Podcast Workflows - Tips for podcasters to grow their podcast without wasting time.

Real quick before we get started. I want to tell you about a brand new newsletter I'm launching called Podcast Workflows. Each week I'll do a deep dive on how a popular podcaster creates their show. From start to finish, from processes to tools. Then, I'll tell you how you can apply what they do to improve your own podcast workflow.

You can sign up over at [podcastworkflows.com]. One breakdown delivered to your inbox weekly. Learn the tools and processes used by top podcasters and reclaim hours of your life, every week. That's [podcastworkflows.com]. Sign up today, for FREE.

Generative AI is all the rage these days, and it seems like there's a new tool every five minutes. They range from summarization to email composition to full video and voice fabrication. But how can podcasters benefit from AI tools? Research, scheduling, script writing. What about full on voice synthesis and replacement?

Would you believe this isn't my real voice? Maybe you did believe that isn't my real voice. I, for one, don't think it sounds very much like me, but when I did this on my live stream recently, a lot of people said it sounded just like me. I sent a voice message to my wife and she didn't question its authenticity. It's only when I asked my brother point blank, sending him the audio, does this sound like me that he said it did not.

So what can AI tools do for podcasters? Can they replace us? I don't think so. I don't think that there's enough emotion, shockingly in the Robot Joe version of this cold intro, but there are lots of ways that generative AI can help podcasters. And I'm going to dive into some of those use cases and tools today, here on the Profitable Podcaster.

Welcome to the Profitable Podcaster where I help coaches, course creators and authors launch and grow their podcasts to help them build authority, generate more leads, and sell more through actionable advice and expert tested systems.

I'm your host, Joe Casabona and I've been podcasting for over 10 years. I've generated millions of downloads and hundreds of thousands of dollars from my shows. I'll teach you everything I know here on The Profitable Podcaster.

Okay. So this is going to be a typical what, why, how kind of format. I'm gonna tell you what you can use generative AI for, why you should or could use it for that, and then how, as in the tools. So what you can do, and then the tools you can use.

I'll say right off the bat, I think I said this, I said this in the cold open… I used elevenLabs for the synthesized voice. I personally don't think it's there yet. I have seen people using it to create videos of themselves from like a couple of different tools, and they're like, “This can, this can make it so I can make videos without actually having to turn on a camera.” And I'm like, “Yeah. If you wanna create crappy videos, right? So, if you go back and listen to the cold open again, you'll probably hear a cold emotionless reading that is a close facsimile to my voice, but not that close. And definitely not as animated as my voice is. So, that's what I think I would love to hear your feedback on back on Twitter. So you can tweet at me @jcasabona. I'm also on Mastodon and Blue Sky for some reason.

Anyway, let's talk about what you can use generative AI for? There are a couple of obvious ones: I think, research and transcripts are probably the biggest ones, right? You can open up a chat with ChatGPT, asking a bunch of questions.

That could be your jump off point. So for example, I used ChatGPT to do some research on a podcast for a client where I didn't know much about the area of expertise. I am sure I've said this before and I used ChatGPT to learn a little bit about that area of expertise. And then subsequently come up with ideas for the podcast.

Transcripts is another one. I am actively using that right now with Descript. A transcript is being generated as I speak.

There's also otter.ai.

Riverside offers that now. There are lots of transcription tools out there, and they are getting better by the day.

Summarization and social posts are two that I'll put together, right? And this is generally where you would feed in a transcript. And then have it come up with a summary and possible social posts. Your mileage may vary on this, but I think that this is a really powerful tool, atleast as a starting point, right? Let's say you have recorded an interview with somebody. You don't quite remember the main points you talked about. Maybe you didn't take very good notes that day. Maybe you don't take notes at all. You can feed the transcript into one of these tools that I'll talk about in a minute, and then you can get a potential summary. You can get potential social posts.

I have seen pretty good summarization in craft, which is my note taking app. I have not been able to get a good summary from ChatGPT because the transcripts are usually much longer than the allowed, they call it a token size, basically the allowed input. So again, your mileage may vary on that. I'll talk about a couple of tools later. That you can use for those things specifically.

you can also use it for artwork, right? So there are a few tools that will generate art and logos and things like that. There are also AI assist, like assistance that can help you again, this might be good for idea generation. I am not in a position, I'm not a graphic designer by any way, shape, or form. I'm not a prompt master. So like maybe with the perfect prompt, you can come up with artwork for your podcast. But by and large, I think again, this would be probably a good starting point or if you get your tweaks, your prompts, and your tweets just right, like maybe you can tweak it to make it good.

