In short
The “Dark Side with Dave” bonus episode is a conversation about using AI to “vibe code” software quickly and cheaply, including rebuilding a Quip-based show-notes app, creating a modern Swift Mac app from open-source ham-radio software, and debating long-term impacts on programmers and pricing for AI-assisted development.
Guests/backgrounds
Dave Bittner (Cybersecurity podcaster, “The Cyber Wire” host; also co-hosts “Hacking Humans” with Joe Kerrigan and “Caveat” with Ben Yellen mentioned). Jason and Brian are present as collaborators; Dave Bittner is the main host, with Jason as the primary “vibe coding” builder.
Key claims
AI can rebuild complex software fast with minimal breakage when guided by experienced developers; unit tests can run automatically before deployment; costs can be controlled via token budgets; AI can “raise the floor” for learning and coding; cynicism about AI is “cracked” as users get past an initial laziness hump.
Notable examples
Rebuilding Quip show notes into a faster, more feature-rich app (font sizing, speaker color highlighting) using OpenAI “Astra” model; deploying to a $12/month VPS and securing it in ~25 minutes; converting ham-radio open-source software into a signed/distributable modern Swift Mac app in ~3 hours; using Docker/Postgres-based local testing with 115+ unit tests; rebuilding a ham-radio digital-protocol app that requires a second ham radio to test.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOVibe Coding New Show Notes App
0:45 to 1:56
Discussion about the new show notes app built by Jason and its features.
“We used to use an app called Quip from Salesforce, and they're sunsetting it.”
Cost Efficiency of the New App
1:56 to 3:25
The cost benefits and additional features of the new app compared to Quip.
“And that was grandfathered in because if we were to pay for it at the normal price, we would have been paying$60 a month for the three of us to be on that piece of software, which is a collaborative document software.”
Experience and Value in Software Development
3:25 to 4:36
Exploration of how years of programming experience influence project pricing.
“If I was going to charge somebody this back in the day, this would have been a 50 to a hundred thousand dollar piece of software easy.”
Efficiency in Software Creation
4:36 to 5:35
Discussion on the efficiency gained from experience in software development.
“You know, it's nice to be the boss when I can just tell somebody who is, you know, somewhat capable, but not really capable because I'm repeating myself a lot.”
Automated Testing and Deployment
5:35 to 7:42
How the new app uses automated testing and deployment processes.
“So I imagine you gave it a crap ton of instructions.”
Rebuilding Open Source Software for Ham Radio
7:42 to 11:21
Dave discusses his experience rebuilding software for ham radio using AI.
“The fun part about the VPS server is I went and I said, make me a VPS server on vulture.com, picked the$12 a month package and gave it my SSH key.”
Challenges of Testing Ham Radio Software
11:21 to 13:33
Difficulties in testing software that interfaces with various ham radio protocols.
“So I don't know what it's doing under the hood.”
Fun of Modern Coding Solutions
13:33 to 14:00
Reflecting on the joy of solving coding problems with modern tools.
“And they're like, well, you fucking pay me then, you know?”
Embracing Technology for Repairs
14:00 to 14:59
Learn how using AI can assist in DIY projects like ham radio repairs.
“You know, it's like throwing matchsticks on the ground in front of a vampire.”
Cynicism and AI's Impact on Programming
15:00 to 16:50
Explore the shift from cynicism towards the benefits of AI in programming.
“To what degree do you feel like your guard of cynicism that you drape yourself in is being taken off of your head as you experience, as you do these various vibe coding projects and they're successful?”
Show all 20 chapters
The Future of Programming Jobs
16:51 to 18:58
Discuss the changing landscape of junior programming roles with AI advancements.
“There's no entry path anymore because nobody's learning the skills that you and I have learned over the years and years and years of being a programmer, which is what helps you be really good at using these tools now.”
AI as an Enabler for Learning
18:59 to 19:22
Understand how AI can serve as a gateway for curious minds to learn coding.
“I think our curiosity is going to override that.”
Cheating and AI in Education
19:23 to 20:30
Analyze the implications of AI use in educational contexts, including cheating.
“I'm particularly interested in your notion that this will help raise the floor.”
