Build, Buy, or Wait? The AI Content Battle with Jeff Umbro, CEO at The Podglomerate

(Intro Narrator at 00:00:00) Today, we're talking to Jeff Umbro, CEO at The Podglomerate, about the future of podcasts and what might happen to human connection. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:17) But as I was preparing, I learned something that I didn't know we would be talking about. You guys made a fully AI podcast? Like, as in the characters speaking in it are artificial intelligence?

(Jeff Umbro at 00:00:29) Yeah. Give me a second. I'll show it to you.

(Joel Beasley at 00:00:32) Okay. I just saw that Google NotebookLM thing.

(Jeff Umbro at 00:00:36) Oh, yeah.

(Joel Beasley at 00:00:36) Have you seen that?

(Jeff Umbro at 00:00:37) I have. But you tagged me in that, actually, I think.

(Joel Beasley at 00:00:40) Oh, I did?

(Jeff Umbro at 00:00:41) Yeah. Yeah. But I, yes, I've seen it. I haven't really done a deep dive, but there's a big Ars Technica article out there that I haven't read yet that's basically talking about how awful it all is and how there are now people like, AI bots are reviewing the AI-generated podcast or book or whatever it was. So, yeah. Here, let me share my screen with you so you can see this. So this is the video version of the podcast that we made. Okay. None of this is humans.

(Joel Beasley at 00:01:19) None of it's humans. "Welcome to En Route, a show that dives into the rich histories of countries, cities, and cultures from around the world." Okay. Yeah. En Route. Why do we keep saying that we're AI? Why can't we just pass it off as, like, they're humans? "And produced by..."

(Jeff Umbro at 00:01:35) Yeah. Why say AI a bunch? Because I don't want to pass off that creation is something that we did on our own, for both sides of that coin. On the one hand, to be blunt, if it's not good or if the pronunciations are off or if the translation is wrong, then I don't want to be blamed for that. But then on the same note, that entire thing is generated from stuff that we pulled from Wikipedia. So it feels weird to kind of take credit for it. And ultimately, we didn't release that. I don't plan on releasing that.

(Joel Beasley at 00:02:14) Oh, really?

(Jeff Umbro at 00:02:14) Yeah. A big part of the reason why is that it's good. It's much better than I would have anticipated, but it is not up to the standards of what I would want to put out in the world.

(Joel Beasley at 00:02:25) What about it specifically?

(Jeff Umbro at 00:02:28) Well, you heard the big flashy introduction and that sounded pretty cool. And there's a lot of novelty because it is a robot. But there's no fact-checking. It feels a little funny to be taking the script from Wikipedia because while it is Creative Commons, it's not stuff that we actually did the research on ourselves. There is the cultural aspect of that where, if we were going to put that out, there are people that may view it differently. For example, if we did an episode about Germany or the Roman Empire or something, or even an episode about Ukraine, what exists on Wikipedia is not necessarily going to be the same thing that is culturally acceptable for somebody in Ukraine or somebody who is Jewish listening to it, something about the Nazis or something like that. And really, the only reason we produced that in the first place was to test the tools and because there was a little inkling in the back of our heads that this could be really cool, that we can produce this thing at scale really easily and put it out in the world. But at the end of the day, the risks associated with putting that out there, especially as pretending that it's a real person, in my opinion, outweigh the benefits that we would see on the other side of that. Because, I mean, if I'm being honest, that's not a show that's going to generate millions of downloads and get tons and tons of revenue behind it. That's possible, but it's not going to happen overnight.

(Joel Beasley at 00:04:12) Yeah. That's interesting. So I've been having this conversation. I've got five people in my network who are competitors to us that produce podcasts that I've just gotten to know. And when I kicked it around through email and through LinkedIn and stuff like that, the reaction I got from them was actually not one I was expecting at all. Because most of us are forward-thinking with technology and adapting tools. But it seems that there's this barrier that I don't seem to have. I seem to have no barrier in the sense that, oh, yeah. In the future, there's AI avatars. They are everything from models to authors to speakers to singers, and people will see their profiles. They'll appear alongside of us on Instagrams and X and all those social medias, and we'll judge them based on their ability to put content out.

(Jeff Umbro at 00:05:07) Right. I'll say this, that there's two conversations that I'm having. There's the one that's happening in front of a microphone like this. And then there's a second conversation happening where it's, people are talking about, "Hey, could we do this and just not put our names on it?" And, yeah, a lot of people have brought that up to me. We're not going to do that, but the question has been asked many times.

