Mel Varghese is a licensed psychologist with a PhD and the founder of the Selling the Couch podcast, which has passed 1.8 million downloads and 300 episodes. He could not get AI to produce content that sounded like him. Working with Dan, he went from rejecting the output outright to a content workflow built around his own voice.
Mel was sitting on nearly ten years of podcast episodes with no efficient way to repurpose them into social media content. He tried integrating AI on his own, and it did not go well. "I had been trying to figure out how to integrate AI into our podcasting workflow for just a couple of weeks and quickly felt overwhelmed," Mel said. Worse, the output sounded generic. "Sometimes with AI, it sounds like way too AI. I'm like, No, I ain't doing that."
"I couldn't quite figure out how to use AI in a way that accentuates what I'm doing versus strictly just fully replacing," he said.
Mel had known Dan for years and had noticed him posting about AI on LinkedIn. When Dan approached him about working together, Mel said yes. "Okay, I need help," he remembers thinking. "I trust Dan, and I know that he's going to take really good care."
The Approach
Dan wrote brand voice training documents from Mel's own material and built a guided content workflow around them, as part of an AI implementation engagement, so the AI could work from Mel's voice.
Dan started with a list of what Mel already had. They put transcripts of Mel's solo episodes into a shared Google Drive, and Mel added notes on his brand voice and what mattered to him, so the AI had Mel's own material to learn from.
Next, Dan wrote the brand voice training documents, which spell out Mel's tone, vocabulary, storytelling patterns, what to use and what to avoid, and how to keep his conversational style. "This is pages and pages of stuff that he was able to generate for me based on my brand voice," Mel said.
Each episode now moves through five steps.
| Step | What Happens |
|---|---|
| 1. Capture | Mel records the episode as usual |
| 2. Transcribe | The transcript goes into the shared Google Drive |
| 3. Process | The AI works through the transcript using the voice documents |
| 4. Produce | The AI drafts social media posts in Mel's voice |
| 5. Review | Mel reviews and publishes, and keeps final editorial control |
The Results
"This has made my workflow so much easier," Mel said. "The ROI is so worth it. It'll blow your mind."
Podcast content now moves into social media posts through a repeatable, AI-assisted process, so nobody starts from a blank page for each platform. Every draft starts from Mel's voice documents.
What This Looks Like for Your Content
If you are sitting on hours of episodes, webinars, or client conversations and the AI keeps producing output that does not sound like you, you can book an AI strategy call and bring one episode.
