AIMinds #004 | Justin and Jason of Transistor.fm

About this episode

Justin Jackson is the co-founder of Transistor.fm, a podcast hosting platform. He and his partner, Jon Buda, started the company in 2018. Today, the company has a team of 6 and hosts nearly 27,000 podcasts.

Jason Pearl, a Senior Software Engineer at Transistor.fm, played a big role in developing its AI-powered transcription feature. Jason has held various senior positions in product and engineering and was previously the co-founder of Deepworld, a Massively Multiplayer game with 1.4mm registered players.

Listen to the episode on Spotify, Apple Podcast, Podcast addicts, Castbox. You can also watch this episode on YouTube.

In this episode of AIMinds, we were delighted to have Justin Jackson and Jason Pearl of Transistor.fm share their insights and experiences in the world of podcasting and AI transcription. Here are some key takeaways from the episode:

1. The Journey: Justin Jackson shared his deep-rooted passion for audio broadcasting, dating back to his childhood experiences with talk radio. His foray into podcasting in 2012 has laid the foundation for many professional successes, emphasizing the vital role of podcasting in his career.

2. Embracing Slow Media: The discussion revealed the value of slow media, particularly in the podcasting realm. Justin elaborated on the unique aspect of podcasting that allows for a deeper, uninterrupted connection with the audience, fostering a sense of community and loyalty.

3. Overcoming Challenges in AI: The co-founders delved into the challenges of integrating AI transcription into their platform, emphasizing the importance of thoughtful user experience and the balance between technological power and user-friendly design.

4. Pricing and Product Design: The discussion provided a fascinating insight into the complexities of pricing as a fundamental aspect of product design, especially in the context of a service like AI transcription.

This conversation offers an unfiltered look into the evolving landscape of podcasting and AI technologies and is a testament to the vital implications for content creators and listeners alike. Transistor.fm's ongoing commitment to innovation and user-centric design positions them at the forefront of pioneering advancements in the podcasting domain.

As we traverse the ever-evolving podcasting terrain, it is through such discussions that we gain deeper insights into the transformative potential of podcasting and AI integration for creators and listeners. Tune in to the full episode to explore the intricate dynamics of podcasting and AI transcription and gain valuable insights from the minds behind Transistor.fm.

Fun Fact: The team at Transistor.fm prioritizes building a lean team of passionate individuals focused on customer service and creating a good work-life balance for themselves.

Show Notes

00:00 Building relationships, learning, and connecting through podcasting.
04:47 Small team, passionate, customer-focused, and relaxed.
08:35 Transistor: Podcast hosting with distribution and analytics.
11:45 Podcasting builds deep connection, offers flexibility.
14:29 Technology solutions must address user needs creatively.
19:46 Invite collaborators, one price for multiple shows.
20:48 Podcast explores challenges and opportunities of pricing.

More Quotes from Justin and Jason

Transcript

Demetrios:

All right, fellas, we are here today to talk about your journey. You are coming at us from Transistor FM. Justin, can you give us a bit of background on yourself?

Justin Jackson:

I started podcasting in 2012, I think, and was definitely into the idea of radio. Some of these are transistor radios, which is on brand. I grew up in Alberta, which is like, kind of conservative country here in Canada. And so there's lots of talk radio stations and had lots of good memories of driving around in the truck with my dad, listening to talk radio. Even, like, I remember being a kid, and once I got my license and I was driving by myself, I would turn off the radio and pretend I was on the show. The host would say something, and then I'd pretend I was, like, talking back and forth. So the idea of broadcasting via audio was always really appealing to me. And when I moved to this town, actually, it's a pretty small town, and I wanted to stay connected with the broader tech community.

Justin Jackson:

One of the first things I did was start a podcast called product people, which was just interviewing product people from all over the world. And that experience, I think basically almost everything, every good thing that's happened in my career since then can be traced back to starting that podcast.

Jason Pearl:

Amazing.

Justin Jackson:

The relationships I built, the connections I made, the things I learned, and then just the act of building an audience and putting something out there that people connect with was really significant. So, yeah, podcasting, for me, has been a big part of my professional life, starting around 2012. I think that's kind of how things work a lot of times. It's why it's hard to reverse engineer any entrepreneurial success, because often people will tell one part of the story, but there's usually quite a few layers. And for me personally, if I didn't have that layer of really being into talk radio, I don't know if I would have been as interested in podcasting if I didn't have the layers. I mean, this is when those early computer experiences are foundational. Without that experience making things on computers, starting in the kind of becoming teenagers, right when the web was coming out. So, like, the mosaic web browser, I think, came out in 93.