(Joel Beasley at 00:05:32) I want to do it. I don't know if I would want, I don't want to put my name on it and be like, "This is a Joel Beasley or Podchaser production." But I believe that, okay, when the LLMs first came out, there was this one dude that wrote and published 900 books in three months.

(Jeff Umbro at 00:05:49) Wow.

(Joel Beasley at 00:05:50) And he made great money.

(Jeff Umbro at 00:05:52) Wasn't there that guy on Spotify who published a bunch of AI-generated music who just got sued?

(Joel Beasley at 00:05:57) Yeah. Yeah. Yeah.

(Jeff Umbro at 00:05:57) Yeah. I mean, I think part of the hesitation, in my opinion, is also that nobody really understands. It's the Wild West. A lot of people don't really understand the limitations at this point. And I think a lot of people are a little bit fearful of being first and getting it wrong.

(Joel Beasley at 00:06:16) Oh, most people are scared.

(Jeff Umbro at 00:06:17) Yeah. Yeah. Yeah.

(Joel Beasley at 00:06:19) I tend not to be. But I would create just a phantom co and be like, "Yeah, XYZ Media," and then just start putting out. I would make profiles of people, digital people, through, you know, using Venice. By the way, if you haven't checked out Venice, venice.ai, I think it is. Have you seen that before?

(Jeff Umbro at 00:06:38) I don't think so.

(Joel Beasley at 00:06:40) Uncensored models.

(Jeff Umbro at 00:06:42) Oh, okay.

(Joel Beasley at 00:06:44) Okay. venice.ai. Uncensored model. So it's basically, it's the same quality as a GPT or a Claude as far as interaction with it, but there's no ideology programmed on top of it. It's just the raw model.

(Jeff Umbro at 00:06:59) Understood. Yeah. It's interesting because I don't know. I'm very curious how this is all going to work, especially how it's going to impact production. And there are a lot of tools that I'm really excited about that are pretty fascinating to me. I will also say that, simultaneously, parallel to that, I have yet to really meet anybody who is taking anything that's generated by a lot of these models that they're using straight out of the box. Everybody is being incredibly thoughtful or trying to be incredibly thoughtful about what goes into their models and what the inputs are that are going to generate whatever comes out of it. And then everyone I know is spending a lot of time reviewing these things to make sure that they're being developed properly. And maybe it gets better over time, but I don't see it today. I guess, put another way, I feel like a lot of people in podcast production, at least, if you're operating at a beginner to novice level, some of these tools can make your stuff really, really incredible and bring you to a more intermediate level. But if you're already operating in the advanced space, then these things really, maybe they make your life a little bit more convenient, but they're really not going to do anything to advance the craft. And that's the big thing I've really been noticing across the board. I think people who are just hobbyists should be using ChatGPT to help write their show notes. But I think that if you find this big, robust production company that has a team who's doing that and doing it well, ChatGPT is not going to do a better job.

(Joel Beasley at 00:08:57) I disagree.

(Jeff Umbro at 00:08:57) Okay. Yeah. No worries.

(Joel Beasley at 00:08:57) I would say, I would say what I would do is, first of all, we all, so my background is the software development world. Right? So we kind of blanket everything with GPT, but there's a lot of models and they can be fine-tuned. What I would do is I would take historical performance of the good, high-quality human stuff and custom train an LLM—not just paste it into GPT—custom train an LLM with embeddings and everything to understand these concepts and then replicate it for future episodes.

(Jeff Umbro at 00:09:28) Yeah. Well, that's the thing. I think that is the question. It's like that's working, and it's a data-driven approach. But as you've seen and as everybody has seen, sometimes what works is not what the data is telling you. And so I don't know. I think there's an argument to be made on both sides of this. And, you know, we've talked about this before. I'm experimenting with all this stuff and trying to learn it. I'm just not convinced as much as you are anyway that this is the future.

