How to create a podcast with 2 AI tools
Combine Claude + ElevenLabs to create personalized AI podcasts.
Many of you probably noticed that I got excited when Google released NotebookLM's podcast feature. My inbox exploded with questions about it. Everyone started creating basic podcast summaries from their content - but something felt off.
I spent weeks testing NotebookLM, diving deep into its capabilities, discovering what worked and what didn't. Here's the thing: most people are using these AI podcast tools like simple text-to-speech converters. Upload content, get a robotic summary, done. I’m not saying NotebookLM isn’t a good tool. It is an amazing research tool, just not for podcasts.
I even wrote a ridiculously popular guide on it here.
But the real podcasting breakthrough came when I started experimenting with combining different AI tools in ways they weren't necessarily designed for. I wanted something more natural, more personalized, something that would actually engage listeners or at least bring me closer to that goal.
Let me show you exactly what I mean.
If you've tried NotebookLM's podcast feature, you know the issue - you're stuck with two identical AI hosts having what feels like the most corporate conversation ever. They sound painfully unrelatable, often padding conversations with filler that adds zero value.
We need a completely different approach if AI podcasts are actually going to be good and provide value. Instead of accepting the limitations of any single tool, I started experimenting with combining Claude's natural language capabilities with ElevenLabs' voice generation.
The goal wasn't just to create another AI podcast - it was to develop a system that could transform written content into engaging audio while maintaining your unique voice and style.
What you're about to learn is a framework for creating AI-powered podcasts that sound (more) natural, engage your audience, and scale your content in ways that automated tools simply can't match.