Monetizing AI-driven Software Development
20:31 to 22:36
Learn about potential business models for leveraging AI in software development.
“And that, you know, people that are engaged and interested, this is just going to be a tool that really, really helps them.”
Efficiency Gains from AI Tools
22:37 to 23:41
Explore how AI tools can streamline tasks and improve productivity in projects.
“So you're kind of, you could be their AI Sherpa.”
Transforming Laziness into Productivity
23:42 to 25:28
Discover how to turn initial laziness in using AI into enhanced productivity.
“If you'd like to sponsor Jason and the show, please write it.”
The Evolution of AI and Its Applications
25:29 to 27:55
Reflect on the rapid advancements in AI technology and its broad applications.
“Once they get past that hump, They just use it as a tool to make their day better.”
The Ethics of AI Applications
28:00 to 29:10
Explore the ethical concerns surrounding the use of AI in sensitive areas.
“I would not try and devise a bioweapon with it.”
Best Practices for AI Development
29:10 to 30:42
Learn key tips for building and documenting AI projects effectively.
“But the final thing I have to say is everybody's going to come back to me and say, but hey, aren't you building something?”
Closing Thoughts and Collaboration Tools
30:42 to 31:22
Discuss the importance of collaboration tools and final thoughts.
“I know I've been long-winded on this today, but it's just fun.”
Transcript
Automatic transcript. May contain errors.0:00The Dark Side with Dave.
0:07Brian Schulmeister:Welcome to The Dark Side with Dave featuring Cybersecurity's hardest working podcaster Dave Bittner. While he's not covering the daily cyber beat on the Cyber Wire, he's busting scams with Joe Kerrigan on Hacking Humans, tackling privacy with Ben Yellen on Caveat, and proves there's always room for one more cybersecurity podcast with only malware in the building. Hello, Dave Bittner.
0:27Dave Bittner:Hello, Jason. Hello, Brian. And it's good to be back. Seems like the weeks fly by, don't they? They do. Yes, they do.
0:35Brian Schulmeister:Yes, they do. So, Dave, I have spent the last couple days vibe coding our new show notes app. And you are, besides Brian, the first person to see it outside in the wild. And I'm curious. We used to use an app called Quip from Salesforce, and they're sunsetting it. So that is going away. So I rebuilt it using the eyes of the A's.
1:03Dave Bittner:I did not use super intelligence or super duper intelligence, but I did use the codex to build this thing out. And I'm curious, what do you think of it? I'd say so far so good. I mean, it's very comparable to what we were using before, so I feel comfortable in it. My initial reaction was, oh, I got to increase the font size. there's a slider for that yeah so i just so i did that and that was fine uh but uh it now it seems fine maybe a slightly less colorful than what we had before but uh i there's nothing i don't have any complaints it all seems to to work and i think we slid into this new version with little uh
1:50Brian Schulmeister:resistance at all so well done thank you thank you so the thing the reason i bring this up is we spent$30 a month on Quip. And that was grandfathered in because if we were to pay for it at the normal price, we would have been paying$60 a month for the three of us to be on that piece of software, which is a collaborative document software. We can collaborate in real time with our documents and do all sorts of stuff. The new version that I made, it has like 18 more features that we need. Like it highlights the colors of the different speakers so we can tell who's doing what and all sorts of stuff.
2:24Brian Schulmeister:this took me seven hours of just waiting for it to do its thing i spent maybe an hour prompting it and the cost in tokens was 14 oh now i take this software i put it on a 12 a month vps from vulture and i just saved us a bunch of money and it's better and faster so i was just i you know i shit on vibe coding all the time. And I think that if a regular person came in to do this, it would, they would have spent a hell of a lot longer because I know what I'm doing as far as how to build software. It's exactly what I told you, Jason.
3:03Jason DeFillippo:I was like, this is, this is fantastic. But the reason it's fantastic and the reason it worked is because you've got 20 plus years of programming experience behind you. We've got 10 years of using Quip behind us. You knew exactly what needed to happen. You knew how to get there. And if somebody just said, just tried to do it, that didn't have all that, it wouldn't, it would take months and thousands of dollars. Yeah.
3:26Brian Schulmeister:If I was going to charge somebody this back in the day, this would have been a 50 to a hundred thousand dollar piece of software easy. That was my next question.
3:33Dave Bittner:So if someone came to you and said, we want you to execute this, knowing what you know now, how do you, how do you factor in your 20 years of expertise into the bottom line when this you you did this in a day or two
3:51Brian Schulmeister:yeah that's the thing it's like how do you factor that in it's like i do you charge per project do you charge hourly you know it's like if i charged hourly i would never eat again so you have to you have to take into account it'd be like a thousand dollars an hour to actually get something done
4:05Dave Bittner:back when we were you know back in my video days when we were you know the early days of uh like 3D rendering and that sort of stuff. We had the realization that if you charged by the hour for rendering, you were not motivated to update your computers because the slower they were, the more money you made. Right. Yeah. So we just started making everything project based. Yeah. It just worked better that way.
4:31Jason DeFillippo:I mean, for the 20 plus years I was running my company, we couldn't do hourly. It was everything was project based. So. Yeah.
4:38Brian Schulmeister:And I found that too with because I used to do podcasts by the hour and it's like I can do I am so fast now it's like if I do it by the hour again I wouldn't eat so you have to do it by the project because otherwise that experience like goes out the window it's like I am so efficient with everything that I do because I've got 25 30 years behind me and it's like the kids nowadays are just like oh you know I vibe coded my way out of this and so my hatred for the vibe coder still stands I do enjoy I do Do enjoy doing it, though, because this was a shit ton of fun. You know, it's nice to be the boss when I can just tell somebody who is, you know, somewhat capable, but not really capable because I'm repeating myself a lot.
5:21Brian Schulmeister:A lot of that time is me going, no, not like that, like this. Like that sandwich video that I posted about somebody trying to tell an AI how to cut a sandwich in half, you know? There's a lot of that going on.
5:32Dave Bittner:But I'm curious, like the first time this came out of the gate. So I imagine you gave it a crap ton of instructions. No. No?
5:41Jason DeFillippo:Oh, really? No, he pointed it at Quip and said, rebuild that.
5:45Brian Schulmeister:I said, go to the Quip website and rebuild this. That's literally what I did. And it came back and I've been actually taking features out that it put in automatically for stuff that I'm like that we never use and then customizing it. So when the first round comes back, how many stupid broken things are in there? Almost none. Wow. Almost none. It worked almost flawlessly out of the gate. There were, you know, formatting things and just little tweaks. But as far as functionality went, it went like pretty much straight up. And I'm using the new Astra model from OpenAI on high, which is, you know, like, you know, the top of the line open AI model.
6:29Brian Schulmeister:I don't, me and fucking Claude are going to go out back and have it out one of these days because that motherfucker does not listen to me anymore like he used to. But, yeah, the new Astra, it works pretty damn well. Got yourself a new girlfriend. Yeah. Pretty much it.
6:46Jason DeFillippo:She gets you.
6:47Brian Schulmeister:She does. She totally gets me. So it came back pretty quick. And every time that I put in a prompt now to modify something, I know how to word it as a programmer would word it to make it do the thing that it needs to do. I know the caveats along the way. I was like, do this, don't do this, make sure this is in here, look to these pages, make sure you update this. So it's very much knowing what, it's not like, make this clicky-clicky go there. It's not like that.
7:17Dave Bittner:Is it doing its own testing along the way?
7:20Brian Schulmeister:Absolutely. It has the version that I have right now, and I didn't tell it to do any of this. It will do over 115 unit tests based on any update that I make to make sure everything works before it deploys. And it runs locally in a testing environment that it built itself using Docker and Postgres and all this other stuff before it will deploy to the VPS server. The fun part about the VPS server is I went and I said, make me a VPS server on vulture.com, picked the$12 a month package and gave it my SSH key. And when I went, when we were done building it and I had it locally running, I said, go secure this server, update the server with the software needed to run this package and install it and write the manual.
8:05Brian Schulmeister:So if I ever needed to do this on my own without you, I could do it. 25 minutes later, VPS was all set up, locked down, SSH was locked. All the security packages were updated and the software was running and I sent it to Brian. It's fucking crazy.
8:20Dave Bittner:So how do you deal with the budgeting? Like you said, it cost$14. How did you make it so that, you know, seven hours later it wasn't$1 ,000?