(Joel Beasley at 00:09:59) Yeah. Oh, I am. I'm very convinced. And I can help bring you over to the right side here shortly. Okay. But I actually heard a pretty good explanation from Ethan Mollick, this guy, "Co-Intelligence," working with AI. I follow him on LinkedIn and Twitter and all that. And he gave some example where he said, a lot of people misinterpret. So what OpenAI is doing and what a lot of these model companies are doing is they're building intelligence. But they don't care about your use case. All the people that are building use cases on top of GPT, they don't really care about that. What they care about is getting to AGI. And if you look at OpenAI's spec, they have five levels of AGI and they've just crossed into level two and they're about to cross into level three. And so they're trying to do that. And then they're basically putting it out to the world and being like, "All right. Here's this intelligent digital being that we made. Now do with it what you will." Right? And so it's, if you just walk up to the intelligent being and it knows whatever it happens to know and you just start asking the questions, it's going to be moderately good at a lot of different things. It's going to be really great at some things. Right? But what they also have given you, and not just OpenAI, all the other models and the technology stack as a whole, what they've also given you is the ability to work and train the model. So it's like if I give you a PhD person and they walk into your business, PhD, brilliant individual, and they walk into Podglomerate, your business, and you're just like, "Give me the answer to this problem," it's going to be fairly generic. But if you show them all of your business and teach them about all the intricacies of your business for two years, then they're going to have really good insight. Right? And so that's the lacking point. They're putting their intelligence out.

(Jeff Umbro at 00:11:49) Yeah. But just to push back on that, I don't disagree with that. But if you're putting in that much energy into training this thing, sure, at some point, that is going to be really useful to you. But are you putting in more than you're getting out, in the short term at least?

(Joel Beasley at 00:12:06) Well, no. They're burning through cash like crazy.

(Jeff Umbro at 00:12:08) Yeah. And so that's the thing. I don't know. I go back and forth on it. It's like the AI conversation is interesting to me, but it's not exciting to me. I think that there are more compelling things happening that I'm putting my attention towards. And, yes, is there the fear that I'm going to miss the boat and we all get technologized out of a job? That's there for sure. But I feel like even if you're right and everything is able to be optimized to the point where these little ghosts in the machines are able to do the things that we're doing better than we're doing them, I don't think that the human element of that is going to go away. Do you think that people won't be working in the future for jobs that can be done by these things?

(Joel Beasley at 00:13:02) Yes and no. I like to use the horses analogy.

(Jeff Umbro at 00:13:05) Yeah.

(Joel Beasley at 00:13:05) So if you look at horses pre-cars, it's involved in everybody's lives and everybody has them. There's tons of horse jobs. There's a huge horse economy. Now, it still exists. There's still millionaires that have horse businesses. I mean, it's still a thing. It's just not at the scale at which it once was. And I think that's—

(Jeff Umbro at 00:13:23) It's going to be like artisanal podcast production and the humans are doing it.

(Joel Beasley at 00:13:28) Yeah. The wealthy people will have it handmade by the top people in the world, and then the masses will have super high-quality product that's just AI-generated.

(Jeff Umbro at 00:13:36) I don't know. That scares the shit out of me.

(Joel Beasley at 00:13:39) It does. But I'm optimistic, though. So here's a, I'm a hybrid. I'm not all e/acc, which is this term about whatever computers are, they answer everything. I'm very optimistic. And what I'm trying to do is, without filters, understand the time scale. It's like I'm standing on an island and I see a tsunami coming. I'm like, "All right. We're going to get wrecked by the tsunami." Right? But I can't tell if that's one year away, three years away, or five years away. I can't tell. I'm bad at telling the scale of distances, but I can see it and I know it's coming. And so what I'm trying to do is figure that out. So my question of if this is happening is just completely answered, unequivocally, yes. But now, how long is that going to take? If you want to help better understand how poorly you and I as humans understand scale, my favorite thing to remind myself of how bad I am at this is to know that 110 years ago, we got electricity. Okay. So we went from no electricity, no power lines, nothing. And that was 110, that's two people ago.

(Jeff Umbro at 00:14:53) Well, I just listened to this fascinating podcast from Outside/In from New Hampshire Public Radio about the history of refrigerators. And that was about 100 years ago. And there's a guy who, I'm going to butcher this. It's in the episode. You guys should check it out. But there's a guy who tried to live his life and consume his food the way that they would have before refrigerators. And the first thing that he realized is that he couldn't make a cheeseburger because, you know, the cheese comes in the spring. You slaughter the cows in the fall. You need the grain to come in the spring in order to get the bread, and lettuce and tomatoes grow in different seasons. And it kind of ruined a lot of things for me because they're saying that the apples that I buy, outside of the fall, they're probably left over from the previous fall, and they're just in these cold storage units where they've removed all the airflow.

(Jeff Umbro at 00:15:47) And anyway, it is what it is. I love apples, and I'm gonna keep eating them. But it really put things in perspective. And to your point about how we didn't have electricity 110 years ago, we didn't have flight until 80 or 90 years ago.