Justin Jackson:

I mean, that was when I was 13 years old. So all of those layers kind of build up and eventually culminate in maybe a business bet that pays off.

Jason Pearl:

I've probably been coding since I was like, eleven or something. So 30 something years later, like, in another life, I probably would have been some sort of physical engineer, like electrical engineer or something. But the feedback loop from coding is so tight, it's hard to give up that instant hit of success that you can get. So I sort of glommed onto that pretty early, I guess. I went to school, not for software development at all. I got enamored with finance and watched Wall street and thought that stuff was all super cool, so I went to school for that. And I'm, like, hearing from my friends as they're coming back from their first jobs and they're talking about how they're just, like, cold calling people 12 hours a day and trying to hawk whatever they're supposed to hawk. And that sounds terrible what you're doing.

Jason Pearl:

So I wound up sort of switching gears and realizing that software was like, it wasn't, like, clear that was a career path at the time. It was like, just becoming something that seemed like you could do professionally.

Justin Jackson:

Jason's done a lot of cool other stuff in software as well. He was part of a video game company for a while.

Jason Pearl:

I ran a little video game studio with a couple of friends, and we built a massively multiplayer, like, 2d crafting game. They got pretty popular, I think. All said and done, we had one and a half million registered users.

Justin Jackson:

Yeah, I think this is one of the cool things about running a small company is that we've been able to bring in all these people we've met throughout our lives and have someone like Jason's caliber on the team, and we're so small. I think Jason described at one time, you're like, the nice thing about transistor is like, we're all killers. We just all show up and we're really good at what we do. We're passionate about what we do, and there's very little cruft. It's just like a lean team trying to build amazing product, really customer focused, really customer service focused, but at the same time trying to build a really good life for ourselves and having good time off, having, slowing down during the holidays, being more relaxed, like, we have no KPIs right now, really. That's what we were talking about before. We can just trust everybody on the team to kind of show up every day and do great work. And, yeah, it's been really fun.

Jason Pearl:

You can see that in the product. A lot of people often comment, like, how simple things are and how uncluttered everything is, and a lot of that is because we just have to ruthlessly pare down what we're building because of the time available. And those constraints, I think, lead to a good product.

Demetrios:

You're being very deliberate and you're being okay with it being that deliberate and not feeling like, I need to add these 20 new features, because that's what people are asking for.

Justin Jackson:

John and I, we developed this strategy of wait and see, so something would come up and it'd be like, we got to do this, or the company's going to fail. And then we started realizing we could just wait and see what happened. And often it ended up not being a big deal, or it ended up manifesting itself, evolving in a way, in a different direction than we had originally thought. And then you can also see what everyone else has done, the reception from everyone else in a lot of ways. AI transcription, which is what we've just added, was that kind of feature. We've been talking about it for years.

Jason Pearl:

For a long time. Yeah.

Justin Jackson:

And it just kind of always bubbled around. Should we be thinking about, know, Jason was doing experiments on his own with different models, and eventually it kind of bubbles up and then it reaches this high watermark and it's like, okay, now it feels like we should act like now is the time.

Jason Pearl:

Part of it. Is the technology kind of caught up with our needs? The availability of a lot of this stuff, at least as far as I can tell, is pretty new at scale and at a cost that is viable, that we can do it or we can price it at a level that our customers find palatable. So we're plugging in Deepgram transcriptions now, obviously, which is why we're chatting. And the speed and the quality of it has been just phenomenal. We've been really pleased with what we're getting out. We've got a handful of users beta testing it now, transcribing their podcasts, and I think it's going swimmingly.

Demetrios:

Give us a bit of background on the tool.

Justin Jackson:

I mean, simply Transistor is podcast hosting the same way that you need web hosting for a website, you need podcast hosting for your podcast. We help people host their podcast. They upload their content to us, and then we distribute it via RSS to Apple podcasts, Spotify, Pocketcasts, Google podcasts, all of the different players. And then we also offer additional services on top. So, podcast analytics, you get a dashboard with how each episode is performing. We have built in podcast websites for your shows so people can go to your domain name and see all your episodes and subscribe from there and get show notes. We have dynamic ad insertion, which is another feature that Jason built once he came on. So you can dynamically insert an announcement to your podcast listeners.

Justin Jackson:

You can tell them about an event, you can sell, actually actual ad space, but that's essentially the business. It's a content management system for podcasters. They create their episodes in transistor. They publish them and manage their kind of publishing workflow through us, and then we do the hosting.

