Production",
  "duration_seconds": 3600,
  "transcript": [
    {
      "speaker": "Host",
      "text": "Today we're discussing AI implementation in manufacturing.",
      "timestamp_start": 0,
      "timestamp_end": 15
    },
    {
      "speaker": "Guest",
      "text": "The main challenge is data quality and integration.",
      "timestamp_start": 16,
      "timestamp_end": 35
    }
  ],
  "audio_url": "https://meetgeek.s3.amazonaws.com/mtg_12345.mp3"
}

In Zapier, this is your trigger. Set up a Zap that activates whenever a new meeting is completed in Meetgeek. Meetgeek's Zapier integration provides the full transcript and audio URL automatically.

Step 2: Summarise Content with Resoomer AI

Once you have the transcript, you need to distil it into coherent show notes. Resoomer AI accepts long-form text and returns structured summaries. You'll format the Meetgeek transcript output as plain text and send it to Resoomer.

Configure Resoomer to summarise at different compression levels. For show notes, aim for 40-50% compression. For social media teasers, use 20-25% compression to create punchy bullet points.

The Resoomer API endpoint for text summarisation:

Copy
POST https://api.resoomer.com/summarize
Content-Type: application/json

{
  "text": "{{ transcript_text }}",
  "type": "abstract",
  "compression": 50,
  "language": "en"
}

In Zapier, use a formatter step to convert the Meetgeek transcript array into continuous text:

Copy
{{ transcript[].text }}

Then pass this to Resoomer. The response includes the summary text and key points:

Copy
{
  "summary": "The episode discusses AI implementation challenges in manufacturing. Key topics include data quality issues, integration complexities, and ROI measurement. Speakers emphasise the importance of starting with clear use cases and building internal expertise before scaling.",
  "key_points": [
    "Data quality is the primary barrier to AI adoption",
    "Integration with existing systems requires careful planning",
    "Teams need both technical and domain expertise"
  ],
  "original_length": 2450,
  "summary_length": 320
}

Step 3: Extract Clips and Generate Social Content with Clipwing

Now you need to identify the best moments in the audio and create clips. This is where Clipwing excels. You can either specify timestamps manually or use Clipwing's detection to identify speaker changes, pauses, and topic shifts.

For a more automated approach, analyse the Meetgeek transcript to identify strong quotes or key discussion points. Look for moments where speakers introduce new ideas or provide especially quotable statements. Extract those timestamp ranges and send them to Clipwing.

The Clipwing API for clip generation:

Copy
POST https://api.clipwing.com/clips/create
Authorization: Bearer YOUR_CLIPWING_API_KEY
Content-Type: application/json

{
  "source_url": "{{ audio_url }}",
  "clips": [
    {
      "start_time": 240,
      "end_time": 420,
      "title": "Data Quality Challenges",
      "format": "square"
    },
    {
      "start_time": 900,
      "end_time": 1080,
      "title": "Integration Strategy",
      "format": "vertical"
    }
  ],
  "output_format": "mp3",
  "include_captions": true
}

Clipwing returns processed clips with captions and timestamps:

Copy
{
  "clips": [
    {
      "clip_id": "clip_001",
      "url": "https://clipwing.s3.amazonaws.com/clip_001.mp3",
      "duration": 3,
      "captions": "Data quality is the primary barrier to AI adoption"
    }
  ],
  "status": "complete"
}

Complete Zapier Workflow Structure

Here's how the full workflow flows in Zapier:

1. Trigger: Meetgeek meeting completed → webhook to Zapier.

2. Step 1: Format transcript text using Zapier formatter.

3. Step 2: Send formatted text to Resoomer AI summarisation endpoint.

4. Step 3: Extract timestamps from original transcript for key moments (use a Zapier Code action to parse the transcript and identify speakers with quoted content).

5. Step 4: Call Clipwing API with extracted timestamp ranges.

6. Step 5: Combine summarised content, transcripts, and clip URLs into a structured document.

7. Step 6: Create a Google Doc or Notion page with show notes, chapters, timestamps, and links to social clips.

8. Step 7: Post social clips and summaries to Twitter, LinkedIn, and your podcast platform using native integrations.

A basic Zapier Code step to extract key timestamps might look like this:

Copyconst transcript = inputData.transcript;
const keyMoments = [];

for (let i = 0; i < transcript.length; i++) {
  const block = transcript[i];
  
  // Identify moments with quotation marks or emphatic language
  if (block.text.includes('"') || block.text.includes('!')) {
    keyMoments.push({
      start: block.timestamp_start,
      end: block.timestamp_end,
      text: block.text,
      speaker: block.speaker
    });
  }
}

return [{ keyMoments }];

In n8n, this same workflow is more explicit. You'd use dedicated HTTP nodes for each API call, with explicit JSON mapping between each step. The advantage is transparency; the disadvantage is more configuration upfront. Make (Integromat) sits somewhere in the middle with a visual flow editor.

Handling Output and Distribution

The final step is packaging everything. Create a Zapier action that generates a formatted document. Use Google Docs API to create a new document with the following structure:

Copy
## Summary
[summary from Resoomer]

## Key Points
- [key_point_1]
- [key_point_2]
- [key_point_3]

## Timestamps
00:00 - Introduction
04:00 - [topic 1]
15:30 - [topic 2]

## Social Clips
[List of clip URLs from Clipwing]

## Full Transcript
[Full transcript with speaker names and timestamps]

Use a final Zapier action to post to your podcast platform's API or to social media platforms via native Zapier integrations.

The Manual Alternative

If you want more control over which moments become clips or how content is summarised, you can pause the automation at key steps and inject human judgement.

After Meetgeek generates the transcript, pause the workflow and ask a team member to review and flag the best moments for clipping. Add these manually to a spreadsheet or form, then resume the automated workflow using those hand-selected timestamps.

Alternatively, summarise manually using Resoomer as a writing aid. Paste the transcript into Resoomer, review the output, edit it for your audience and tone, then feed the edited summary into the rest of the workflow.

This hybrid approach is worth considering if your podcast has a distinct voice or if certain episodes require editorial nuance. The automation saves time on grunt work whilst preserving quality control where it matters most.

Pro Tips

1. Manage API Rate Limits Proactively

Meetgeek and Resoomer have rate limits on their free tiers. If you're publishing multiple episodes per week, upgrade to their paid plans or implement backoff logic in your orchestration tool. In Zapier, add delays between steps using the "Delay" action. In n8n, use the "Wait" node. Resoomer's API typically allows 10 requests per minute on the starter plan; if you exceed that, requests fail and your workflow stalls.

2. Test with Short Clips First

Before automating your entire back catalogue, test with a single short episode, 15-20 minutes maximum. Verify that timestamps are accurate, summaries are coherent, and clips export in the format you need. Meetgeek's timestamp accuracy can drift if there's background noise, and Resoomer sometimes struggles with highly technical jargon.

3. Use Clipwing's Caption Sync

Clipwing can auto-generate captions for clips, but review them for accuracy. If your podcast has guest speakers with strong accents, captions may be garbled, which looks unprofessional on social media. Set Clipwing to include captions, then use a Zapier step to send clips to a team member for a quick 30-second review before publishing.

4. Build in Error Notifications

Workflows fail silently sometimes. Add a Gmail or Slack action at the end of your Zapier Zap so you receive a notification when the workflow completes successfully. Better yet, send notifications only on failure. In Zapier, use conditional logic ("Only continue if...") to branch: on success, send a success notification; on error, send a detailed error report.

5. Archive Raw Files for Reuse

Store the original audio file, full transcript, and API responses in a cloud folder (Google Drive, S3) for each episode. This means you can regenerate clips or summaries later without re-processing the audio. Set up Zapier or n8n to automatically save JSON responses from Resoomer and Clipwing to a specific folder.

6. Optimise Clip Length for Each Platform

Different platforms have different optimal clip lengths. TikTok and Instagram Reels perform best at 15-30 seconds. LinkedIn performs better with 1-2 minute clips. Twitter is fine with 15-60 seconds. Configure Clipwing to generate multiple versions of each clip at different durations, then route them to different social platforms using Zapier's conditional logic.

Cost Breakdown

If you're publishing only one episode per month, you can reduce costs significantly by using free or starter tiers: Meetgeek Free ($0), Resoomer Free ($0, with limited summaries), Clipwing Free (3 clips/month), Zapier Free ($0, with 100 tasks/month). Your monthly cost drops to $0, though you'll hit API limits quickly.

For teams publishing weekly, the Professional tiers above are essential to avoid hitting rate limits. Consider this your baseline investment in removing editorial bottlenecks.