(Jeff Umbro at 00:16:05) We didn't have cars in everybody's driveways until the forties or fifties. I get it. There's a pace of technology innovation that is going to come, and it is gonna be the kind of thing where hindsight's 2020 and when it's here, we're gonna be like, how did we not know it was coming? But I don't know. Maybe I'm just being dense.

(Joel Beasley at 00:16:31) Let me talk about the optimism here.

(Jeff Umbro at 00:16:33) Okay.

(Joel Beasley at 00:16:34) Okay. So the optimism that I experience is, and I got this from another guy I talked to who owns a podcast production company that's probably 50 times the size of you and I's combined. Right? And I was like, hey, what do you think about this?

(Joel Beasley at 00:16:48) I had dinner with him in Nashville this past week. And he said, he goes, okay, I understand. Because I sent him the Notebook LM thing privately. And he's like, I see that.

(Joel Beasley at 00:16:59) But human relationships and interaction with each other and audience building aren't gonna go away. And I'm like, he's exactly right. They're not gonna go away. Also, people are gonna develop preferences. I believe you and I could be sitting here in five years, Jeff.

(Joel Beasley at 00:17:15) And it's a common thing in our culture to have a preference if you consume AI-generated content or if you follow any AI-type people. And I think there will be a preference, especially for the next couple generations, for organic or human-based content. And I think that'll drive the prices up, and whoever is doing it well will do well.

(Joel Beasley at 00:17:38) So I'm not concerned about this as far as destroying the business. I'm not concerned about that aspect at all. But I am, I'm just paying attention to what's coming at us.

(Jeff Umbro at 00:17:50) Yeah. I don't disagree with the idea that we should all be paying attention. I do disagree with the idea that this is inevitable. And I don't think, and maybe I'll be eating these words in a few years, but I do not think that this is gonna be something that completely replaces humans in terms of the work that we're doing. And I just don't think that it's very good at it.

(Jeff Umbro at 00:18:12) And maybe these things get better at storytelling and we're able to have better inputs to make them better at storytelling. I just haven't seen it, and I don't know. I will see. I just don't know that that's something that's actually in our future. And, you know, hopefully, I mean, hopefully I'm right and you're wrong in terms of the business stuff, but also I'm not going into this super naively saying it's not gonna happen.

(Jeff Umbro at 00:18:44) I'm paying close attention. I actually this morning did a training with my staff about some of the AI tools that I'm using and seeing every day in order to make them aware of all of that. So it's a thing. It's very much happening. And you're right, you can't just be blind to it, but I don't agree with you in terms of where I think this is going in the future.

(Joel Beasley at 00:19:08) Well, tell me where you think it's going.

(Jeff Umbro at 00:19:12) I guess maybe your point about people having a choice and there'll be human-curated stuff, and then there's gonna be AI-generated stuff out there, and they're gonna live side by side. That's probably somewhat true, but I think the human stuff is still gonna be the vast majority of what we're seeing out there. If for no other reason than what are humans gonna do if they get worked out of these jobs?

(Joel Beasley at 00:19:36) Well, they're gonna be spending time with each other in person.

(Jeff Umbro at 00:19:39) Well, people have been saying that forever though, and that's not happening. What is the Peter Thiel thing? We were promised flying cars and we got Twitter. I think time and time again, we figured out that if we reduce people's need to work, they're just gonna find other ways to do the work. There are certainly some things that will be made better by us using these tools.

(Jeff Umbro at 00:20:06) And even in podcast production, you'd be kinda crazy not to use these LLMs at the very least to give you the V1 of your show notes or something, which will save people time. Where I do think that there is a limit to what will be done by these things, though. And it's gonna be different for every industry. But I don't know. Maybe I'm just rosy-eyed and too idealistic.

(Jeff Umbro at 00:20:36) But I don't know that it's something that is gonna work. I also don't know why people want it, if I'm being honest. It strikes me as something that's going to disrupt everybody's lives in a negative way.

(Joel Beasley at 00:20:57) Yeah. But we're bad at long-term stuff, assuming. So short-term, if you walk up to 20 employees who's making, you know, a couple, let's just say they're making good money, and you offer them a tool for $20 a month that does 80% of their job, you think they're not gonna take it? That's why they're doing it, because it makes their lives easier.

(Jeff Umbro at 00:21:15) Yeah. Yeah. I don't know. I think that you're not wrong. I think that it's not gonna end that way.