8:32Brian Schulmeister:I'm using the$100 a month OpenAI package, the ProPlan. Okay. So I'm not using the API for this. So if I was using the API, you could actually track it and see how much you're spending on it. I'm going by the percentage of tokens that I've used for the week because you get a weekly token allowance. So I take it, divide it by five weeks, and then take the 100, divide it by five, 20 bucks a week, use 47 % or whatever, 48 % of the tokens, and then boom. That's how you get your number. So for that part, it was easy. Now, I mentioned all of this in service of the challenge that you laid down to me, which was to go take the open source software that you use for your ham radio and rebuild it.
9:17Brian Schulmeister:And I did. And I sent it to you.
9:19Dave Bittner:Yes. And this was all a lead up to that. Well, I have opened it. And let me just say off the bat that I am dazzled by how it looks like mission accomplished. You and by you, I mean you and your little AI gremlins. or gnomes, perhaps, did a great job. And I'm curious, like, what did you tell it look and feel wise? Did you say make this look like a Mac app or, you know, use modern visual standards? Like, how did you tell it to break out of that old Windows 95 look and feel?
9:54Brian Schulmeister:Well, the first thing I did was I told it that we were going to take this open source software and we were going to turn it into a modern Swift app for the Mac. So I wanted it to use Swift for everything. That's all the apps that I use now because the Swift is like the kind of the go-to. So I don't want to use any legacy stuff. So I'm like, turn this into a modern looking Swift app that we're going to compile for a Mac that I need to be able to sign and distribute to my friend Dave, who has a ham radio. And I created a little Git repo with the source that you gave me and let it run. And that's what it came up with out of the gate on its own.
10:31Brian Schulmeister:I gave it no other design instructions besides make this a modern Swift app for a Mac.
10:38Dave Bittner:How long did it take it to do it?
10:40Brian Schulmeister:Three hours. It took three hours with a couple little tweaks. Because there was that one really ugly screenshot that was very, very Windows 95 with the graphs and all that stuff with the graphics. That one I had to go back and tell it a couple times. I'm like, no, go back and fix that one too. Because that was like a subset of the code. And it went back and fixed that one too, I think. So I can't test it because I don't have a ham radio. That's why it was interesting. I wanted to see how it worked, if it worked, from first run to sending to you the first time.
11:10Dave Bittner:Well, I still have to hose it up to my ham radio, so I've been holding off on that. As I mentioned to you, I was under the weather over the weekend, so I promise you by the next time we get together, I will have done that. It's interesting because I think the first time I booted it up, as it was sort of initializing itself, one of the old windows popped up from the old version ever so briefly and then it disappeared and the modern interface came up and I went, ooh, smoking mirrors.
11:39Brian Schulmeister:Jazz hands. Yes, exactly.
11:42Dave Bittner:Exactly. So I don't know what it's doing under the hood. I don't care as long as it works. Yeah. But the other thing that this made me realize when, because first of all, like I mentioned this offhanded is wouldn't this be a nice thing to do? and you, you know, very nicely and you just did it and you sent it to me and you're like, oh, I just did this. And I was like, oh, I just created a bunch of work for myself.
12:12Jason DeFillippo:Because now I have to test it.
12:15Dave Bittner:Well, but that's the thing. Like an app like this, it made me realize why there aren't a whole lot of different versions of an app that does this. because this app does so many things with so many different protocols. There's no way I can test them all. The other thing about this particular app is that in order to use it, you need a friend who also has a ham radio and is the other side. You know, there's not really like a null modem for communicating with yourself with this sort of thing. So I have to figure that part of it out as well. And I will. But yeah, so it was a nice sort of recalibration for me about what it takes to build and maintain a project like this.
13:03Dave Bittner:It's not just the behind the scenes. It's not just the technical. that when you're dealing with dozens of models of ham radios and dozens of digital protocols, that's a crazy grid to try to test against. And so, you know, respect to the folks who made the original. I don't know how they manage it. And I can understand why they don't update it very often.
13:28Brian Schulmeister:So, yeah, it saves a lot of time and a lot of late nights and a lot of frustration and a lot of, you know, under appreciation coming to them for people who are just like, my radio doesn't work. And they're like, well, you fucking pay me then, you know? Right. Exactly.
