Modern Software Podcast: Autonomous SDLC : A Developers Perspective - David Robertson on AI-Generated Books, Autonomous SDLC, and Modern Development Tools

Modern Software Podcast: David Robertson on AI-Generated Books, Autonomous SDLC, and Modern Development Tools

In this episode of the Modern Software Podcast, host Mike Verinder sits down with David Robertson, Director of Enterprise Software at Exterior Finance, for a wide-ranging conversation on AI, software automation, and modern development practices.

Mike and David discuss:

• AI-Generated Content: David shares his journey of experimenting with ChatGPT to write and publish books, from ideation to publication in under 10 days. He explains the challenges of prompt engineering, AI consistency, and generating AI-driven cover art for his second book—a children's picture book.
• The Autonomous SDLC: They explore the idea of an autonomous software development life cycle (SDLC), covering how AI tools can generate test cases, write code, automate testing, and monitor production environments. Mike shares insights on current tools, autonomous testing innovations, and the investment hurdles preventing widespread adoption.
• Modern Development Tools: David dives into the power of GitHub Copilot for accelerating development workflows, from writing boilerplate code to generating unit tests and learning Python for emerging AI frameworks like LangChain and Semantic Kernel. They also touch on the rise of tools like Cursor AI Studio and the role of AI in enterprise coding environments.
• Security Testing and DevOps: The conversation shifts to the growing importance of security testing in modern software development. David discusses integrating static and dynamic analysis tools into CI/CD pipelines to identify vulnerabilities and ensure compliance—critical for finance organizations.
• Testing Trends: Mike provides valuable insights into the current state of automated testing, including API testing, mobile testing challenges, and the role of tools like Selenium, BrowserStack, and Sauce Labs. They explore why Selenium remains dominant in the industry despite new innovations in AI testing.

Along the way, Mike and David reminisce about their shared history at Drive Financial, the evolution of development tools, late-night deployments, and the value of strong teams in software projects.

This episode is packed with insights into AI, testing trends, and the future of automation—perfect for developers, testers, and software enthusiasts alike.

🔔 Subscribe to Modern Software for more interviews and discussions on software development, testing, and innovation!

Topics Covered:

• AI-generated books and images
• Autonomous testing and SDLC
• GitHub Copilot and modern coding tools
• Security testing trends and DevOps integration
• The challenges of mobile testing and enterprise tools

Transcript

Mike (0:1.944)
Hey, Dave, welcome to Modern Software Podcast. Today I'm talking with David Robertson. He is the director of enterprise software for Exterior Finance. But Dave and I kind of go way back. We known each other from Drive Financial and some previous work that we've done together. We've stayed in touch for a good 20 years and wings and beer.
things like that. welcome to the podcast, David. How's it going,

Dave Robertson (0:37.357)
Going well, yeah, thanks Mike, it's good to be on. So yeah, appreciate the invite.

Mike (0:43.330)
Yeah, so ⁓ I am continuing this discussion around autonomous testing or not just autonomous testing, but really the autonomous SDLC ⁓ and ⁓ where that journey is and how long that journey is going to be. ⁓ we're going to get to a state of an autonomous SDLC or is that a 10 year state or
next year state or, or what that really looks like. Um, but that's a, that's a big journey, man. So that, that starts, you know, obviously testing is an aspect development is an aspect requirements are an aspect of that. Um, but, uh, anyway, so I, you know, I know we, we have both played a lot in AI lately. Um, I guess you would call us AI enthusiasts a little bit.
Um, and so I, know, one of the things that, that I've seen that you've done in, uh, recently is, uh, you have a couple of books that are out, uh, that you have done with AI and tell, tell me about that, man. Tell me about that journey a little bit and, and what you've done so far.

Dave Robertson (2:2.967)
Yeah, sure. You mentioned the AI trend, right? So me, like a lot of other folks, are trying to stay current with a lot of what's happening in the marketplace. And so honestly, I got a little bored this summer and just decided on a personal kind side hustle, side trip kind of thing. Hey, let's kind of reach out there a little bit and see what can I apply AI to. And I just decided, hey, let's try it with book.

Mike (2:3.406)
all of us.