(Jeff Umbro at 00:21:22) And I don't have a good reason for why other than I just don't think any of it is actually that effective. And I totally get the argument that it's gonna get better. But until I see it, it's a hard thing for me to grasp.

(Joel Beasley at 00:21:38) Yeah. And I completely understand too. Right? I talk with developers, software engineers who've been spending their entire years building this stuff, and even them. Some of them have a hard time seeing it.

(Jeff Umbro at 00:21:48) Yeah.

(Joel Beasley at 00:21:48) And so it's not uncommon. But—

(Jeff Umbro at 00:21:54) Yeah. Well, I'm curious though, because obviously we're gonna continue going down this road, and you know, we should have a conversation in a year and two years and three years, and we should talk about where this actually went. I do have very strong convictions for a lot of technology gains in a lot of different industries. For example, I'm incredibly bullish on Spotify right now. I think that Spotify at some point is going to become the future platform for audio everything.

(Jeff Umbro at 00:22:29) You're gonna be listening to books, music, podcasts, your Sunday church—

(Joel Beasley at 00:22:38) Sermon?

(Jeff Umbro at 00:22:39) Yes. Your Sunday sermon. If it's audio, it's gonna exist there. And I think that there's a few examples of that. That is coming in the future in all different industries.

(Jeff Umbro at 00:22:53) I don't know. It's probably my famous last words and, you know, I wasn't intending on coming onto the show on the record to say these things, but—

(Joel Beasley at 00:23:01) I know. This is why it's fun.

(Jeff Umbro at 00:23:03) Yeah. I don't know. I don't see it happening. And I do think, yeah, and I don't think it's happening because I don't think, at the end of the day, I don't think most people want it to happen.

(Joel Beasley at 00:23:14) Yeah. So I would disagree. I'd say most people, they don't want the negative effects of it to happen. If you ask them, do you want AI to come in and take your job and do all this stuff, they'll say no. But if you say, you want AI to come in and do 95% of your job for $20 a month, they'll say yes.

(Jeff Umbro at 00:23:27) Yeah. But then they don't have a job.

(Joel Beasley at 00:23:30) Like I said, people are really bad. Look at the national savings rates and how people manage their finances on a national scale. People are really bad at long-term stuff. It's sub-20% that are good at long-term strategies. You're in a weird world because you're pretty good at long-term stuff because you're a founder.

(Joel Beasley at 00:23:49) Right? You have to make long-term decisions, but most people don't.

(Jeff Umbro at 00:23:52) I get it. And I've seen it with a lot of my friends and stuff. And I've seen the stats and everything, but I don't know. Maybe I just choose to look at the positive side.

(Joel Beasley at 00:24:01) So positive side, I think the businesses that we're in, production businesses, are gonna do great. I do think that the things that are on my radar as things that I believe would slow down this from happening. So let's just say that in 100 years, it's all happened. Okay? In 100 years, AI, there are AIs walking around the street and you can't tell the difference from them from humans, just like these humanoid robots.

(Joel Beasley at 00:24:25) Right? So if we're good with that, we'll get there. Even if it's not 100 years, let's say 1,000 years. So now it's like, okay, we got there. Now we're just in between where we are today and there. And so the question is, how is it gonna play out? What could slow it down? One of the things that could slow it down is, hey, our government, when healthy and working properly, is supposed to serve the people. Right?

(Joel Beasley at 00:24:47) So we, the people, if we get together and we vote that there needs to be some type of restrictions on this, then we can do that. And I think that'll help. However, historically, the government is very slow to adapt, and they only take action after lots of deaths. So the safety bureaus, seat belts, all of these things usually only happen after a ton of pain. It's not proactive, it's reactive.

(Joel Beasley at 00:25:15) So the job loss will, I believe the job loss will happen. I believe there will be a lot of pain and then I believe they'll step in and say, you know—

(Jeff Umbro at 00:25:26) Well, do you remember when Sam Altman went in front of Congress last year for a hearing on AI? By the end of it, they were literally asking him if he wanted to come on board and write the regulations for how AI should be served throughout the country and how it should operate. And so to your point, that is not a healthy response, offering the guy who owns the largest LLM company to write the rules for how the LLM companies can operate. That's terrifying, I think.

(Joel Beasley at 00:25:57) It's a double-edged sword. Right? Because you have to have people who are competent on it. But you should, if you're gonna put the board together, you need a board of people who are competent in various areas. Competent in the business side of the LLMs, competent in the ethics side of the LLMs, competent in the actual programming of the—

(Jeff Umbro at 00:26:16) Well, even that though. Isn't that how the OpenAI board was set up?