Proofreading is another place where you can use AI tools, right? So again, if you have your summary, you come up with a summary. Generally what I will do because what my experience in my experience, having these tools summarize my podcasts, they always say the speaker or the podcaster, they always make it in the third person. And so I will usually take that output and then feed it into ChatGPT I use.

I'm pretty sure, so I use Raycast.ai. Raycast is a launcher for the Mac. And they, as I record this recently rolled out Raycast.ai, which builds an AI chat into their tool. I am pretty sure.

Yeah. It uses chat GPT so you can basically have ChatGPT on the desktop. I think it uses GPT-4 because I'm paying for it, like I'm paying for Raycast.ai. But they don't actually powered by GPT and Turbo modes. Early access will be included in a separate plan. So I'll probably be upgrading to GPT-4 inside of Raycast, assuming it's not the same $20 a month as ChatGPT-4 itself. Either way, that's neither here and nor there. What I'll do is I'll feed in the summary or the social posts and I'll say something like, you know, reword this to being the first person, reword this to make it a friendlier conversational tone and stuff like that. And I suspect, the more I use these tools, the more they'll learn how I work.

So, research, transcripts, summarization, social post, artwork, proofreading. Those are the ones that I think a lot of people on like first blush, right? Research, transcript, summarization, like those things were the first ones to come to mind for me.

Another one is who to interview. So you can ask one of these tools who within this area of expertise can I interview for my podcast? You need to be careful with that one though, because at least the last time I tried it, about half of the list that both ChatGPT and Google Bard gave me were people who are dead and therefore cannot be interviewed. And that's even if it knows who's living and who's dead, the data from at least ChatGPT is from I think, 2021. It's never current to present day or even within a month of present day. So anybody who died between the update and the present day could still be included, right? So that's something to look out for. But again, it'll give you some potential ideas, right? And this is not something I've tried in a while. Maybe the updated models are better. Maybe Google is actively indexing more often. Though from what I've seen, ChatGPT is still better though. I think it really depends on who you ask and what you're using it for.

Bing, which is not one that I've really considered, apparently sites its sources and is doing live indexing and stuff like that, which is why you have a limited number of queries.

So those are the things that you think about, right? Maybe finding contact information could work. Maybe finding podcasts to swap with, right? You describe your podcast, ask it for similar podcasts, but again, podcasts are subject to Podfeed a lot. So if a database has not been updated in a year, a lot of podcasts could be gone by then.

Now, the one or two things I didn't think of that I was just testing and worked pretty well were production and promotion schedule. So the prompt could read like this: I have 12 episodes that I want to publish between now and August. I want to publish every other week. What does my production schedule look like? And then it'll give you a list, right, of fortnightly dates or every other week dates. Then you could say, I want to create a promotion schedule where I promote each episode 1, 14 and 30 days after the episode airs using the production schedule you just built. Give me a promotion schedule for each episode which is pretty cool, right? And then I asked it to make a table so that I can easily visualize the schedule. Then I asked it to find any overlapping dates, right? Because if I'm publishing every Monday, then invariably I'm going to publish episode one and then promote episode one on the same day I am publishing episode two, if it's a fortnightly podcast, right? So you can visualize how that promotion and production schedule are going to work, which I think is pretty cool.

So those are the things that you can do. Research, transcripts, summarization, social posts, artwork, proofreading, production and promotion, schedule, maybe figuring out who to interview, and things like that.

Why you would wanna do this stuff? So, I mean, research, I'm old enough to remember when Wikipedia came out, right? Wikipedia was kind of treated the way AI tools are being treated today, which is to say that people in academic spaces were against Wikipedia altogether. So they didn't allow it to be used for research papers.That has since changed, and something that we thought about in college was using it as a starting point, right? Because in Wikipedia you have to cite your sources, so you could just jump to the sources section and get the primary sources for the article.

AI tools can be the same way, right? It gives you a jump off point or at least find and verify that information to give you terms. For example, I didn't know fishermen were called anglers, right? I didn't. That's something that ChatGPT taught me, right? And so with research, it could be a jump off point.

