WHAT IS IT?

Podcast Growth Engine is a complete content repurposing system that converts podcast episodes into 20+ pieces of platform-optimized content, automatically scored by viral potential, with A/B test variants generated for every headline.

You get two editions:

1. CLI EDITION (for developers and automation) A single Python script that processes audio files or transcripts and outputs JSON with content pieces, viral scores, weekly calendars, and A/B test variants. Full control. Easy to customize. Built for integration into your own products.
2. DESKTOP EDITION (for creators using Claude) Five copy-paste prompts that work in any Claude interface. No code. No setup. Just transcript in, 20+ pieces out. Perfect for white-labeling or offering as a service to non-technical clients.

Both produce the same output: 3 Twitter threads, 2 LinkedIn articles, 2 YouTube Shorts scripts, 2 TikTok scripts, 2 newsletter sections, 2 blog outlines, 5 quote cards, and 3 email subject lines. Every piece scored. Every headline tested.

HOW IT WORKS

THE PROCESS (Five Steps)

• Step 1: Extract Content Atoms. Feed a podcast transcript to Claude. The AI identifies 15-20 high-impact moments: quotes, stories, frameworks, controversial takes, data points, and predictions. Each atom gets tagged with which platforms work best.
• Step 2: Generate Content Pieces. From those atoms, generate 20+ pieces of platform-native content. A Twitter thread isn’t a rehashed LinkedIn post; it’s optimized for Twitter’s format, rhythm, and psychology. Same with every platform.
• Step 3: Score by Viral Potential Every piece gets scored 0-100 using a three-part formula: Novelty (40%) + Controversy (30%) + Utility (30%). This isn’t gut feel. It’s math. Pieces that score 80+ get priority. Those below 40 get cut.
• Step 4: Generate A/B Variants. Every headline, hook, and subject line gets three versions: the original, a curiosity gap version, and a benefit-driven version. You don’t guess which wins. Your clients run the test and find out.
• Step 5: Build a Weekly Calendar. All scored, deduplicated content gets assembled into a realistic weekly schedule. Twitter gets its peak times. LinkedIn gets Tuesday-Thursday. Shorts get evening slots. Calendar respects every platform’s best practices.

Output: Ready-to-publish pieces, viral scores, calendar, A/B variants, and statistics.

Time to first episode: 20 minutes for creators. 5 minutes for developers (after setup).

HOW MUCH TIME THIS SAVES

Manual podcast repurposing: 8-12 hours per episode, Podcast Growth Engine: 20 minutes per episode, ROI: 6-10x time savings

For an agency with 5 clients posting weekly:

• Manual approach: 40-60 hours per week spent on content repurposing
• Growth Engine: 100 minutes per week
• Hours saved: 39-60 per week
• Billable hours freed up: 5-7 hours per client per month

At a $150/hour average billable rate, $1,050-1,400 per client per month is freed up.

This is a tool that pays for itself on the first two clients.

WHAT YOU GET IN THE BOX

CLI EDITION (22 KB)

• repurpose.py (single Python script, 18.5 KB)
• ab_test.py (A/B testing + statistical analysis, 14.3 KB)
• setup.sh (one-command installation)
• requirements.txt (all dependencies listed)
• .env.example (API key template)
• AGENTS.md (system prompts, fully documented)
• mcp_config.json (MCP servers configured and ready)
• README.md (complete technical documentation)

DESKTOP EDITION (11 KB)

• SKILL_DESKTOP.md (5 copy-paste prompts for Claude)
• AGENTS.md (system prompt reference)
• START_HERE.txt (quick start guide)
• mcp_config.json (for future integrations)

Everything is documented. Everything is customizable. Everything is production-ready.

WHO IS THIS FOR?

THIS IS BUILT FOR AI AUTOMATORS AND AGENT BUILDERS

You build products for other people. You understand the value of great tooling. You know that the difference between a mediocre product and a great one isn’t the idea, it’s execution.

Podcast Growth Engine is for you if:

• You want to expand your product line without building from scratch.
• You have clients or customers who need podcast content repurposing, and you want to do it faster than manually.
• You want to white-label a complete solution and resell it under your brand.
• You want to offer podcast content as a service and need the automation behind it.
• You want to integrate podcast processing into your existing AI agent product.
• You want code you can read, understand, and customize instead of a black box.

