Tom Canfarotta

Posted on Jun 9 • Edited on Jun 10

🎙️ MailCast.ai: Email-to-Podcast-RSS-Feed

#webdev #devchallenge #postmarkchallenge #api

A submission for the Postmark Challenge: Inbox Innovators

💡 What I Built

Hey dev community! 👋

I built an Email-to-Podcast RSS Service that automatically converts emails into podcast episodes! Simply forward any email to the service, and within minutes you'll get a professionally produced podcast episode delivered to your personal RSS feed 🎧.

Here's the magic workflow:

User forwards email ➝ Postmark receives it ➝ Webhook extracts content ➝
AI fetches full articles from links ➝ OpenAI writes podcast script ➝
ElevenLabs generates natural speech ➝ Episode added to RSS feed ➝
User gets email with podcast links 🎉

🎥 Live Demo

📧 Try it yourself:
Forward any email (newsletter, article, etc.) to: [email protected]

Within 60-90 seconds, you'll receive:

Direct MP3 download link
Your personal RSS feed URL
Subscribe in any podcast app!

💻 Code Repository

The project is open-source and available on GitHub:
🔗 https://github.com/tcanfarotta/email-to-podcast-rss

Key components:

src/routes/webhook.js: Postmark webhook handler
src/services/emailProcessor.js: AI-powered content processing
src/routes/rss.js: Dynamic RSS feed generation

Core Webhook Handler

// src/routes/webhook.js
webhookRouter.post('/postmark', async (req, res) => {
  const email = {
    from: emailData.From,
    subject: emailData.Subject,
    textBody: emailData.TextBody,
    htmlBody: emailData.HtmlBody
  };

  // Process asynchronously - don't block Postmark
  processEmail(email).catch(error => {
    console.error('Error processing email:', error);
  });

  res.status(200).json({ message: 'Email received' });
});

AI-Powered Processing

// src/services/emailProcessor.js
async function processEmail(email) {
  // Extract content, including from "Continue Reading" links
  const content = await prepareEmailContent(email);

  // Generate podcast script with OpenAI
  const aiResponse = await generatePodcastDialogue(content, email.subject);

  // Convert to speech with ElevenLabs
  await generateAudio(podcastScript, outputFile);

  // Add to RSS feed
  await addPodcastEpisode(episode);

  // Send email with podcast links
  await sendEmailReply(email.from, episodeTitle, rssUrl);
}

🛠️ How I Built It

This project was an exciting journey combining email automation, AI, and podcast technology!

🧠 The AI Stack

I needed fast, reliable AI services for two tasks:

Script Generation:

✅ OpenAI GPT-4o - Creates natural, conversational podcast scripts
Transforms dry email content into engaging monologues
Generates catchy episode titles automatically

Text-to-Speech:

✅ ElevenLabs Turbo v2.5 - Natural, expressive voices
Handles emotions and pacing beautifully
Fast generation (usually < 30 seconds)

🔗 Smart Content Extraction

The coolest feature? Automatic article fetching!

Many newsletters just have teasers with "Continue Reading" links. My service:

Detects these links intelligently
Fetches the full article content
Converts it into the podcast script

// Intelligent link detection
const readMoreRegex = /(continue\s+reading|read\s+more|view\s+article)[\s\S]{0,200}(https?:\/\/[^\s]+)/gi;

// Fetch and extract main content
const webContent = await WebFetch(link, 'Extract the main article content');

📻 Personal RSS Feeds

Each email sender gets their own unique RSS feed:

https://yoursite.com/rss/feed/{unique-id}
Contains only their converted podcasts
Compatible with all major podcast apps

🚀 Production Setup

Node.js + Express:

Modular architecture with separate routers
Async processing to handle webhooks quickly
File-based storage for simplicity

Deployment considerations:

Static file serving for MP3s
Proper CORS headers for RSS feeds
Error handling with retries for AI services

💸 Cost Optimization

To keep costs manageable:

Implemented caching for repeated content
Limited script length to ~1500 characters
Added retry logic with exponential backoff
File size limits on stored podcasts

🔧 Challenges & Solutions

Challenge 1: Email Loop Prevention
Initially worried about email loops when sending replies!

Solution:

// Only process emails TO the service, not FROM it
if (email.from === process.env.POSTMARK_FROM_EMAIL) {
  return; // Skip processing
}

Challenge 2: Content Extraction
HTML emails are messy - scripts, ads, footers everywhere.

Solution:
Built a robust HTML-to-text converter that:

Preserves link context
Removes signatures and forwards
Cleans up email artifacts

Challenge 3: Rate Limits
Both OpenAI and ElevenLabs have rate limits.

Solution:

// Retry with exponential backoff
for (let attempt = 1; attempt <= maxRetries; attempt++) {
  try {
    // Make API call
  } catch (error) {
    if (error.response?.status === 429) {
      await new Promise(resolve => 
        setTimeout(resolve, attempt * 2000)
      );
      continue;
    }
  }
}

🎯 Use Cases

This tool is perfect for:

📰 Converting newsletters to podcasts for commutes
📚 Making long articles accessible while exercising
🎧 Building a personal audio library from emails
♿ Accessibility for vision-impaired users

🙌 Final Thoughts

This project combines the best of email automation and AI to solve a real problem - we get SO much great content via email but not enough time to read it all!

Big thanks to Postmark for their reliable email infrastructure and Dev.to for organizing this challenge!

The magic happens when you combine: