0:08 [music] All right, welcome back to the business of Laravel podcast where I talk
0:12 to business leaders who are working in and with Laravel. My guest today is Avid
0:17 Call, founder of Podscad, Podscan and I don't know, do you say the FM? Anyway,
0:21 founder of Podscan or Podcad.fm podcaster, writer, programmer,
0:26 bootstrapper, influencer guy, like all sorts of sorts of things. So Arva, just
0:29 kind of like tell us like who are you and what you about? How do you tell
0:31 people what you do? >> Yeah, that thanks so much man. I I could
0:35 go into like a 20 minute story right now. I usually do when I try to explain
0:39 who I am and what I do. But yeah, I'm just a software developer who's a little
0:42 bit entrepreneurial. >> And I I had the big fortune of of
0:47 building a software business back in in 2017, selling it in 2019 and then
0:51 getting to a point where I could just do whatever I wanted, which is when I
0:56 started diving into Laravel and into building web apps, more web apps because
0:59 you never stop building web apps even if you if you have an exit, at least me,
1:04 the kind of software founder that I am. So yeah, I'm I'm German originally. I
1:09 now live in Canada. I'm building, you know, software applications currently.
1:13 Podscan, like you said, not with the FM, I don't think. You know what? If we were
1:17 in the dot era, I would probably say it, >> right? [laughter]
1:19 >> Yeah. But it's just not a thing we do so much anymore, right? Like, yeah, your
1:22 domain is not a part of your identity. So, it's >> really not. Yeah. And I do have, and
1:26 this is one of the weirdest things about Podskin. I do have the com domain, yet
1:29 it still lives on the FM. >> That was that was one of the things I
1:32 wanted to ask you. >> That is pretty wild. And I I it's all
1:35 about the not wanting to do the migration. [laughter] That's
1:39 >> and I wondered the same thing. We we started as a co. We bought the com and
1:42 we're still doing a lot of things in the co cuz I was like I don't want to deal
1:44 with moving it all over. So >> that's the the weirdest forwarding I've
1:49 ever done is forwarding a.com to afm. But my domain rating at at this domain
1:53 is just growing and growing because I do a lot of programmatic SEO and a lot of
1:56 people are connecting, you know, their their whatever links to to my thing. So
2:01 that works for me and I don't think I'll ever change it because why would I? And
2:05 FM is great. It's a radio thing and it's about podcasts, you know, it's audio.
2:09 >> But yeah, so my my background is uh I'm an engineer, always have been. And um
2:13 now I'm kind of a soloreneur >> and I've been building stuff and growing
2:18 it and obviously also like talking about it in public. I do a lot of building in
2:21 public where I just share what I do on on Twitter. I think that's what I still
2:25 call it cuz I'm old. >> Twitter Twitter for life. will never not
2:28 be Twitter cuz I'm not going to not going to fall for that stuff. But uh
2:33 yeah, build I'm building pretty much anything I do in front of the people
2:37 that I think either it's for or that are like me. >> So that makes sense.
2:41 >> Okay. So I want to hear the whole backstory to what got you here and
2:44 everything like that. But I do want to ask the first question I'm supposed to
2:47 ask which I failed to do is what is your business and how is Laravel involved in
2:50 your business? So obviously there's lots of businesses but Podcan is kind of your
2:54 main thing right now. So what is Podcan and how is Laravel involved? So, Potscan
3:00 is my attempt at transcribing every single podcast episode that comes out in
3:04 the world as soon as possible and then making that transcription
3:09 accessible for search, for alerting, for, you know, data analysis or
3:14 whatever. So, it's effectively a podcast data platform that initially was ideated
3:21 as Google alerts for podcasts. That was that's that was kind of the theme that I
3:24 started with because I needed that for another project that I had
3:27 >> and it quickly turned into wait just have the world's largest transcript
3:32 database that has like full transcription with speakers and all that
3:35 for every single show out there. That must be possible, right? So that was
3:40 nobody's doing it. the competitors only track um podcast like metadata like
3:44 titles and descriptions because effectively the whole world of
3:47 podcasting and I'm already like deviating into this this whole space but
3:52 it's all built on the RSS technology right podcast you would think with
3:56 Spotify and Apple is like walt garden stuff but it really is just a walt
3:59 garden sitting on top of open source technology as it usually is right and we
4:03 know that well in the Laravel community as well but um the effectively we we
4:09 scrape all those those RSS feeds We we try to track all podcasts and then we
4:14 just ingest it using um OpenAI's whisper and and other comparable like models
4:18 transcription models to just turn audio into text and then there's there's a
4:22 whole whole lot of stuff going on lots of AI work as well to analyze and you
4:27 know take this data and a big old database and an API and that's pretty
4:32 much what this is. So POSCAN is both a thing for builders. I have a lot of lot
4:36 of the customers of this project are other founders who built on top of
4:40 podcast data right for advertising reasons for um the PR reasons like if
4:46 there's a an a conversation where a brand is kind of currently experiencing
4:50 a lot of trouble that needs to be dealt with right so people need to know about
4:53 it and need to handle it so that's that's the builder side and obviously
4:56 marketing departments PR departments use it just to track things to search for
5:00 things to do research and all that kind of stuff as well it's a pretty gigantic
5:03 project and I'm effective ly running it alone. And >> so that was one of my questions because
5:08 I just recently signed up for Podscan and it's it's impressive. It's very
5:12 expansive. There's a lot going on. It's very easy to understand. And I was going
5:15 to ask, do you have designers on board? Do you have other developers or is it
5:17 mainly you? >> So when it comes to development, it's it's all me and Claude at this point.
5:24 >> Yeah. You know, it's it's always an AI involved at some point. But when it
5:28 comes to design, um I had a lot of help from uh Nick Grerfeld. He's uh is a guy
5:33 who has been designing a lot of my projects. He's been been designing my
5:36 YouTube thumbnails for a while and you know just he's a UX designer and
5:40 somebody who really knows how to do this kind of stuff. So he helps me a lot with
5:42 that too. >> Yet most of it is just me trying stuff out and sticking with the things that
5:49 that work, right? That's that's pretty much what that is. But it's a mostly
5:54 solo run project. I'm trying to expand it a little bit but not on the dev side
5:57 at all. Like I've never hired a developer for this because it's you know
6:01 you don't ever say anything is feature complete but it is feature complete
6:06 enough not to warrant like massive resources in terms of development but
6:09 what it's not complete is customer complete. So you know sales and
6:13 marketing yeah >> do you find it uh what's the experience
6:18 like as a as a developer who's also doing sales and marketing? I feel like a
6:22 lot of us are just kind of like I can build it and if they come we're great.
6:25 But the idea of getting the people to come is a very difficult kind of
6:27 transition. >> I had to fight a lot of my like preconceived notions when it comes to
6:35 what marketing and sales is like I most developers and not not sure if that is
6:39 just my experience and one of my peers or like a generalizable one but most
6:44 developers are kind of siloed away from the marketing and sales departments of
6:47 the companies that they work for or even education. When I went to university for
6:52 computer science back a couple of decades ago, I I was everybody was kind
6:57 of talking the business people down and they were talking like as as if again if
7:03 you build it they will come. We know that's kind of a warning, not a
7:07 guarantee, right? So yet it was still perceived as if a product is good
7:11 enough, marketing is going to be so easy the people who then do the marketing
7:15 does don't have to be as smart as we are building the project. That was kind of
7:20 built into my my education as a developer, which is horrible. And I had
7:24 to do a lot of deprogramming to understand, no, no, no, no. This is
7:30 actually a a completely different yet comparably relevant part of the journey that you
7:35 are on as a founder, right? As a developer, you probably will never have
7:38 to touch marketing if you don't want to. But if you are a software founder,
7:41 >> this is going to be the bread and butter of how people get to know your product,
7:45 how they get exposed to it, right? And you could do it in so many ways.
7:48 Building a public one of them. If you have audience overlap or if you have
7:51 potential network effects where people can tell their peers that they know
7:55 about your product and you get get it in that way, that works too. But sometimes
7:59 you just have to do paid advertising or you have to do content marketing. You
8:02 have to write blog posts and it's going to be annoying because you don't want
8:06 to, right? And developers never want to make stuff that doesn't immediately show
8:10 that it's working. At least that's what my experience as well because we're so
8:13 binary, right? Things work, they don't work. But some things they work two
8:17 months from now, two years from now, but we have to do them today. It's it's been
8:20 a hard journey for me to understand that this is relevant and I needed a lot of
8:25 help along the way. Obviously, a lot of people in our community are really good
8:29 at this like in the Laravel community in in particular. There there are a couple
8:33 couple names and probably get get to them in the conversation as well that
8:37 just people when people show up and talk about their thing you see how much
8:42 resonance that makes in the community and in in their potential like group of
8:47 prospective customers and if you get inspired by that you're already way
8:50 further than most people who just think this is crap I don't want to do it. So
8:53 >> yeah. >> Well, and so one of the things that I
8:56 think that folks in the Larvo community who've been super successful have had
8:59 going for them is that they are building for the community that they already have
9:03 a reputation in. And so you know, you mentioned the idea of building in
9:05 public. You either want to be building for people like you or for people who
9:09 need the thing you're building. Were you already in the podcast world? Did you
9:13 already have a name and a reputation and connections there when you started
9:15 building Podcan or did you have to intentionally insert yourself in that
9:19 space as you were building it? It's it's a funny one because the POTS scan is
9:23 lives at the intersection of the world of podcasting which is its own industry
9:28 and the world of people who have now started to become aware of podcasting as
9:31 a potential target for their marketing or research efforts. So it's a
9:36 completely disjoint yet also slowly overlapping group like the ven diagram
9:40 is moving in into each other right those groups. So, it's been um it's not been
9:45 as hard to go into the podcasting community because a lot of the people
9:48 building podcasting products that I can integrate with, they are my founder
9:52 peers, right? The the transistor of M people, the the Riverside people, like
9:56 any any of these these tech companies that facilitate podcasting, [snorts]
10:01 hosting, creating, distribution, any of this, I already had kind of an in with
10:05 them being a little bit of a vocal person in the software community, right?
10:17 hard for me and the marketing folks because like a a medtech company, some some kind of
10:24 health company and getting into their PR department for somebody who's never had
10:28 any contact with this world. That has been the hard part and that is the part
10:32 that I struggle with the most. It's easy to sell to founders if your peers and
10:35 your friends are founders. >> Easy. It's never easy but it's easier to
10:38 do this. It's an inroad >> but to people who actually care about
10:43 LinkedIn you know [laughter] like people who care about reputation and who care
10:46 about the status games and I don't want to belittle that it's just part of the
10:48 game that they play >> different world yeah >> to to get into that mindset and to
10:54 fulfill the perceived need that they have >> where but when I talk to a founder it's
10:59 like do you have an API that has this and I'm like yes I do okay need
11:04 fulfilled right but for somebody in a in a PR department it's like yeah I need to
11:08 be able to to report this to my my higherups within 20 minutes of X or
11:13 else, you know, like that is their need and to to sell to them. So different.
11:19 >> So podcasting and founder world was easy, but the other side of things,
11:23 that's the hard part. And I guess that's always the hard part with SAS, the
11:26 moment you get out of your little early adopter bubble. Like it's so easy to
11:30 sell to early adopters because that's literally what they like, buy new things
11:34 and try them out. But for people beyond the chasm, that's a different story.
11:39 >> And we talk often like at least at least in the bootstrap founders world, we say
11:43 a lot of people will be like, "Oh, I saw a need for realtors, so I'm going to
11:45 build an app for realtors." And then we have to ask the question of do you
11:48 actually like realtors? Do you do you like the idea of going to a real realtor
11:51 conference and think about the things they value, you know, and so for you,
11:53 you're in this kind of world of marketing and advertising and whether or
11:57 not you, you know, like entering that world, um, you're clearly showing that
12:00 you have the ability to empathize with their needs, right? To figure that out.
12:04 What what are some things you've had to do that maybe you didn't expect? Maybe
12:07 spending time on LinkedIn, maybe doing paid ads, but there are any other
12:10 things. Do you go to conferences? Like how do you even learn about that world
12:12 and get connected there? >> So, I've I've never done a demo call in
12:17 my life before I started POTSC. That's that's maybe the thing like showing
12:20 somebody how to use software was so alien to me because I've always sold to
12:23 people who knew knew how to use software. >> Yeah. >> Right. To me, yeah, this is
12:27 self-explanatory. This button says export data. Obviously, this is going to
12:30 export a CSV file. Like, [laughter] sure, right? Why would you need another
12:33 dialogue and click and select what you want? You could just do this right in
12:37 your Excel or whatever. So, selling to nontechnical people often requires
12:41 building this this foundation layer of trust that has nothing to do with
12:45 software. And that was new to me because a software business to me is you sell
12:49 access to software. People want software, people talk about software,
12:53 people want their software challenges met and that's it. But in a in a world
12:58 of less technical and more um organizational data driven I guess um is
13:02 what these people are or like result driven you have to care about the job to
13:06 be done more than the tools that it's done with. So that that was a refocusing
13:11 effort for me to go away from no I can't just add another dialogue into my thing
13:15 and expect them to find it. I have to to figure out what their process is what
13:19 their actual input data data transformation output data process is.
13:23 Learn about the input data that they get that they have to work with. Learn about
13:26 the output data that is expected of them that they need to deliver and see how
13:30 they currently solve this problem. Typical jobs to be done methodology and
13:35 try to map as much of my internal process that the the software tool does
13:39 right onto their process while being adaptable, customizable and flexible.
13:42 >> No big deal. >> Super hard. Super hard with an industry
13:45 that you don't understand. That's the thing when when you said this just now
13:49 like when you don't know how real estate works like or when you don't like real
13:53 estate people like I I would I I don't think I would ever build for real estate
13:56 people not because I don't like them but I certainly know that there's so much
14:02 detail and intricacy to that world and so much like human factor in there too
14:07 right where trust and and reliability and connections and you know pushing a
14:11 thing there so you get something back from them later that is super hard to
14:16 map on software like software is a binary thing but human relationships are
14:20 constantly evolving. >> I would stay away from that market.
14:24 Plus, you know, they don't have the best reputation. That's the other side of
14:26 things. But >> it's it's mapping your your [snorts] your technical process or your your
14:33 vision of how to solve this technically into the messy realworld version of that
14:43 in all of these things that you only get to see when people say no this is not
14:47 for me and then you have to dig and have to figure out well why is this not for
14:52 you right what in there is the thing that um that is that is not working for
14:57 you here that I thought was working for you but it's not right like what is the
15:00 detail that I'm missing do you even know is it an unknown unknown it just feels
15:04 like it's not there super complicated >> h you're making me think of Michelle
15:09 Hansen's um deploy empathy and kind of for those who don't know she's a a big
15:13 advocate of um interviewing your customers and kind of understanding them
15:15 better and it's kind of weird that someone needs to be an advocate of that
15:19 like of course we do that right but I I really appreciate what you're saying
15:22 about like the the level of assumption that we have that working with technical
15:26 users they're going to know the same things we do about how to use software
15:30 and how software should be organized and so I'm sitting here thinking that your
15:33 work is just to find these people and convince them to use their app and
15:36 you're like that's half the game that I have to make it actually work for them
15:40 in the way that their brains think that is different from my way of thinking.
15:43 >> Yeah, they have to they have to see that that that's also a thing like adoption
15:47 of tools like this like pots or any tool that is an expert tool for another group
15:52 of people, another kind of industry. People have to see that you get them
15:56 >> in your product and that's incredibly hard to communicate if you're not from
16:01 that world. lingo like the phrasing right to me I never understood certain
16:06 things to mean a certain thing but there's a lot of nuance between a list
16:10 and a collection for people who use this all the time who use list building or
16:14 who want to collect data right it's it's this is a lame example but like these
16:18 tiny terms they they make up a significant part of other people's
16:22 professional lives and if you get one of them wrong you're out because you're not
16:24 one of them. >> Yeah. So that AI helps with that a lot
16:30 cuz I can I can tell my my cloud code or whatever I use to just look at this UI
16:36 from the perspective of a professional XYZ and then tell me what they're what
16:41 the tools that they're currently using, what what phrases they use, what terms
16:45 they use, and if there is a a kind of a miscommunication here or not. Like one
16:48 of one of the tools that I use both in my writing and in my my software, my
16:52 code writing, well, don't write code anymore. I just say yes, yes, this is
16:56 fine. But let's let's get to that at a different point. Um, one of the tools
16:59 that I use is kind of the the devil's advocate version of uh AI usage. I just
17:05 show it a thing and I tell it if you are this person that I want to like this,
17:10 >> what would you say is missing here? What would you say is not working? This works
17:13 really well with blog posts, with articles. Like if you're somebody who's
17:18 an who's an expert nerd in the industry and you read this, what's your your
17:21 attack vector? How would you say this is a crappy article? And then you fix it.
17:25 Same goes for code. Same goes for interface. Like if you're an expert
17:28 here, what would you say is missing here? It's really really useful. And it
17:32 it means you don't have to have your AI write a single line of code. It can just
17:35 look at it and tell you something. So if you're an a hardcore artistal coder, you
17:40 you might still find use for AI tools this way. >> And that's very valuable because I'm I
17:46 I'm constantly asked by prospective clients, how do you guys use, you know,
17:51 AI? and constantly asked on podcasts and stuff, do you think AI is going to take
17:56 all our jobs away? And I tend to be more conservative when it comes to use in AI
18:00 in generating code um and also in generating content. But one of the
18:04 things I'm seeing as a trend is people talking about regardless of what you
18:07 think about from a generation perspective, from a review perspective,
18:11 it is helpful to do a first round of hey, you might have missed this or if I
18:13 were a newcomer, I wouldn't understand this or uh you know the rest of your
18:17 application has this particular type of code in place and this particular piece
18:20 doesn't. And to me that's very interesting because a it preserves the
18:25 humanity and creation and b I think it fits better with what LLMs are capable
18:30 of doing which is saying I'm looking at your code as a pattern and I'm looking
18:34 at this other code as a pattern or I'm looking at this writing as a pattern.
18:36 I'm looking the writing in your competitor app as a pattern whatever and
18:38 I'm just telling you how these patterns do or don't match. I'm like that that
18:44 puts AI more in useful tool category which I I love and out of magician
18:48 that's going to completely transform your life category which I just don't
18:51 think is the right way to think about it. So that's I really appreciate you
18:53 bringing that up. >> Yeah, it's it's it's not taking your job
18:56 anymore if it makes you way more effective at your job. >> But you still are required to drive it.
19:02 It's it's a pair programming situation and and >> I think that to me is how we're going to
19:07 be using AI going forward like in in a in a normal capacity like not as like an
19:11 expert tool to do the things for us but as a a constant kind of wise mentor
19:15 sitting beside us helping us with everything and that is not just for
19:19 coding that is for cooking that is for for any activity that you might need
19:23 some expert insight on you can always ask right these tools are effectively
19:28 free comparatively right in terms of what they cost So having somebody like
19:33 for let's stay with coding here like whenever you have whenever you commit
19:36 whenever you push your code to your to your main branch or whatever why not run
19:42 an AI security sweep right before it right why not have like a a prompt that
19:46 says check whatever I've changed here for glaring security holes that cost you
19:52 30 seconds but it will spot like yesterday I had this like I I was
19:55 deploying not two days ago let's be honest two days ago doesn't matter right
19:59 but you know I had a I had a situation where I had built something where I have
20:03 a search bar that I wanted to put on my homepage that it directly integrates
20:08 with Podscan search which is sizable. I have like 45 million podcast episodes
20:12 somewhere in an open search cluster that full transcript that are searched. So it
20:16 costs a lot to do a search. So I wanted to build a system around this that um is
20:22 capable of preventing abuse or at least making sure that people can't overuse
20:26 it. Right? So, I had the caching system and I was counting how many searches
20:30 coming in per session and all of that. I had all of that built. It was great. And
20:33 then just before I committed it or before I pushed it, I was like, "Okay,
20:36 let's do another security sweep." And it and I checked and it said, "Well, your
20:40 cache timeout is 0 seconds >> because I set it to zero for testing
20:45 locally so I wouldn't run into my own limits, but I hadn't set it back." So,
20:49 it was saying, "This is great. you have a cache and all that, but if somebody
20:52 were to hammer your system, it would take one client to start like a thousand
20:56 searches at the same time. You might want to change it back. And I was like,
21:00 phew. Because that was my doing, right? I programmed this yet, it found my own
21:06 forgetting to set it back to an hour or a day or so. This is so valuable and you
21:10 can completely automate it away. It's quite useful. Well, and it's interesting
21:14 because historically I would and I think this is still the case. I would expect
21:17 that a policy that we have at Titan, which is that every line of code needs
21:20 to be seen by another person. So, normally that happens through code
21:22 review, occasionally through pair programming. So, we catch those by
21:26 saying, "Hey, every person should review their code again in a different user
21:28 interface, usually the GitHub pull request interface, and then also have
21:32 somebody else see it." Um, but to your point, you know, you're a solo
21:36 developer. You don't have that ability to have, hey, you know, have your other
21:38 programmer look at it. Well, what if it's just me? Uh, that's your other
21:41 programmer. >> That's right. That and that is my other
21:44 everything. That is my marketing helper. That is my my cold email like
21:51 proofreader. That is my um blog article fact checker, you know, with research
21:54 mode or whatever. Like anything that I need a second, third, 100th pair of eyes
21:59 on, I have some agentic tool to help me with it. Like this has been the boon for
22:04 bootstrappers. like AI agentic systems are if you overuse them, they vipe code
22:09 you into oblivion obviously, but if you if you use them just right, right, the
22:14 Goldilocks AI zone, then you have some extremely capable not like not not an a
22:19 genius because these things are still guessing and and what do I call them?
22:25 Um, they are gaslighting engines, right? That's that's kind of what these things
22:30 are. But if you pre-prompt them well, if you tell them not to gaslight you is
22:33 kind of the fix this, don't make mistakes kind of prompt, but you know,
22:37 like if you tell them to be as critical and as like aggressively contrarian as
22:42 possible, then you don't have your sickopantic machine that those things
22:45 tend to be. >> And then they are extremely helpful because they will point out the flaws
22:51 that a person that you pay to point them out would probably not point out because
22:55 even though you pay them, they still want to keep getting paid. So you have
23:00 your your 8020 there. So it's um it's quite helpful to have this extremely
23:04 honest when correctly configured thing helping you do things that otherwise you
23:09 would have to pay people a lot of money. >> So I know that you create an incredible
23:12 amount of content. You know you've created courses about how to use Twitter
23:15 well. You've created courses about you know how I I remember you had a podcast
23:19 episode recently about how you use AI like throughout your whole business.
23:23 >> If someone is hearing let's just kind of wrap up the AI part. If someone is
23:27 hearing your AI approach, is there one resource? Would it be that that podcast
23:30 episode or is there somewhere else where like if I want to learn about how Avid
23:34 uses AI, this is the place to go? >> I think my podcast generally is a good
23:37 resource of the things that I'm doing because I just yap about whatever I'm
23:41 doing at any given moment every week. Yeah. Like that's that's kind of what
23:44 this is for, right? I try to put it into a topic, kind of a sensible like,
23:49 >> you know, scoped thing. But yeah, that episode if we can can put that in the
23:52 show notes or something would probably be very helpful because it's a good
23:55 starting point, right? You don't need to do everything, but you can do certain
23:59 things and they're very very easy if you do. >> Yeah, I appreciate that. So, one of the
24:03 things that we're often trying to focus on as podcast is figure out um why do
24:08 people choose to use Laravel and then if somebody were to want to get to where
24:11 you are, what would it look like for them to get there? So, those are kind of
24:14 like the two angles. So, let's let's focus on the Laravel for a bit and then
24:17 we'll go to that like how people can get where you are. you your first big hit uh
24:21 you you said you know it happened you sold it and then you learned Laravel
24:25 what motivated you to get into Laravel at that point >> it's it's pretty funny because my my
24:30 programming story well it began with Tobo Pascal let's be real like that's
24:32 that's >> let's go the whole way back there let's do it
24:36 >> that's like school time coding that that we actually learned that in in high
24:40 school we had a like a 50some year old teacher who just understood enough to
24:44 pascal to teach it to us and obviously we outpaced them that the people cared
24:50 about within hours of picking it up. But my my actual real first programming job
24:56 was in PHP back in in 2001 or so. So we're talking PHP4. We're talking typo
25:01 3. I was writing typo 3 back uh backend extensions. That was as horrible as it
25:06 gets back in the day. It it was not juicy. It was not like flashy. It was
25:10 just building. We didn't even have git back in the day. We were building on top
25:14 of not even SVN. We were building on top of CVS. That's how far back I
25:17 >> was going to say it was at CVS. Oh my gosh. >> We were not FTPing into our server
25:22 anymore, at least not all the time, but we were doing that. So that's kind of my
25:26 my first foray into programming was uh web programming at least was PHP in an
25:29 older version. >> And then I I took the the JavaScript
25:33 train like so many people did, which funny enough led to my first like uh
25:38 Silicon Valley job, which was great. Like I it coffee script got me there, so
25:43 you know, can't complain about that, right? Yeah. Back in the day. That's
25:46 that's have not heard that word at all. >> 2012. Yeah, that's that's pre-angular.
25:50 That's that's pre-react and all of that. It was good times. >> And um at some point um I ran into a
25:59 company that wanted to build an IoT platform and they were looking at
26:04 Elixir. They were looking at the Elixir language which is kind of sitting on top
26:08 of the Erling VM. So it's it's it's the Erling language but kind of better with
26:12 some Rubyesque syntax. is still functional programming like the complete
26:16 it's the complete opposite of PHP right it it is compiled it runs on top of a VM
26:21 it's highly functional it has a rather small community it uh it you have to
26:27 spawn a process to run it and all of that I mean we do that too with FBM but
26:29 doesn't matter right it's it's completely the opposite of how we deploy
26:33 PHP and it was great it helped me a lot it uh it helped me to understand
26:36 functional programming which is really interesting like data transformation at
26:41 the core of of coding is is really fun to to always think in in how to
26:44 transform data and and just shift it around. That's cool. So, I worked for
26:49 that company for a bit. That's like 2015 or so, couple years. And then I started
26:53 building a side project which was Feedback Panda. And Feedback Panda was
26:57 the business that I ultimately sold for for a couple million dollars. That's my,
27:00 you know, my my big exit that I had. And that was built in Elixir Phoenix as
27:05 well, right? Fenix being so it it was built in that because that was the
27:08 language that I was building in at the time. And I was like, hey, I do this for
27:11 work. might as well moonlight with the exact same toolkit because I know
27:15 exactly how to get things built. And I built that and I I deployed it and I got
27:20 horrible uh like problems like finding a good host to run this on and all of
27:24 that. But figured all of this out. Uh got several uh thousand customers. We
27:29 got to 50ome thousand MR with that business. >> And then we sold it two years later to a
27:33 private equity company and had our big exit and life was great. Like that's
27:37 pretty much that story. It was >> fantastic. And and so once that happened
27:41 I I fell into the the devoid of not knowing what to do which is something
27:45 that happens to a lot of founders and I started writing the the boots founder
27:48 which is my blog and my podcast and all that stuff started happening at that
27:51 point. I just started sharing my memories my experience from before and
27:55 built on top of that just you know every week I wanted to uh release something
27:58 tell people about something that I'm either thinking about now or thought
28:01 about in the past. It's kind of how the podcast and the the blog and the
28:04 newsletter started. It's a media business journey, I guess. Um, which is
28:10 >> by far by far now my most profitable endeavor because of sponsorships and
28:13 that kind of stuff. Funny enough, really, right? Podscan is not as
28:16 profitable yet as that other thing, but that other thing is also since 2019. So,
28:21 you know, that's six years in the making. >> Um, and and that's the first time I got
28:26 into contact >> with the Laravel community must have been around yeah 2019 2020 on Twitter,
28:33 right? I was starting to build a Twitter presence and a Twitter audience and I
28:38 started seeing obviously people like Taylor and other people in the community
28:44 just talking about this PHP thing that I had completely forgotten about, right?
28:47 Because once you're in the JavaScript ecosystem, everything is solved with
28:50 JavaScript. JavaScript on the front end, JavaScript in the back end, JavaScript
28:53 on mobile or whatever. Everything JavaScript is the greatest language
28:56 ever, right? For for people in that community. And I I think for them it
29:00 probably is. Yet, you know, if you take a step back, it's this weird thing where
29:03 you set up your setup tool so you can set up a setup tool to then deploy your
29:06 [laughter] code somewhere maybe. Right? It's it's it's complicated and Laravel
29:11 is not complicated. I I figured that out immediately. I I so I was looking for a
29:15 new project to to build to get into Laravel because I really wanted to.
29:20 >> So, I built a thing called podline.fm, which is not podcan.fm, FM, but it is a
29:25 I I wanted kind of a voice recording tool to allow people to call in and, you
29:30 know, message me and I could play that on my podcast cuz I love the the Seth
29:34 Seth Goden Akimbo podcast where he has a lot of people calling in and asking
29:37 questions and he he goes through them and answers them. And I wanted something
29:40 like this, but I looked at the tool that he was using and it looked like it was
29:43 made from the '90s, like the in the '90s, just from the from the tech. And I
29:47 was like, I can do this as well and I can probably do it better. So I built it
29:52 in like two weeks, you know, people tech tech people, you cannot stop us from
29:56 rebuilding software that's already there and I deployed it and it I had a couple
30:00 people trying it out. But I wanted to find more people >> and I was like where do I find all of
30:05 these these people that I can sell this to? Well, probably on podcasts where
30:09 people are already asking for send us a voice message or you know leave us a
30:14 message at this number or you whatever. We use the what was it called?
30:19 um you pipe voice pipe or something speak pipe was the name of the
30:22 competitor and if somebody were to mention speakpipe I would know that I
30:26 could reach out to them and say we have something better >> but there was no such scanning tool for
30:31 podcasts so that's what pod pods scan was so now I had built a laravel tool
30:35 with pod line and I already had transcription in there because I took
30:39 messages in transcribed them real quick on the CPU and then sent that as an
30:43 email to the the owner of the podcast so I just built another Laravel application
30:49 that would try and fetch all of these podcasts from their RSS feeds, download
30:54 the MP3, transcribe them, and then save that to a database, make it searchable,
30:57 make it alertable. >> And what came out of it was um a single
31:02 instance web server, which is still my main application, some some AWS EC2
31:08 instance that runs bots.fm, FM and a fleet of Hetso servers, like probably 20
31:13 or so that are also Laravel applications, but all they do is
31:16 communicate with my main server back and forth to fetch URLs, transcribe them
31:20 locally because they're GPU enabled servers on Hetner. They cost me like 200
31:26 something bucks a month or€180 or something, which is almost nothing
31:30 for a GPU that you can run a month long. They not they do 247 they transcribe and
31:36 um analyze like with a local LLM and sometimes if it's too complicated they
31:40 go to OpenAI and back >> and they they they get a URL and they
31:44 send a transcript back to my main server and they fetch a new URL send it's a
31:48 message cue right so that's that's what potskin is it's one big Larwell
31:51 application that handles all the you know the the logic the user interfaces
31:55 those components and you know the recurring stuff and a fleet of servers
32:00 all deployed on Larable Forge that uh or through Laravel Forge I guess that's
32:04 handle all of the >> transcription and that kind of stuff.
32:08 This is prelear cloud. This is all still kind of self self-managed but works
32:12 pretty well and is extremely cheap to run. Like the hardest and most expensive
32:17 part to get right was the database because this is now a six terbte
32:22 >> database, right? This is all information. Sure. >> What was getting that right? What did it
32:25 look like? What are you even using for your for your full text search? get
32:30 getting that right was orchestrating um a kind of a split in the data. So I have
32:36 I have all the original text saved in this main MySQL database on Amazon AWS
32:43 like RDS which is extremely expensive. This is by far the largest part of my
32:47 expenses. Also the reason why it took me so long to get to profitability if you
32:51 pay $5,000 a month for your one database, right? That's just that's just
32:56 what that is. So that's where I'm at. And um the search the full text search
33:00 is an open search cluster also running on AWS that is kind of bifurcated that
33:04 is split like I I funnel all my transcripts in there. I don't don't have
33:08 all the information in my open search cluster like all the the metadata. I
33:11 don't need that. I just need the transcript the title and the description
33:15 and and these these get synchronized all the time. Um what do I use for that?
33:20 Scout. I I use again Laravel has everything. That's the great thing about
33:23 this framework is yeah just scout. There's an an open search library for
33:29 scout that adds another provider, right? First, initially I used Miley search a
33:33 lot, but the the data the data set just outgrew the capacity of the ingestion of
33:39 my Miley search cluster. Not the search search was super fast. But if you throw
33:44 50,000 1 megabyte transcripts into a Miley search cluster every single day,
33:49 >> it just gets stuck trying to ingest them. At least at the version that I did
33:52 back in the day. I think I've been working with the Miley search people to
33:55 make it better and more feasible and some parts of my my product still use
33:59 Miley search as well but >> I have an open search cluster because
34:03 you know elastic search open search is just what you use for search if you have
34:06 massive data sets like this. >> Yeah. >> Got it.
34:10 >> That's that's that's my my Laravel uh >> that's a Laravel back story.
34:12 >> That's really what that is. >> So one of the things that often comes
34:16 out of this podcast is somebody hears somebody and they're like, "Oh my gosh,
34:18 that's the life I want to be living." And I mean, even personally, like I
34:23 would love for Titan to be 40% product and I, you know, I'm just still working
34:27 on that sales and marketing hook that you're talking about, you know, like I
34:29 can build a product, but I'm still working on that. >> So, if someone were to want to at some
34:35 point in the future be where you are, uh, if they want to be a solo bootstrap
34:39 founder, if they want to kind of make their own thing, can we talk about
34:42 influences that got you here? And I'm talking books, podcasts, YouTubetubes,
34:46 experiences, just like what got you to where you are today that somebody else
34:48 could kind of say, "Yeah, okay. I'm going to do that." >> My the biggest thing, name, book, or
34:53 whatever that comes to mind is is a person, and that's Rob Walling, like of
34:57 the Startups for the Rest of Us podcast. I I've been to a lot of micro comps and
35:02 I've chatted with him a lot and I've read his books and I've listened to his
35:07 show um his shows, right, and his presence on other shows, the the the
35:11 micro community and all of that. everything he's been building has been
35:14 extremely useful for me and people like me. So like that is a person that I
35:20 would unequivocally follow and purchase every single piece of product that he
35:24 releases from because he is kind of one of us from from the very first hour like
35:29 he's been building up the terms that we now use in the in the solo founder
35:33 bootstrapper community through his work and his presence in the field. So he
35:37 recently recently a couple couple months ago year ago or so released uh another
35:44 book like his his third or something >> forgot the name because I'm really
35:46 horrible with titles and names which is funny for somebody who works with
35:50 podcast names and titles every day but um the the most recent book any book by
35:54 Rob Bowling is something that I have like very very close to whatever I
36:01 needed right um that that his work is paramount to consume when you're
36:05 building a software business just to get >> Is this the SAS playbook that you're
36:07 talking about? >> That's the one. Yeah, the SAS playbook.
36:10 Even though that is actually for like later stage, right? Start small, stay
36:14 small is is also one I I highly recommend. But the actual that he has is
36:18 the podcast. >> Like follow along with the startups for
36:22 for the rest of us podcast and you will not forget things along the way because
36:27 he brings them up reliably every every year around you. You'll find refresh
36:31 renewed versions of this stuff coming into into your ears. highly recommend
36:34 it. >> People don't answer that often enough, but and that's always my answer, too.
36:39 And even folks at Titan where they're like, you know what, like I know how to
36:41 build an app, but I want to become more of a product person. I'm like, I learned
36:45 what I know about product by listening to startups rest of us religiously for
36:49 years. So, I'm like, just go do that. I don't know if you should start from
36:51 today or you should start from back then, but like just go listen to that
36:54 podcast. I have not read the books, though. So, you're you put something on
36:56 my reading list. >> Yeah. Yeah. I mean, the the great thing
37:00 about the the podcast too, if you start like with episode one, which is like
37:03 what is this now? like 12, 15 years ago. It's been a while. You also see somebody
37:09 building a media business from a pretty crappy microphone to what it is now,
37:13 right? It's it's also inspirational if you are in your own media journey. If
37:16 you want to do your own podcast, if you want to do your own writing, your own
37:19 authorship, your own online presence, you're on the influencer situation, even
37:24 though we we don't like to be called influencers uh in our community, but you
37:29 >> people are influential if they have a sizable audience. That's that's what it
37:32 is. But if you want to see that journey happen in front of you, you you look
37:35 into like how these podcasts started. Like same for me, right? That my first
37:39 episode sounds like I recorded it through a potato. But it it still is
37:44 valuable, useful, and and a story that to this day resonates with people. So
37:49 it's yeah, that's that's something I recommend. Other other book
37:51 recommendation that I'm just going to throw that in because it's so such a
37:55 controversial pick, I guess, is um Hooked the by Nal. Um it's how to build
38:02 habit forming products. That's the subtitle of the book. And Nal is um it's
38:06 it's an interesting story because he talks about like how to get people
38:10 hooked into a product. And if you look at the reality that we live in, all the
38:15 gamification and all the the triggers and the variable rewards that we get,
38:20 all this this uh loot boxes and people like opening, you know, mystery boxes
38:23 and stuff. Yeah, >> that is the underlying principle of how
38:28 people gamble all the time and how people come back to things that are
38:31 maybe good for them, maybe hurtful to them. So read this book with the
38:35 perspective that what you learn in there is either going to be very helpful for
38:40 your product or very hurtful potentially if abused for large parts of society.
38:43 That's why it's a >> be both helpful for your product and
38:47 hurtful for large swats of society. >> And that's the thing it's it's not
38:50 exclusive. So it's it's important though to understand how this works because
38:54 this also helps you step out of it which is funny because Niel wrote a second
38:57 book which is called what is it uninterruptible or something which is
39:01 how to step out of the hooked cycle. So you know it's a it's a pretty funny one
39:04 but that's also a recommendation that I would like to drop because it's about
39:08 product and about consumer psychology and we don't think about that enough as
39:12 as tech founders. Well, it's fun that I I just got to on board into your
39:16 application just very recently and one of the things that you did really well
39:21 was recognize that during your trial um you know you really want to get them
39:24 hooked. You want to get your end user to say during this trial I have found the
39:28 value and so for example you know I wasn't getting any alerts and I get a I
39:32 think automated message although it might have been you personally but
39:34 regardless I get a message from the system that says hey it's been a couple
39:37 days you know we're only 3 days into your 7-day trial you haven't got any
39:41 alerts here's something you could do to potentially expand your alerts to be
39:44 less narrowly focused so you can actually see some content because you
39:48 know that if I see my name popping up or Titan's name popping or whatever I'm
39:52 like oh this is a valuable tool and if I spend in seven days and there's no
39:56 alerts. I'm going be like, you know, so I I thought that was clever. I
39:59 definitely could tell that you paid attention to it. It didn't feel like it
40:03 was a manipulative thing. I I felt like you're like, if you're doing the trial,
40:06 you want to get value out of this tool. Here's ways to get more value. So, it's
40:10 helpful for me to think about like an an ethical way of trying to make sure that
40:15 people are able to get value out of this tool. >> Cool. Well, let me pull back the veil
40:19 for like a a second here because that is both a manual and an automated thing at
40:24 the same time. So, I I have built a system that just lists everybody who
40:27 comes into my product and what they have done, right? Have they set up an alert?
40:32 Did that alert find any mentions? Like, have they played with the API? Have they
40:35 turned this on, that on? And I have all of this data available because I track
40:38 it internally using like the activity log. Uh I think that's the name of of
40:42 that, you know, library or something. I just every activity in the system gets a
40:47 log. So on that on that list I have the the name of the person email and when
40:51 they joined and all of that and all the activities could condensed into a couple
40:55 numbers and then I have a button that says generate email. So the there's a
40:59 prompt that pulls all the activities you have and have not done into plus plus
41:05 your email address plus your name and the the thing that we think you do into
41:12 a a chat GPT or an GPT4 something prompt comes back with an email which I then
41:17 manually read through and fix. Like some of the things I put into your email were
41:20 from OpenAI and some were from mine cuz I knew about you, right? some somewhere
41:24 just me thinking that would be a helpful thing and then I I send that email out
41:28 from from the system at that point. So we we do we scan the domain that you uh
41:32 that you sign up with figure out what is on that page and give you good
41:36 suggestion from that through AI. That's one of the other useful uses of AI is to
41:40 do some background research on your customers and suggest good things to
41:44 them like from from what you think they might do and might >> well and I love this idea of you using
41:50 AI to augment your personal processes. Like it's the same as you said at the
41:53 beginning. Beginning it was you know I'm writing the code but I'm getting AI to
41:56 review it. I'm writing this but this blog post and I'm getting to review it.
41:59 And now you're saying, I want to write these sort of emails, but it would be
42:02 prohibitive for me to write these sort of emails to the number of clientele I
42:07 have unless I have a nonAI automated tool making the list, pushing content
42:11 into an AI tool, generating the email, and then I, a human being, I'm now
42:15 augmented in my ability to email all these people by the fact that my
42:19 automated a non AI and nonI systems get me something that I can act on. I'm
42:23 like, I'm doing the same with VAS, you I'm just like, I can't do all the things
42:27 I need to do, but if the VA does 80% of the work or my business development
42:30 manager does 90% of the work, I can do the last 10%. So, I'm just like, yeah,
42:33 >> I get it. >> It's kind of that that's the thing,
42:36 right? The last 10% and pretty much the first 10% as well, right? Somebody has
42:39 to start and and figure out what where's the priority? What should they do? And
42:43 then they do it. And then you kind of do the review and condensation and like
42:46 figure out if it's right or not. And I think that's what AI smart AI usage is
42:49 about. like the first 10 and the last 10% are yours input and output really
42:54 like confirmation and prompting or whatever and the the in the middle the
42:59 agent be that a human being a virtual agent or a real agent a fake agent you
43:03 know made up agent whatever that might be that that work can be done in the
43:08 middle it's like a 1080 10 kind of rule I think that that generally that that
43:11 worked before AI right that's how we used executive and and virtual
43:15 assistants forever but now we can map that onto the artificial process as well
43:17 I think that makes sense >> that makes a ton of sense I told you this at the beginning and
43:21 what I told you was we're going to get to the end of 45 minutes and I'm going
43:24 to say we have a million other things to talk about and that's where we are. But
43:27 there is one question I always end everything with and it is most fun to
43:30 ask this question to people who've had a successful exit before.
43:33 >> The question is if somebody were to offer you $100 million for your product
43:36 today and just said, you know, here's $100 million. Walk away for your
43:38 product. What do you start doing tomorrow? >> Yeah, build another product. What else
43:43 would I do? Like I I enjoy this. Like the obviously the process of building
43:47 something for people and figuring it out is is always fun. It's just much less
43:51 strenuous when you have some money, right? Like why why do founders get into
43:55 the serial founder mode? Well, because every single time they step up into more
44:01 financial security, the risk of failure becomes less of a of a problem. I've
44:05 personally financed Potscan for many many months before we got some some
44:08 funding from the docom company fund and that wasn't much but it was enough to
44:12 kind of stabilize the business to get it into profitability. I if I couldn't have
44:16 get gotten it there, I probably would have tried to derisk it and build
44:19 something else, right? So, you you you just pick a little riskier of a project
44:23 next time around, but I would still code. Love this. I want to sit in front
44:27 of my monitor and and tell it what to do. This this is great.
44:31 >> That's brilliant. Um Avid, I had such a great time talking to you. I really
44:33 appreciate you hanging out. Thank you for sharing all of this. If somebody is
44:37 interested in continuing to learn from you, you mentioned the podcast and
44:40 Twitter. Are there any other ways? Anything else you want to plug or
44:43 anything they should do to keep connected to you? >> Man, I I talk about Podscan all the time
44:49 on on my Twitter thing like TBF.fm is where my podcast lives and pots.fm is
44:53 where Podskan lives if you want to look into podcasts and figure that out. But
44:57 yeah, just just follow me anywhere. You you will I I will not stop yapping.
45:01 That's uh that's for sure. >> It's But it's good stuff though, so keep
45:03 it up. >> Appreciate it. >> Well, thank you so much for coming on. I
45:06 really appreciate you. >> Thanks so much. Thanks for having me.
45:09 >> Of course. That's it for the rest of