(Joel Beasley at 00:26:21) I don't know. I don't know.

(Jeff Umbro at 00:26:24) There is a nonprofit board that owns the for-profit company. And yeah, the idea was that they would put a bunch of people on that board for all of the reasons that you just mentioned. You know, the ethical, the business, the moral side.

(Jeff Umbro at 00:26:42) And at this point, everybody except the business side has quit. Yeah. They don't exist anymore. So I think that's, yeah. That's capitalism, baby.

(Jeff Umbro at 00:26:54) Everybody's just gonna continue to do what's best for their own wallets. And I don't know. I think we're in a really bad place in general. And to your point, our government should, in theory, be functioning in order to help us with this stuff and be forward-thinking here. But we're still arguing about stuff that the broader public figured out 50 years ago, you know?

(Joel Beasley at 00:27:18) Yeah. Yeah. And we live in a clown world. But we're working through it. We're figuring it out.

(Joel Beasley at 00:27:24) I still maintain my optimistic bent. Right? I think things are gonna work out. And then I think the best way to make sure things work out are to do good work, build strong relationships, and then every once in a while, take a peek up from your work, look up from your desk, and see what's around you and what's coming towards you, what's gotten closer since your last check-in. And so for me, I've been following pretty adamantly AI, humanoid robots, and small modular nuclear reactors for 15 years.

(Joel Beasley at 00:27:52) And so I have a recurring event in my phone once a year, and I go check in on these different spaces. And lately, the AI space has been, whoo, on fire.

(Jeff Umbro at 00:28:02) I saw that, I think it was Microsoft wants to open up Three Mile Island's nuclear plant so they can use it to power their data facility. It's funny because I love the idea of opening up nuclear plants, but probably other things that we could use that power for. But—

(Joel Beasley at 00:28:22) Well, they're gonna be, it's yeah. And then they're pushing it. Right? They need the power to do the data centers, and nuclear has the best energy density ratio. And so they're gonna be doing that.

(Joel Beasley at 00:28:34) Yeah. I've seen several of the top tech companies. I think Oracle did something too, but several of the top tech companies have funded nuclear projects because it's—

(Jeff Umbro at 00:28:43) Including Sam Altman. I know we keep going back to him, but it's so funny. He has his hands in everything. I actually worked with him very, very briefly in 2016.

(Jeff Umbro at 00:28:57) He founded this voter registration startup called VotePlz, V-O-T-E-P-L-Z. And for four months, I went and lived in San Francisco, and I helped run the comms for that organization. Oh, nice. I didn't know who he was at the time. He was doing something with Reddit, and he was at Y Combinator still.

(Jeff Umbro at 00:29:15) And he was in an office on the other side of the room. He'd show up once a week and throw his feet up, his legs up on the desk, and would talk through this stuff. I only ever had two conversations with him. But it's so funny because now he's set to rule the world. And it's just, like, the guy who had the voter registration startup in my mind.

(Joel Beasley at 00:29:38) Yeah. Well, I mean, the models are all out there. Right? And so you can go access them. Venice AI is hosting many of the models. And so the, let's call them alt models from the mainstream ones.

(Joel Beasley at 00:29:50) They're getting just as good, if not, they're basically parity with the commercial models now.

(Jeff Umbro at 00:29:56) Well, there was, I think it was an episode of This American Life or something. But they did a whole podcast about one of the guys who helped develop ChatGPT in early version before they put any of the filters on it, had given access to all of his friends at a wedding party or something to use this thing. And they actually had a lot of conversations with, I think they published a book of poetry that this thing wrote. And I'd have to look it up, but it was a beautiful episode. And by the end of it, I'm like, oh, what are we doing?

(Jeff Umbro at 00:30:34) But anyway, well, are you listening to any podcast about AI or produced by AI? Or outside of the realm of functionally and technologically how this works?

(Joel Beasley at 00:30:45) So I'm gonna be on the anti-AI, pro-human side when content comes out. I will pay more for, I look at concerts and stuff as a parallel. Right? The artists now, their revenues are so far down from streaming.

(Joel Beasley at 00:30:59) They make all their money on concerts and touring. I think we're gonna see a push to in-person events. So I think we're gonna see more like All In. If you see them, they do live podcast recordings. Daily Wire does live podcast record. I think we'll see a move towards that. And I think we'll see, I think the conference industry is going to explode. I think it's going to grow significantly because we're, we are soon approaching, Jeff, the moment where you don't know if I'm actually here or if I sent my digital twin. But if we're in person, you know that I'm there. And so we're a trust economy.