13:43Dave Bittner:Exactly. Yeah.
Read the full transcript
13:46Brian Schulmeister:But I think I just think it's fun that, you know, when we can do this stuff now and yeah, I'd still make fun of vibe coders. I'm going to make fun of vibe coders, whatever. Just because It's fun, you know, but you throw a Gen X coder a problem, Dave, I'm going to solve it no matter what. That's right. That's right. It's true. You know, it's like throwing matchsticks on the ground in front of a vampire. They're going to count them. They're going to fucking count them. Yeah.
14:15Dave Bittner:Yeah. Yes. Believe me. I understand that impulse as I'm, you know, in the midst of repairing my ham radio power supply that blew, you know.
14:28Brian Schulmeister:Oh, not another power supply.
14:30Dave Bittner:But he's got all the tools now.
14:32Brian Schulmeister:And the windmill. And the windmill.
14:35Dave Bittner:I have all the tools, but I'm doing the same thing I did when I built or when I repaired the piano, which is I'm using ChatGPT to walk me through the repair. And it's great. It's a great companion.
14:58Dave Bittner:So here's my question, though. To what degree do you feel like your guard of cynicism that you drape yourself in is being taken off of your head as you experience, as you do these various vibe coding projects and they're successful?
15:17Brian Schulmeister:So as these models get better every time, which we knew it was going to happen, we knew every time you get a new one, it's going to be whatever you're working on is the worst it's ever going to be with with a few exceptions, because they have gone back a couple of times and really screwed the pooch. I'm enjoying it. I'm enjoying it a lot because. I understand what it's doing for the most part when it builds things. So for me, it's literally like I finally have my own team of people that I can tell what to do. They're, you know, and I'm just, I enjoy that part. I'm still a horrible businessman, so I'm never going to make any money off of it.
15:54Brian Schulmeister:And while I understand that, you know, the kids nowadays can find this extremely intoxicating and I, you know, from what you, from what you asked Dave, my cynicism is, is, you know, pretty much razor thin at this point. I fucking love using it. I have it open all the time. I feel bad when I'm not actually using it because I'm like, I have this tool here that I can just point anything at it and say, make me software. and it's really good at software. There's a lot of things it's not good at, but as far as software goes, it's terrible at emotions. So am I. So we get along really well. It's great for me.
16:30Brian Schulmeister:One of my best friends. But, you know, as far as building software goes, I am just, I am like a kid in a candy store, you know? I've got, yeah.
16:39Jason DeFillippo:Do you think about the longer term impact of it though? Do you? No. Well, I know you tend not to think about long-term impact because you're convinced you're not going to be here in about 10 years anyways. But for the, like, everybody, like, the problem that we always talk about is nobody's hiring junior programmers anymore. There's no entry path anymore because nobody's learning the skills that you and I have learned over the years and years and years of being a programmer, which is what helps you be really good at using these tools now. Or do you just think the tools will get sufficiently good enough that we will never need entry-level programmers anymore?
17:12Jason DeFillippo:And it will get to the point that those dumb fucks that used to project manage us that didn't know anything will be able to do what you're doing now.
17:19Brian Schulmeister:I have a very good answer for that. I think that the people who like to code are going to learn how to code no matter what. I think people like us, people like Dave, we're tinkerers. We love to learn. I think those people aren't going away. And I think this is a great gateway to actually learning how to code. Because you can say, make me this. And then you can go look at how it made it. And if you want to learn, if you have the curiosity, those people are still going to be around. They're not going anywhere. You know, engineers are not going anywhere. You know, anybody with curiosity for the sciences, they're not going to go anywhere.
17:56Brian Schulmeister:Just the standard. It kind of comes back to what I said at the beginning of the AI revolution is here's mediocrity. Here's the people below. Here's the people above. It is eventually going to compress all of those people. but I'm starting to think more along the lines of the people that are up here are going to stay up here. The people up down below, it's going to bring them up from, you know, being dumber than dirt to at least decent at doing what the fuck they're doing. The smart people are going to learn how to use this as a tool and not really rely on it the way that we used to. I mean, we're seeing this with writers.