Dave Robertson (2:33.335)
publishing, authoring, all of these kinds of things. Never really got into that before. And so it was really just kind of like, let's just start from nothing, essentially, and see how far we can go. And so that's exactly what I did. I started with Chat GPT, came up with a basic outline, some very, very simple research. But my goal for the first one was to publish everything, start to finish, get it in the marketplace in a week. And so I wasn't really looking.

Mike (3:2.467)
Wow.

Dave Robertson (3:3.371)
Yeah, I wasn't really looking to create some grand novel that everyone would think was deserving some Pulitzer Prize or anything. I just kind of threw some stuff together and started with ChatTPD to generate some content.

Mike (3:19.502)
So from from ideation all the way to I'm guessing Amazon publishing.

Dave Robertson (3:29.057)
Yeah, Amazon was actually one of the laggards. I used a platform that will broadcast the distribution to different sellers, walmart.com, Barnes and Noble also included Amazon, but there were several roadblocks actually with Amazon because of the AI content that was in it. so that one took, I think, more on the line of 10 days instead of the seven calendar days that I gave myself this artificial deadline.

Mike (3:58.584)
took 10 days just to publish it.

Dave Robertson (3:58.723)
Yeah, it was all worked. 10 days start to finish. So I created cover art. I generated, it was about 14,000 words. So more of a novella short story, you know, kind of thing, right? But yeah, it took a little while and then wrote a few roadblocks through Amazon that I had to get around. I wound up publishing directly through KDP, Kindle Direct Publishing is what they call it, to get around some of this stuff.

Mike (4:10.764)
Yeah.

Mike (4:22.702)
Well, you said little. mean, I guess the book was about that thick. You know, I guess it was little, but it wasn't like a paragraph. I mean, it was was a real story. Right. So I read the first one or I bought the first one. I haven't read the second one and you're working on a third one. Is that right? Yeah.

Dave Robertson (4:26.761)
It's about 60 pages or so. Yeah. Yeah. Yeah. Yeah. Exactly. Yeah.

Dave Robertson (4:39.479)
Yeah, thanks Mike.

Dave Robertson (4:46.541)
Yeah, so the second one, again, to kind of challenge AI, the first one was basically generating words. Can AI generate text, words, content, and be at least decent as far as the stuff that it generates? The second one, the first one is about a volcano in Costa Rica, right? And so I did a little bit of research, found some German explorers that had been there back in the day.

Mike (5:3.096)
Yeah, what was the story about David?

Mike (5:12.290)
Okay.

Dave Robertson (5:16.013)
found a map on I think Wiki Commons and stuff like that. And then just kind of put together a fictional story that I made up and said, hey, I know you write the story, right? Here's kind of the outline of the book and chapters and whatnot.

Mike (5:31.406)
Okay, so you gave you gave chat GPT kind of an outline of what you you wanted. Did you and you said here are three main characters or something and.

Dave Robertson (5:45.089)
Yeah, that's actually where it got terribly complicated. I started with just the basic outline, said, write a story. And after about 12 or 1500 words, it had already gotten to kind of the climax of the story. I'm thinking, no, this is not going to work. So there's a whole lot of prompt engineering that went into this. And I basically started having it generate chapters at a time, which was a lot easier for the AI to kind of manage and handle.

Mike (6:7.917)
Yeah.
Yeah.

Dave Robertson (6:13.871)
But then my prompts kept getting longer as the book goes on because as we all know AI is non-deterministic, but I was telling a consistent deterministic story. So in one case, it actually resurrected someone who had died in a previous chapter. And so I continuously had to go back and edit my prompts to clarify some of those characters and basically how certain chapters would end to maintain consistency in the story.

Mike (6:40.042)
Yeah. So I would assume if you have a chapter and you just say, okay, three people, one of me, they decide they want to go on a journey, something like that. And, uh, I'm assuming you have to read your own chapter before you tell it like what to make in the second chapter. Unlike how you want the story to sort of, uh, deviate or evolve.

Dave Robertson (7:8.525)
That's basically what the prompt was. I think by the time I was halfway through writing chapters in the book, my prompt was almost 500 characters that I was inputting for every chapter so that it would maintain consistency. And like you said, oh yeah, this has already happened. So, you know, don't bring this person back or whatever, right? So, yeah.

Mike (7:20.908)
Wow.