(Joel Beasley at 00:31:37) Right? We do business with people that we trust. And so I think that'll force us to do more stuff in person together, or with our existing relationships we had before the boom took place. Right? Well, do—

(Jeff Umbro at 00:31:50) You remember? Yeah. Yes. I agree with all of that. But I think that has nothing to do with AI. I think people are just on their computers too much.

(Jeff Umbro at 00:32:02) And I think that people want more in-person events. I think COVID really highlighted that and put it under a magnifying glass that people really need community and third spaces more than they have it. Really, what I'm talking about is Evan Ratliff, who is the founder of Atavist and was one of the hosts of the Longform podcast. He put out a show a few months ago called Shell Game, and it's really well done.

(Jeff Umbro at 00:32:29) His wife is the former head of audio at the New York Times. And so it's this super reported thing where he basically had an AI version of himself live his entire life. And I won't spoil it, but he goes through the—

(Joel Beasley at 00:32:46) Spoiler it.

(Jeff Umbro at 00:32:47) Yeah. Well, no, I mean, I didn't love it personally. He got a lot of praise throughout, but ultimately he had this version of himself, this robot Evan, went in and would try and cancel his credit card or call customer service or something, the most basic tasks he couldn't do. And again, that is probably representation more of where we are in that process than the future.

(Jeff Umbro at 00:33:21) But it really underlined to me everything that I'm thinking is what's actually happening when people try and act on it. Separate from that, there was an article, I think, last year in Wired that was all about generative AI podcasts. And there's been a couple in the last few weeks that they're all here. And I remember the headline was something like, "Generative AI Podcasts Are Here. Prepared to Be Bored." And to me, that is this world.

(Jeff Umbro at 00:33:55) It's like the stuff that is being produced just isn't that good. And the podcast that I showed you in the beginning of this call that we produced, that was one of several reasons why we didn't want to release it. It just wasn't that good. It was effective, and it did exactly what we wanted it to, but it was devoid of anything human. And we weren't able to put in any flavor or narrative or voice.

(Jeff Umbro at 00:34:25) And you're literally just listening to a Wikipedia article read to you, why you should go to—

(Joel Beasley at 00:34:31) The portal. Did you have software developers involved on this, or is this just you guys?

(Jeff Umbro at 00:34:34) No. And I'm not saying that it's impossible to do it. This was out of the box, more or less. We had somebody going through and tweaking it a little bit.

(Jeff Umbro at 00:34:43) But I could be wrong here. But I do stand by everything I've said.

(Joel Beasley at 00:34:49) Yeah. No. I mean, they're your experiences, right? And you had them.

(Joel Beasley at 00:34:52) Yeah. Look. I think we're really close to it being better than humans. I mean, there was that report Ethan put out. They took creativity, a creative test essentially, and had things created and they showed two sets to humans and you got to vote on which one was more creative and the AI won and they didn't know which one it was blind, right? So the AI won out. So the AI was more creative at human creative tasks than the humans were.

(Joel Beasley at 00:35:18) Right? So the AI won. So the AI was more creative at human creative tasks than the humans were—

(Jeff Umbro at 00:35:25) across the—I mean, I'll check it out. Yeah. I hadn't heard that. Yeah. And, you know, I'm—

(Joel Beasley at 00:35:30) not running any of these studies, right? I just follow these people and read it.

(Jeff Umbro at 00:35:35) Well, that's also super subjective. It's, you know, boring or creative is in the eye of the beholder.

(Joel Beasley at 00:35:40) Yeah. Yeah. That is true. Nice. What are two tips for growing your audience for the people that listen that also have their own shows?

(Jeff Umbro at 00:35:48) Sure. So the two free tips that I'll give you, and then I'll give you one more—

(Joel Beasley at 00:35:53) Tips. Yeah. Two free, one pay. All right.

(Jeff Umbro at 00:35:56) Done. So the free ones are, there is a big difference between marketing and marketability, and you should make sure that your show is actually worthy of bringing in net new people. I look at this the same way I look at a website bounce rate. You can drive as much traffic as you want to your show, but if it's not sticky, then people are just going to leave. And that applies to your existing listeners too.