18:32Brian Schulmeister:We're seeing this with everybody else. The people who love to write, the people who love their jobs, it's not affecting them. It's actually making them better at it. To get to your point, Dave, I think my cynicism is definitely cracked quite a bit because I think that the human nature behind what we do and what we love to do is going to override the laziness factor that I was originally feeling that everybody was going to kind of fall into. It's like, oh, well, I don't have to do that. I can just type a thing and go. I think our curiosity is going to override that. And the people that matter, at least.
19:05Brian Schulmeister:You know, there's always going to be the unwashed middle who just don't give a fuck and they want to go have a Big Mac and go watch Friends. So that's not going away. But I think the smarter people in society and the people who are more ambitious and have drive, I think this could actually help them some. It's a shocker, I know.
19:23Dave Bittner:I'm particularly interested in your notion that this will help raise the floor. you know it'll provide capabilities to people who otherwise might not have opportunities it'll be an enabler for them that I think that's a really interesting take yeah it's just what I've seen
19:42Brian Schulmeister:lately with you know people that I know that maybe aren't on the you know the deeper end of the pool have been using it a lot you know it seems they seem to be learning how to use it to help them get ahead and bring them up from where they were. So, you know, I also have seen some people use it to cheat on tests for regulatory certificates and things like that, where they should probably learn the material. But, you know, come see, come see.
20:13Jason DeFillippo:Cheating is going to cheat.
20:14Brian Schulmeister:Cheating is as old as time. So, yeah. I mean, there's no different from them having another window opening and Googling it. It's just a little faster to take a picture of the quiz and say, what's the answer than it is to go Google, Google, Google, you know, it's same.
20:28Jason DeFillippo:I think I've kind of been on board with your take on this, Jason, for a little bit now. And that, you know, people that are engaged and interested, this is just going to be a tool that really, really helps them. What I'm always more intrigued by, and I don't know if there's an answer for this, is more the financial perspective, because we were just talking about, you know, how do you price out a project like something like this, especially when the people that you are selling it to or theoretically selling it to at some point will know because they're aware of the tools as well, that this is now just something that used to take two months for us to build and would have been$100 ,000.
21:06Jason DeFillippo:They know that you know that they know that you know that it's actually five hours and a couple tokens. So how do you make a living then?
21:15Dave Bittner:Because we all know. This is why I got, this is one of the major reasons why I got out of the video business was one of my biggest clients came to me and said, we want you to edit a video. We're making a new HR video. And I said, great, who's going to shoot it? And they said, we are on our phones. And I went, okay.
21:32Jason DeFillippo:Yeah, well, and Squarespace and social media killed any web design as business. Like it was just done instantly. And now Squarespace is fucked.
21:40Brian Schulmeister:Yeah. Squarespace is dead. So here's, I've been trying to think about this because I'm trying to put together, you know, a business model for rewriting software for local businesses and small business owners who use subscription software that they pay for every year. So find out what they pay every year. So say it's$100 a month per however many users, how many seats they have. So they spend$1 ,000 a month. If I can get that down to them spending$100 a month, take whatever that difference is, cut it in half, you pay me that for one year's worth of software that I am writing for you, And then from beyond that, you're on your own.
22:19Brian Schulmeister:You can hire me as a, you know, I can keep a maintenance contract after that. But look at what the delta is between what they're going to pay now versus what they were paying. Take a small portion of that or, you know, a decent portion of that. And then you've solved their problem. They're spending less money. You're making more money and you're taking it away from the big conglomerates with the big software that they were paying for that they use probably a tenth of the features anyway.
22:45Dave Bittner:You know? So you're kind of, you could be their AI Sherpa.
22:49Brian Schulmeister:Yeah, just do AI arbitrage between shit software and good software. And also you can customize it then for what they need, which I think is the real kicker. Because I think what I did with our stuff with the show notes software, this is going to save me a million hours if we keep doing this show. It's crazy. And I've written an RSS reader that is also now tied into this. So I read my articles. I press a button and it goes into the show notes automatically so I can then tweak my voice to the stories that are going to come out. And I've got it on my phone. I've got it on my iPads. I've got it on all my Macs, the same news software that then ties into the same back end.
23:32Brian Schulmeister:And this took like no time at all compared to what it would normally take. And, you know, and if other people find use in this, then I pass it along to them. I just give it to them because, you know, it's, it's, I don't know. You need a sponsor, Jason. You need a sponsor. It's just interesting. Right?
23:49Jason DeFillippo:If you'd like to sponsor Jason and the show, please write it.
23:53Brian Schulmeister:Patreon.com slash GOG, baby.
23:56Dave Bittner:So. No, it's, it's fascinating. And I'm, I'm really, I'm intrigued by everything you're saying here because particularly the part about going through the lazy zone, right? and coming out the other side. Because I've had that same thought of what happens with everyone's natural inclination towards laziness. We all have it. I have it. Everybody has it. And so an AI just feeds into that initially. So it's really interesting for me to hear you saying that maybe that's a zone everybody goes through. But for some people, you make it to the other side of that. And now it's being used for all kinds of things that improve your productivity and your happiness because you're having fun, right?
24:43Brian Schulmeister:Yeah. Yeah. I think, and you know, since these things are just basically prediction machines, you, you learn what they're going to, they're going to give back to you. That's why the writing is, is so terrible now. You know, it's like everybody who uses it for writing, they can spot it. They can spot somebody who used AI for anything, an email, a webpage. You can, you know, you become attuned to it. Yeah, you totally become attuned to it. But then you take that and then you put it in your own voice, but it's a good starting point. It gets you 80 % of where you need to be in a short amount of time, but then you can make it your own.
25:20Brian Schulmeister:And I think that people who use, you know, LLMs repeatedly get past that point. That laziness, like you said, there's that hump. Once they get past that hump, They just use it as a tool to make their day better. And I'm seeing that over and over again now, which is the weird thing because I never thought I would. You know, I just thought everything was just going to devolve because we do this show and everything is shit. But I'm actually seeing the other side of it where people are really kind of, you know, it's new technology. We're going to evolve with the technology. And seeing people evolve with it is fascinating.
25:55Brian Schulmeister:It's really fascinating. And just watching it myself, you know, I never thought we would be having this conversation today, ever. I'm like, vibe coding is shit. Fuck these guys. I got 100 ,000 views on my Instagram post about vibe coders are shit. You know? Come on.
26:09Jason DeFillippo:Right. Right. Well, next, Jason, you can point your vibe coding lens at Riverside. Let's rebuild this thing.
26:17Dave Bittner:Ooh. Ooh. Hmm. There's some infrastructure there that they paid for.
26:22Brian Schulmeister:There's some infrastructure there that they paid for that is difficult. That is difficult to do. Mm-hmm. Mm-hmm. But I can't wait to try the new one next week that you send to us. No, this one works. This one works.
26:35Dave Bittner:You got me thinking, Jason, because what I am about to embark on a project of rebuilding the shitty website for my local Rotary Club that I'm a member of. Right. And now I'm thinking, hmm, maybe I'll just this would be a good opportunity to learn something and play with something. And, you know, rather than just going to another platform that has templates, maybe I can make exactly what I need. Absolutely. Do that. Absolutely do that.
27:07Brian Schulmeister:Seriously. And use both of them. Use OpenAI and use Claude at the same time and send them the same prompts. See which one works better for you for what you're trying to do. Experiment. It'll cost you 40 bucks for a month and you can just, you know, beat the crap out of it. And if there's something that you like and you run out of tokens, well, then, you know, drop the hundred bucks. A hundred bucks to build a website is like, what the hell? That's nothing. Right. Exactly. Literally nothing.
27:35Jason DeFillippo:That's what it costs, Jason. You and I would have been like dead a long time ago. I know. A hundred bucks for a website. I was charging a hundred thousand for websites back at the time.
27:43Brian Schulmeister:A hundred dollars was the phone call that I got on to tell me if I wanted to work for you or not. You know, that was it. So I can spend a hundred bucks and build a really nice website. Fuck. Why not?
27:55Dave Bittner:Yeah, interesting. Yeah, and like I said, these things are unit testing against themselves.
28:01Brian Schulmeister:Now, for code, it is really good. I would not try and devise a bioweapon with it.