Demetrios:

I do love this idea of this slow movement. And it makes me realize, yeah, I will occasionally watch podcasts, but for the most part, it's just listening. And it is one of those mediums where you're generally doing something else when you are listening, and you just get to be a fly on the wall, a great conversation, ideally.

Justin Jackson:

So, like, social media is really designed to addict people, right? They want your eyeballs for as long as you can. So as long as you can keep people scrolling, accelerate their heart rate, outrage them or whatever, make them want to comment back or reply back or whatever. That system was not appealing to us. And what was appealing about podcasting is it's old tech, it's slow tech. RSS has been around for a long time. It's very simple. It's distributed. Yeah, it's right in the name.

Justin Jackson:

That's right. And there was something about the simplicity and this idea of slow media and slow tech. Meaning, when you're listening to a podcast, first of all, it doesn't demand your eyeballs, right? It doesn't demand your complete attention. It's not designed really to addict you. It's like you listen to a podcast when you're commuting or when you're walking the dog or when you're working out. You can't also respond right away. You can't have a knee jerk reaction. Like, if somebody says something that upsets me on a show, I have to finish driving home, park the car, say hello to my family, and then find a way to contact them, write them an email, and then say what I think there's just so much more time to have a considered reaction.

Justin Jackson:

It slows everything down. And from a creator perspective, it's different as well, because you're not beholden to the large platforms. So there's lots of stories of people who at one point, the algorithm was rewarding them, and all of a sudden, the algorithm took it away, and then all of a sudden, they have no audience. Whereas podcasting, it's slower to build an audience, but you build a deep connection with the people who are listening. And once you have those folks there, as long as you have your RSS feed and your RSS feed, you can bring it wherever you want. If you decide to leave transistor, you can forward your RSS feed somewhere else and it just forwards the traffic right through. As long as you've got your RSS feed, you've got this connection with your audience. And I heard a couple of our customers, I listened to a lot of our customers'podcasts, and they were talking about what would happen if Twitter went away.

Justin Jackson:

What happened if Twitter just crumbles? And they said, man, well, we'd still have our podcast. There's less people who listen to our podcast than follow us on Twitter. But these are the real know, these are the real fans. And if everything else crumbled, if Twitter went away, if Facebook went away or if Facebook took away our reach or whatever, we'd still have this connection. And these people who wake up every week looking for a show from us know, listen and have built this kind of ongoing, long relationship with us.

Demetrios:

So one thing that I wanted to ask, and Jason, this is probably more directed towards you, but Justin, if you have any antidotes too that you want to throw in there, how have you overcome challenges that you faced while building with AI?

Jason Pearl:

The foundation of all of it for podcasts is to have the information in some sort of format that you can use to do other things. So transcription was clearly like the first thing that we need to nail before we can add any other features around that. So that's sort of where we're starting. There are a lot of other nice things we can do with that or things that we think could be useful. I think we do a better job than a lot of places supporting chapters for podcasts. So you can display at what timestamp you're talking about a certain thing in show notes and YouTube, if you publish there, just like ways you can skip around in an episode and see what's going.

Justin Jackson:

Yeah, like, often technology is like a solution in search of a problem, right? And a lot of AI stuff can feel like that. It's like this is just like you're just trying to cram this solution and create a problem. So the biggest challenge, I think, is to go, is this something that people want? Is this something that podcasters want as creators? Is this something that podcast listeners want? What job does this do for them in their lives? And what is the kind of initial version of that that we could make a bet on and then see what happens? The hard part is actually the user experience. It's the UI. Right. Speaker identification especially is like a huge challenge. It's like on the engineering side, how do we identify that this is the person speaking? And then on the UX side, how can somebody identify who this is and type in their name and then maybe change it and then label this section as this person, but not this section. That takes some thinking and some creativity.

Justin Jackson:

One of my concerns with AI is that because the underlying tech is so powerful, it's just easy to just slap it in there without thinking about the user experience. How are people going to be using this? And how can we make it easy? How can we make it easy or even better, magical? Like, it just feels mean. It would be amazing. Jason's also done a lot of work in our show notes around. Like, you can insert a tag in your show notes that says chapters, and if your podcast editor has put in chapters in your mp3, we'll show those in your show notes. Here's the timestamp. And when you experience that, that feels magical. It's like, whoa, half my show notes are already written.