Summarization. It could be a first draft, right? Most of the summarizing that I get from my shows focus a little bit too much in one area. Like if you listen to my other show, How I Built It, I recently did my 5 Favorite Automation Tools, and I think I gave pretty equal time to each of them. But for some reason they thought I talked about Hazel and Plex a lot.

Plex felt like a throwaway line. I mean, they need to go back and read it, but it definitely wasn't the focus. And I don't think I gave any more time to Hazel than I did to any of the other tools. But ChatGPT apparently the way it or not ChatGPT, the tool I was using, the way it analyzed the transcript felt that I had mentioned Hazel more than anything else. So then the same thing goes with social posts. So like all of these things give you a good starting point. I think.

The promotion and production schedule is really nice because at least it gives you concrete dates for stuff that you can then visualize, right? It also actually gave me these kind of boiler plate types of promotions, right? So, “Hey, the new episode is out” and then two weeks later, “Hey, in case you missed it…”, and then 30 days they said like, “last chance to listen…” which I didn't like that, right? If I were really digging into this, I would say gimme a different reason. Maybe that could be like a monthly recap or something like that.

So, but all of these things give you the opportunity to start without a blank slate because for some people starting is the hardest, right? And if you don't know where to start with your summary, at least ChatGPT gives you a place to do that, right?

I would always say when I was doing web design, my clients don't know what they want. They know what they don't want. And I think that rings true for a lot of us. Right? We don't know where we wanna start, but if you give us the wrong thing, definitely we'll tell you.

All right. So in the last few minutes here, I wanna give you a few tools that are worth looking at.

The first of course is ChatGPT and anything that integrates ChatGPT.

Notion Craft are both note taking apps that have integrated ChatGPT or some other generative ai. So you can highlight text and ask for a summary. That's how I was doing transcripts originally.

Dall.E is a…you know, Dall.E is an image generator.

ElevenLabs is that voice synthesizer.

Brandmark.io is a logo maker. I was not impressed with it, but again, maybe it'll give you some ideas, like you could try it for free and then you would pay for the logo that you like. So maybe that will give you some ideas for potential artwork.

Grammarly, I guess could be considered an AI tool, but the one that I've been using the most, so I'll tell you two. Capsho and Castmagic.io. I've been using Castmagic a lot because as I record this and when I bought it, they had an Appsumo like lifetime deal. That was definitely good enough for me to at least try. So with both of these tools, I'll focus mostly on Castmagic at this point though, because Capsso, I haven't really given a fair try to. You can upload an MP3 and then it will generate a transcript. It will generate quotes, like quotable moments. It'll generate titles, keywords, and introduction. Timestamped overview questions and answers from the show. Clips, a real script, a LinkedIn post, a newsletter, and a Twitter thread. Again, this is not perfect, but the Twitter thread was good enough for me to rework a little bit and publish on Twitter.

So again, this is a starting point, right? I don't know that I would've published a Twitter thread on this episode if I didn't have Castmagic to give me a starting point for things to talk about. I'm really curious to see how this is going to do with interviews, cuz right now I've only tried it for solo shows, but I mean, at the lifetime deal price, it was definitely worth the purchase.

So I think that if you are diving into generative AI today as you listen to this, as I record this, it's mid-May 2023, ChatGPT is definitely a great one. Like it'll help you reword things and proofread and things like that. Summarize stuff. You can also have it create a script, right? I've had it write like 500-2,000 word blog posts based on a topic. It's at least a starting point. So, ChatGPT is definitely one of them.

I use Raycast ai right now. I'll see what the cost analysis is for Raycast plus ChatGPT-4. And then tools like Castmagic.io to take your episode, create a transcript, and then create all of the collateral.

If you wanna take it one step further, there are tools like Bannerbear out there that will use AI to create promotional images. You can feed it in a template. And then it will automatically fill in a lot of the stuff that might fall more on the side of automation.

Canva has some AI assistance built into it now, and so it could help you come up with concepts and things like that.

I'll link all of this in the show notes over at [profitablepodcaster.fm]. It'll also be in the description for your podcast player of choice. But that's it for this episode. Just some ideas for using generative AI and then some potential tools. Check them all out in the show notes.

If you like this episode or you have other ideas, let me know at [profitablepodcastor.fm/feedback].

Thanks so much for listening. And until next time. I can't wait to see what you make.