28:09Jason DeFillippo:Or penicillin. I don't know, try to figure out if there's nuclear weapons on a Chinese ship, or I don't know, target a school somewhere and accidentally kill a bunch of kids.
28:18Brian Schulmeister:Yeah, well, there's lots of reasons not to use it for things.
28:20Dave Bittner:No, there was the joke this week about how the FAA is threatening to use AI for air traffic controllers. And I saw someone with a post that said, oh, you're right. That runway was full of traffic.
28:34Jason DeFillippo:Oops, my bad. Yeah, sorry.
28:39Brian Schulmeister:Oh, man. But yeah, it's a strange, strange new world, I think. And I'm enjoying it. Actually, now I'm enjoying it. Now that my cynicism is kind of, you know, tempered a bit. Yeah. You know, because things change. Things change. I may have been hung on the past just a little bit.
28:58Dave Bittner:But I'm happy to see you so happy. Like, to see you truly and honestly excited about something, energized by something like this. I think that's wonderful.
29:09Brian Schulmeister:Thank you. It's fun. It's fun. But the final thing I have to say is everybody's going to come back to me and say, but hey, aren't you building something? in somebody's backyard by using AI to make the software that you're using, which is the grumpy old geek's prime directive is don't build something in somebody else's backyard. If you know how to do it, you have it self-document along the way so you can move to somebody else's backyard. You can keep hopping backyards. Just be a squatter. Or open a text editor and fix it yourself. You know, that's the thing. This is not like I can everything that I've built, I can move, I can rebuild.
29:52Brian Schulmeister:And I have self-documenting is key. Always put in. So here's my last tip for the day. When you're when you're building a prompt, Dave, this goes with you to start a local Git repo and say every time you make a change, commit that shit with a comment that says what you did and keep a manual of what you're doing so that either a new version of yourself or another LLM can pick up this project from where we left off. It'll save you a million hours because later on what that document does is when you load that project back up, either in a new version or a new model or whatever, it'll read that and it will save that amount of time, save it the tokens of having to read every line of the source code.
30:37Brian Schulmeister:And it's one of those things, just self-documenting as you go has really helped out. I know I've been long-winded on this today, but it's just fun.
30:45Dave Bittner:Oh, yeah. It's great. It's been a great conversation. All right. So I'm going to save all the stuff I put in the show notes for next time in our beautiful new collaboration environment.
30:59Brian Schulmeister:I called it grip, but I should have called it gripe. Oh, yeah. This opportunity.
31:04Dave Bittner:It can be renamed. Just put a little crayon E on the end of it, you know? Yeah.
31:12Brian Schulmeister:By the time you guys refresh, it probably will be. Yeah.
31:15Jason DeFillippo:Make sure it commits and documents that change, Jason. Will do. Will do. All right, guys. All right.
From the publisher
Very Important Links!
Support the show at Patreon! - https://patreon.com/gog
Other ways to support the show! - https://gog.show/donate
Join our Discord! - https://discord.gg/r4ZmSHB
Buy some merch! - https://shop.gog.show/
Recorded on Wednesday, September 23rd, 2026.
Jason, Brian, and Dave head to the Dark Side for a deep dive into vibe coding—and, against all odds, Jason may finally be cracking. After Salesforce decided to sunset Quip, Jason used AI to rebuild the collaborative show-notes system the guys have relied on for a decade, adding more features while cutting the monthly cost to almost nothing. Then he took up Dave’s challenge to modernize an aging open-source ham-radio application, turning it into a native-looking Mac app in a matter of hours. The catch: after decades of programming, Jason knows exactly what to ask for, what to watch for, and when the AI is confidently doing something stupid.
That leads to the bigger question: what happens when software that once took months and tens of thousands of dollars can be built in an afternoon? The guys get into value-based pricing, the disappearing entry-level programming ladder, whether AI makes people lazy or gives curious people superpowers, and how custom software could start eating away at expensive SaaS subscriptions. Jason also lays out what he’s learned from actually building with these tools: keep everything portable, use Git from day one, make the AI document its work, and never build something you can’t eventually maintain yourself. Turns out the future may not belong to vibe coders after all—it may belong to people who already know what the hell they’re doing.
See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