Mike (7:28.994)
Yeah. So, um, that's interesting because it, it makes you want, it makes you try to end the story. Otherwise it would just keep going and going and going. Right. So it makes you, I would imagine you're like, okay, when, when is this going to end?

Dave Robertson (7:47.365)
Yeah.

Dave Robertson (7:54.277)
There was a lot of that going on, yeah, for sure. Yeah.

Mike (7:57.898)
Okay. All right. Wait, what were one, one problem, one thing that I noticed, and I have tried to, uh, write applications a number of times, uh, with chat GPT. And one thing I always notice is that, you know, you keep, when you get, when you start getting long, uh, you know, chat GPT slows down, uh, you know, it starts. It's like it's on dementia or something.

Dave Robertson (8:27.267)
Too many tokens, right? Yeah, what do I do now?

Mike (8:28.622)
It starts forgetting what it did and, and you know, it, it's, it's really bad. The longer it gets, the worse that it gets. And so I have, I found like, if I want to code an application, I have to start small, get chat GPT to write a requirement document.
of what we've done or the databases or maybe a database schema or something like that, I've made with this application and like keep all that documented. And then when I see that it starts getting, uh, it starts losing, I don't know what you call that. I'm sure there's a AI word for that, but

Dave Robertson (9:16.409)
It tends to kind of zone out in the middle of things, right? As content gets larger, yeah.

Mike (9:19.946)
Right. Yeah. When I see that it starts doing that, I say, okay, give me a snapshot of everything that where we are. And I save that document or sets of documents because I usually have one for a database and one for code and whatever. And I say, okay. And I just start a new thread.

Dave Robertson (9:38.817)
Yeah.

Mike (9:45.984)
And then I load all the documents to where we were from that that last thing and and started to start kind of started up again. I don't know if that's the best way or the best approach to do that, but I'll tell you, not doing that and getting stuck in the middle sucks.

Dave Robertson (9:57.187)
Sounds like a lot of work, Mike.

Dave Robertson (10:9.251)
Yeah, yeah, for sure. That's one of the dark sides of all this.

Mike (10:11.826)
and not having a backup plan or a way to deviate after you've done this much work, you know, just sucks. anyway, if you have a better way, I would imagine though, is that the same way you had to do that for writing the book? I mean, you're getting on these big, huge prompts.

Dave Robertson (10:34.401)
Yeah, yeah, that's exactly what it is, right? I had to come up with a very detailed chapter outline for every chapter. And then that still included a large part of the prompt that said, hey, basically, here's what's already happened. Here's kind of the theme of the thing. And then here's where we're going. And you had to throw all of that at the AI. Otherwise, it was all over the place, right?

Mike (10:59.736)
Yeah. Have you tried any other LLMs besides chat GPT?

Dave Robertson (11:5.101)
So we use several others. On the code side, we use GitHub Copilot quite a bit. And it's just recently came out with the option to where you could choose different models. And so I'm exploring some of the anthropic models rather than the GPT model from OpenAI. Different things like that. Yeah, they're all slightly curated for different purposes. Yeah.

Mike (11:23.308)
Yeah.

Mike (11:30.806)
Yeah, I've seen, uh, uh, anthropic, I've seen people use their AP, their chat GPT API with anthropic and reference chat GPT from anthropic. Um, uh, a number of times that I don't, I why, why would you do that? Why, why would you try to go back and forth between LLMs?

Dave Robertson (11:50.051)
Well, each one.

Dave Robertson (11:54.947)
Mostly because they're all large language models, but some are better at different things. I haven't gone into too much of the specifics, but a lot of what I'm reading is a lot of the anthropic models are better for like math and financial calculations. And so a lot of the big players like Vanguard, Fidelity, some of those folks, they're choosing more of those models because that's their realm, right? If you're looking for more generic things, a lot of chat interactions and chat GPT will...
is great for that kind of content, right? Dolly and those kinds of things are better when you're trying to generate images, right? So just depends on your use case.

Mike (12:27.629)
Yeah.

Mike (12:34.834)
Sure. Like I know the images and I've heard Anthropix better on code, quite frankly, than GPT is. And if I wasn't so, if I didn't feel so vested in GPT, I feel like, you know, it's been a last time. I finally know how to like manipulate GPT a little bit to get what I want out of it. so, um, part of me really just doesn't want to.

Dave Robertson (12:41.217)
Yeah. Yeah.