(Jeff Umbro at 00:36:23) Like, you can grow your audience just by retaining your existing listeners as opposed to bringing in net new ones. And some of the ways in which you can make your show stickier is just to pay attention to stuff like your consumption rates, how people are listening, where they're dropping off. Like, are you getting other KPIs that you're seeing outside of the downloads? Like, I don't know, number of fan emails, leads generated, social media engagements. And if you're not doing that, what should you be doing in order to do more of that?

(Jeff Umbro at 00:36:55) So that's one. Second would be cross promos. And that's a really easy free way to grow your show. It could be interviews on other shows, audio swaps, or feed drops. Basically, just find like-minded shows, and you can do that by using places like Rephonic, which is a website that'll give you podcast neighborhoods.

(Jeff Umbro at 00:37:17) Literally looking up your show on Apple and scrolling to the bottom, they'll give you a bunch of recommendations for what other shows people are listening to that are listening to your show. And you can use stuff like Podchaser, where they'll literally give you the email addresses of everybody that's associated with that show. And then you can just shoot them a note and you can run these swaps. You can track attribution and conversion with something like Chartable, and you're able to really see how effective these trades are. And then you can keep iterating to make sure that what you're doing in the future is even more effective.

(Jeff Umbro at 00:37:49) It's a really simple math problem. If you have a 1% conversion rate, and if you have 10,000 impressions, then you know exactly how many people you're going to bring in from running that 10,000 impression ad swap, and you can run multiple on every episode of your show. Anyway, I can get very boring with this. So those are two free ways. One paid thing that I'm testing right now, but early on is incredibly effective.

(Jeff Umbro at 00:38:17) And, Joel, you might be interested in this too, is it's called Introcast. Introcast? Yes. I-N-T-R-O-C-A-S-T dot I-O. It's this really cool program.

(Jeff Umbro at 00:38:31) It's run by this former VC who is doing this, I think, in his free time. But he built a tool where you can run dynamic feed drops. And what I mean by that is you can actually dynamically insert a feed drop of your show into other people's RSS feeds. So I could actually plug in, if both of us were opted into this system, an episode of Podcast Perspectives, my show, as the second episode in the RSS feed for Modern CTO. So anybody who goes to your show on Spotify, for example, will see this most recent episode.

(Jeff Umbro at 00:39:09) And then underneath that, it'll say, "Introducing Podcast Perspectives" or something. And this is a host-read introduction from you, Joel. And then it'll be a three-minute clip of Podcast Perspectives, and then you'll do a CTA at the end. And then whenever I run out of money to promote this or we hit the impression cap or whatever, that episode will be removed from the feed, and it'll stop running.

(Jeff Umbro at 00:39:37) But—Wow. Yeah. And it's cool. And then you can just plug a new one in. And so you can do it on the buy side or the sell side. And it allows you to do this with a little bit less overhead and legwork.

(Jeff Umbro at 00:39:52) And I'm seeing pretty great conversion rates on this so far. We're pretty consistently seeing a sub-$4 cost per acquisition. And because it's opt-in, Joel, you would get the ability to say yes or no to Podcast Perspectives. It's really effective because you're only going to have the ad appear on a like-minded show. And, you know, it's going to appear if I approve it?

(Jeff Umbro at 00:40:15) Correct. Yeah. Oh, okay. Because you also have to read the intro and stuff. So, you know, jury is still out.

(Jeff Umbro at 00:40:24) We're still running our tests on it. But I'm seeing some very retentive listenership coming from this. Interesting. Yeah. So, and there's a thousand more ideas like that.

(Jeff Umbro at 00:40:37) We literally do this all day every day. There are 10 of us full time, and these are the only things that we're ever occupied with. So, well, at work anyway, until the robots take over. So—That'll be a while. Yeah.

(Jeff Umbro at 00:40:52) But yeah. If anybody's ever interested, you can just head to podglomerate.com and reach out. We do this all the time, and I'm happy to talk to anybody about it. So—

(Joel Beasley at 00:41:01) Nice. Nice. Thank you so much for doing this. Podglomerate.com. Jeff, the CEO and founder, he's the man, and he believes that I am right when it comes to AI.

(Jeff Umbro at 00:41:12) Hey. I might be asking you for a job if you are right. So—

(Joel Beasley at 00:41:16) It'll work out. It'll work out.

(Jeff Umbro at 00:41:18) Yeah. I'm sure it will. I'm not concerned. I don't think it's going to happen. But thank you for having me.

(Jeff Umbro at 00:41:23) This was awesome. Really appreciate it.

(Joel Beasley at 00:41:25) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn, or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.