Justin Jackson:

And so a next step for this would be something like, hey, we've already identified all the chapters in this episode for you. It's just right here. We've done the work for you. We're looking for those kinds of experiences, I think. But the engineering side is challenging because you want to make it right and you want to have the accuracy as high as you can, and all the back end stuff has to be right. And then what you put on top is just even more important in some ways. Like how can you make this easy for people, intuitive for people, so that it's not adding more work? And my experience sometimes with AI is like, sometimes there's more work at the end of the day, especially lately, like Chad GPT four or whatever, it feels like it's talking back to me a lot. I'm like, hey, can you do this for me? It's like, I'm sorry, Dave, I can't do that for you.

Justin Jackson:

And then people say you have to just demand that it does it. Don't be polite. And then you ask it three times and then it'll do what? That's not a good experience. What we want is a good experience.

Demetrios:

On the business case of transcription, one thing that I often think about is how you're either going to need to charge more to the end user, pass on that cost to the end user, or you're going to have to let that eat into your margin. How do you think about that?

Justin Jackson:

This is the other advantage of wait and see. So we're like, let's wait and see what happens. And since then, we've had a number of competitors release AI transcription features, and some of them have more features on top. Some of them have less. And then there's been a bunch of third party apps that have come out as well. And often in a market, you're going to be anchored to whatever your competitors are doing. So if your competitors are offering it for free, it's going to be harder to charge. But if your competitors are charging for it, then it's going to be easier, and then you also have some sort of pricing anchor.

Justin Jackson:

It's still a bet. We're going to see how it goes. We've debated pricing on this a lot.

Jason Pearl:

Probably more than any other part of this.

Justin Jackson:

Yeah, talking about challenges, that is a challenge, because pricing is also product design. So when we were launching transistor, I had this intuition as a podcaster that the normal thing at the time was you had to pay an extra monthly fee for every podcast you started. So if you were on Lipson, you were paying them $19 a month for one podcast. If you wanted to start another podcast, you had to pay another monthly fee. And at the time, I'd done a bunch of podcasting experiments. So I had, like, four or five podcast feeds, and I said, john, I think we could charge a little bit more for transistor. Start at $19 a month, but allow people to host unlimited podcasts. So in a transistor account, you can start as many podcasts as you want.

Justin Jackson:

You can invite people to those podcasts to collaborate with you. So, like our podcast editor, he logs into transistor, and he only sees the shows that he's been invited to. That pricing decision ended up being a product decision because some people were looking for that feature. It became a feature of the product. And early on, that was a lot of our growth, was we were really one of the only people doing that. And so anyone that wanted to start more than one show, or if you were an agency or a business with more than one show, or even if you just had a bunch of friends that you podcasted with and you're all individually paying for different shows, you could just bring it all home to transistor and pay one price. So pricing decisions are crucial that you can't make that decision and that decision. We did multiple episodes of our podcast.

Justin Jackson:

We have a podcast called build your SaaS, where we wrestle with a lot of these things. And we did multiple episodes on pricing because it was like, this is a big decision, and we got to make sure we're thinking about everything. Like, what are the repercussions of multiple podcasts per account? What are the repercussions what do we price based on? What's our value metric? How does that interface with our costs, bandwidth, hosting, all that other stuff? Pricing is difficult. It's difficult to do it well. And it also has this tremendous opportunity because it can become a real feature of your product that people are seeking. And if you get the pricing part right, you can attract all sorts of people that you might not normally have attracted. But there's a tension, tension in all these ideas. You got your price, and then you got your costs, you got your margins, you've got.

Justin Jackson:

Are people even wanting that pricing? Is it a whole science and art unto itself? So this one, we just did our team retreat, and we were just like debating it a lot.

Jason Pearl:

This one's different for us because I think it's the first thing that has directly measurable variable costs. Everything else that we've built, it might need more system resources, it might need more storage, but those things sort of. It doesn't scale linearly with the usage. But this is like, if we do 10,000 hours of transcription, we have to pay for 10,000 hours of transcription. So it's a lot different way to have to think about it than we have in the past.

Demetrios:

Guys, I've taken up a ton of your time. I really appreciate you coming on here. I think we can cut it now. And for anybody that is.

Justin Jackson:

I'm just getting fired up now, man. Let's do another 2 hours.

Jason Pearl:

You cut Justin after he's had three coffees, right in the morning.

Demetrios:

Ready for more pricing conversation and AI and all the fun stuff. I really appreciate you all being part of the Deepgram startup community. I want to just publicly say that. And it is super cool to have you in there. And I look forward to all the incredible stuff that we can do together.

Justin Jackson:

Thanks, man. This is great.

Jason Pearl:

Us too.

Justin Jackson:

Thanks, lamb.