Dave Robertson (12:52.791)
Yeah.

Mike (13:4.386)
to do the work of learn what all the problems are with all the other aspects. OK, so you mentioned at work and y'all were using or y'all y'all are using copilot at work.

Dave Robertson (13:9.719)
Right.

Dave Robertson (13:20.269)
Yeah, we use the GitHub Copilot, which is specific to code, right? Windows or Microsoft has, I call them all the Copilots. People ask me, hey, do you use Copilot? The first question is, okay, which one, right? I got the office one, they got the windows one, right? So, yeah, I usually clarify. I'm not like writing in MS Teams over here. can you write me this, this 2009 T-SQL statement to generate a report or whatever. We use GitHub Copilot and that is one of the

Mike (13:22.477)
Yeah.

Mike (13:32.180)
Oh, yeah, sure. Sorry.

Dave Robertson (13:50.295)
the best use cases and well engineered large language model tool that I've seen to date. It's really, really good at a lot of things.

Mike (14:1.302)
So what are you all doing with it? mean, are you asking it to write code for you or to review code that it it matches a certain spec or standard or what?

Dave Robertson (14:8.387)
Um

Dave Robertson (14:14.829)
Yeah. That's one of the things that makes it stand out from a lot of the others. It has a large feature set. And so the chat is fantastic. I use it quite a bit rather than going to Google and Stack Overflow and those kinds of things, right? That was kind of the normal developer path for the last 20 years. Hey, let's go see what's on Stack Overflow. Somebody's done this. Well, now I just chat with GitHub Copilot and it's usually curating that and a bunch of other things to give me similar answers.

Mike (14:32.822)
Yeah.

Mike (14:38.221)
Yeah.

Dave Robertson (14:44.995)
But it's also very good at generating. Well, a lot of us are having to relearn some of these things, right? Yeah, it's like, you know, what happened to all my old tools? They go away. But it's very good at generating code. So a lot of folks are developers, specifically in the last 15, 20 years, they've also used ReSharper a lot, right? It's been a big kind of IntelliSense kind of productivity tool for generating code.

Mike (14:45.198)
kind of the death of my Selenium user group in some ways, but yeah.

Dave Robertson (15:14.529)
anticipating what's Copilot replaces a lot of that, almost all of it. And so it's able to kind of anticipate what I'm doing. They're capturing the full context of my developer tools. I use VS Code or same similar thing for Visual Studio, right? So it's able to gain a lot of insight and kind of anticipate and generate a lot of the boilerplate boring stuff, log statements. It also writes unit tests very well. So yeah, it's just kind of an all around good developer tool.

Mike (15:43.542)
And are we talking done that.

Dave Robertson (15:47.215)
It does a lot of things. because of all the... Yeah. Well, these days it's kind of all the things. So it's really good at SQL. It does PowerShell very well. I've been using it to have it teach me Python while I'm learning things like semantic kernel, Lang chain, and some of these things for a lot of these large language model integration tools and stuff that are coming out. So it's multi-language.

Mike (15:49.806)
I don't know which I'll, I would assume you're still in the C sharp world over there.

Dave Robertson (16:16.683)
It's very good at a lot of those things. It failed me when I was working in SharePoint, but I didn't really care because I don't really like SharePoint. So I was OK with that.

Mike (16:26.922)
Ave.

Dave Robertson (16:30.443)
Oops, did I save that up?

Mike (16:32.392)
No, it's just, you know, I thought you were a big fan of everything Microsoft. Yeah.

Dave Robertson (16:38.443)
Hmm. Only the things that work. SharePoint's not really one of them.

Mike (16:45.096)
You know, there aren't, know, after I guess I've been doing this thing for probably what? 25 years or so. Um, probably YouTube, but, uh, you know, it's, it is odd that out of all of those, and I've gone through a lot of shops, uh, in that time. And I think I can only remember two shops that were.net shops that were just full, full.net shops and dry financial was one of those. And the place I was at.
Right after Drive Financial Benefit Mall was another one of those, most everybody else was a Java shops, open source shops.

Dave Robertson (17:24.899)
And these days it's polyglot. It's all of it, right? I mean, we do a lot of stuff with Angular, a lot of stuff, know, JavaScript, I mentioned.

Mike (17:32.332)
Yeah, you're kind of all over the board now anyway, right?

