I was a freelancer, web developer, and programmer since 14 years old, before pivoting to podcasting and automation.
We programmers have a saying when it come the quality of what our work produces: garbage in, garbage out.
That means your output is a function of your input.
When it comes to tools like ChatGPT, the better your prompt, the better the answer you get.
When it comes to using AI to repurpose your content, the clearer your content, the better the assets AI generates.
I’ve been experimenting with Castmagic over the last few weeks, and something has become very clear:
When I speak off the cuff, sometimes I’m not very clear.
I won’t lay too much credit at the feet of generative AI, but when I have it transcribe and subsequently summarize my thoughts, it seems to focus on 1-2 things…presumably the things it understands the best.
Or perhaps, it’s what I have placed at that assumed “climax” of the episode.
Generative AI and Storytelling
I’m not an expert in Large Language Models (LLMs), but the basics are this: it takes in a TON of text-based data, and uses math to associate the discrete words from those texts.
So for example, it might recognize, due to statistical analysis, that “I want to go to” more often ends with the word “Paris, France” than “Cheyenne, Wyoming” (sorry, Wyoming).
All of that is to say, LLMs likely understand how people tell stories, and makes some assumptions about how we organize our podcasts.
I decided to ask ChatGPT (by way of Raycast AI) what makes a good story:
Then I asked it the most important point of the story:
And when the climax usually happens:
And while Castmagic doesn’t always pick what I talk about at the end, it does pick a point after I’ve built some momentum.
The problem is that’s not how I structure my solo episodes. I usually put the most important information up front.
While there is a place for that (like social media), it doesn’t create compelling content like a story does.
We are Storytellers
There’s a clear difference between the popular podcasts and the struggling ones: a story.
Not in the “Hero rescues the people in trouble” sort of sense. But there is still a clear beginning, a conflict, a climax, and a satisfying resolution.
And if you subscribe to The Hero’s Journey, there is still a hero (the listener), and a guide (the host).
For example, on The Profitable Podcaster, I might tell a story about how I had a client (the hero) struggling to stay consistent.
In the beginning, she starts the podcast, diligently producing an episode each week. But then she falls behind after having to unexpectedly take a week off (here’s the conflict).
So she hires me (the guide) to help her, and we set up a few automations that save her 2 hours per week…there’s the climax. NOW, she can spend those 2 hours creating a back catalog of episodes, staying 1-2 months ahead…a satisfying resolution.
This is much more compelling than me talking at you about how you need to automate parts of your podcast, right?
And doing a little bit of planning from the beginning can help us create better episodes, which then creates better content for repurposing on other platforms.
How Can AI Force Podcasters to be Better Storytellers?
So knowing all of this begs the question: what can we do with this information?
If we want to leverage generative AI for repurposing especially, we need to give it better inputs. Garbage in, garbage out.
That means doing a little bit of planning up front. I think there are two ways you can do it:
- Plan the show on your own, outlining, creating pseudo scripts, and building a narrative around each episode. We need to lean into the story.
- Use AI to help you plan and organize your episode, giving it a topic and crafting a prompt to help you build a better episode.
Here’s an example of the latter:
You can also feed it an outline and ask to improve it:
…or a narrative:
Leveraging AI isn’t just about creating content we normally wouldn’t create. It’s about helping us create better content.
That means using it for both input (researching, organizing, and crafting stories), and output (taking what we’ve created and repurposing it for other platforms).
When you create better inputs, your output will be SO much better.