This is not for podcast creators who just want a tool. (There are cheaper options for them.) This is for people who want to offer this as a service, build on top of it, or integrate it into their product.

WHY IT WORKS

THE ARCHITECTURE THAT MAKES THIS POWERFUL

• Viral Scoring Algorithm Most tools tell you, “Here’s 20 pieces.” This tells you “here’s which 3 will actually go viral and why.” The scoring formula (Novelty × Controversy × Utility) is based on what actually drives engagement. Not opinions. Math. Your clients prioritize correctly. They don’t waste time on filler.
• A/B Testing Framework Built In. Everyone talks about testing. No one implements it. This comes with a complete A/B testing system: create tests, log results, run statistical analysis (Mann-Whitney U test), and get a winner with confidence. Your clients don’t guess. They know what works.
• Platform-Aware Scheduling: The calendar doesn’t just dump all content on the same days. It respects Twitter’s peak times (9 am, 2:30 pm, 5 pm ET). It knows LinkedIn works Tuesday-Thursday mornings. It schedules Shorts for evenings. Every platform gets optimized posting times.
• One-File Architecture (CLI) The entire content generation engine fits in one Python script. You can read it. You can modify it. You can integrate it. You’re not dependent on a SaaS company staying in business. You own the code.
• Prompt-Based (Desktop) The desktop edition uses five careful prompts to guide Claude through the full process. Want to change the tone? Edit the prompt. Want to focus on different platforms? Edit the prompt. Want to customize for a specific audience? Edit the prompt. You’re not locked in.
• Statistically Sound A/B Analysis The A/B testing uses the Mann-Whitney U test (non-parametric), which is the gold standard for small sample sizes. It doesn’t require normal distribution. It requires p < 0.05 AND 15% lift for promotion to playbook. Your clients get real winners, not false positives.
• Built on Proven AI Anthropic Claude for content generation. OpenAI Whisper for transcription. Scipy for statistical analysis. Not custom models. Not unproven frameworks. Industry-standard tools that work.

WHAT MAKES THIS DIFFERENT

• vs. REPURPOSE.IO They generate clips. You generate full content across 8 platforms. No viral scoring. You have it. Higher price ($150-500/month). You’re $99-299. No A/B testing. You have it.
• vs. PODIUM Video-focused only. You handle text and video. More expensive. Better for podcasters who just want clips. Worse for agencies that want complete content.
• vs. DESCRIPT Great for editing. Weak on repurposing. Doesn’t score content by viral potential. Doesn’t generate cross-platform content. Different tools for different jobs.
• WHY THIS BEATS THEM FOR YOUR USE CASE
• You get the full source code. Customizable. White-labelable. Not locked into someone else’s API. You can run this for unlimited clients without per-seat licensing.
• You get viral scoring. Your clients publish the right content first—less waste. Better results.
• You get A/B testing. Your clients can prove what works. That’s how you move from “this tool saved time” to “this tool made me money.”
• You get two editions. CLI for developers. Desktop for non-technical users. One product, two markets.

WHAT YOU CAN DO WITH THIS

YOUR OPTIONS

• OPTION 1: White-Label It Rebrand both editions. Change the colors. Change the copy. Resell as your own product. Your name on it. Your pricing. Your support.
• OPTION 2: Add It to Your Service Offering. You’re an agency. You manage social media for clients. Now you offer “one free week of content repurposing.” Use this as your backend. Save 10 hours per week. That’s your margin.
• OPTION 3: Offer It as a Standalone Service. Create a landing page. Charge $99 per episode (CLI) or $49 per episode (desktop). Take the podcast from the client. Run it through the engine. Send back 20 pieces. Done. Recurring revenue.
• OPTION 4: Build On Top of it. Integrate the CLI into your own product. Your UI. Your branding. Your workflow. The engine is just the backend.
• OPTION 5: Customize for Vertical Content marketing agencies: charge extra for blog outlines. PR agencies: charge extra for quote cards. Influencer management: charge extra for TikTok scripts. One engine. Multiple revenue models.
• OPTION 6: License It to Other Builders. You know other AI product builders. License this to them—recurring revenue from white-label fees.

Most product builders are stuck choosing between “build from scratch” and “trust someone else’s black box.”

This is the third option: code you understand, architecture you can modify, and a product that’s already proven to work.

One episode in. 20 pieces out. In 20 minutes. With your name on it.

That’s what you’re getting.