Dave Robertson (17:34.753)
Yeah. So yeah, it's a lot of automation, YAML and you when you throw the DevOps tools and all this stuff in there, you just, you got to know like 20 different technologies these days just to keep up.

Mike (17:46.158)
So that's an interesting point. exterior is a a I don't know if it's a bank. You classify that as a bank or.

Dave Robertson (17:58.689)
No, they don't take deposits. so technically, or guess legally, it's not a bank. We borrow money to lend money. So it's just a, yeah, it's a financial lender. Yeah.

Mike (18:7.128)
finance drive, not more actually drive is a bank now, but okay.

Dave Robertson (18:11.585)
Yeah, pretty much. Yeah, they are now. Right. That changes things, right? Yeah.

Mike (18:16.896)
OK.
So, okay, interesting. have you ever used cursor? Have you ever tried cursor or used cursor or even know what cursor is?

Dave Robertson (18:35.117)
So interesting story. We're trying to do a POC with Cursor AI Studio, but they won't get back to us. Apparently they have such large demand. Yeah, we've sent out a requisition to them and yeah, they just haven't been able to get to it yet. that's on our list. Yeah, it's pretty cool.

Mike (18:45.955)
Wow.

Mike (18:54.368)
Yeah. So, I, I, I looked at cursor. Um, I, I looked at that company pretty big, but it's such a small company and
I don't mean this and I feel like an old man when I say this, but the, the people that are working at cursor just like fresh out of school, like just like almost just got out of school. Like if you, if you look at all the profiles of all the, and there's only, there's not that many people there, but, and, so was like, man, to, be able to tackle a company like that was such a small skeleton staff and the

Dave Robertson (19:20.907)
Okay. Yeah.

Mike (19:38.786)
I would imagine the demand that once, because once your name gets out there for the type of work that, that they're doing, I I would imagine just everybody would want to a piece, right? Hey, let me try it and see if it works and how it works. So, uh, so yeah, I would imagine you would be going through some, all kinds of growing pains, right? Like when, when you're dealing with a, a staff that young, it's

Dave Robertson (19:55.139)
Yeah, I think they're going through some growing pains right now. Yeah.

Mike (20:8.786)
What is a company? What is a company supposed to look like, right? Not just technical, but, you know, I've never done this before.

Dave Robertson (20:10.947)
Yeah. Right.

Dave Robertson (20:18.627)
Yeah. Yeah, they got one sea level. It does it all.

Mike (20:25.822)
So interesting. yeah, I was that's why I was I was I was wondering if cursor was in enterprises yet. I also know other companies, but these are these are a little different, like the cursors and the GitHub co-pilots are are really made for development shops to start using. Or at least this is my read.
that they're made for development shops to start integrating LLM AI functionality into the development shop. And there's some other tools out there like Replet. And I don't know, there's a couple of a couple others. Bolt and things like that. They're. Almost more of.
Hey, I want to make a product and just drive it, you know, all the way through to deployment, uh, within this one single application. And, uh, I mean, those are not really developer focused. They're, they're more of just, it's more of like what I was talking about earlier. Uh, I have an application that I want to make in, and chat GPT and
How do I break it up and make it right? So I guess maybe those would be.

Mike (21:58.824)
I don't know, smaller applications. I don't know. You certainly couldn't do something enterprise with something like that. So.

Dave Robertson (22:4.843)
It's, I think a lot of that's still trying to shake out these days, right? There's, like you said, there's different levels of it. If you've seen anything from Rider AI, that's kind of a all-in-one platform where you can kind of build your own integrations. But you don't necessarily need developers to do that. They'll take on that plug into your systems and integration. they basically host all the agents and all these things.
Whereas on the flip side, if you're looking at like semantic kernel, lane chain, crew AI, some of these kinds of things, that's more, you're enabling the developers that you already have to plug in a new skill set. Often it's usually Python, some of those kinds of things that may or may not be in-house. And then you can build your own, right? So you may leverage the LLM itself that's hosted by AWS or it's hosted by Azure.

Mike (22:53.336)
Yeah.

Dave Robertson (22:58.679)
But then you're building all the connecting tools, very similar to how you would write APIs and that kind of thing to say, okay, I'm going to call this LLM, but then I'm going to give it this function or this tool. And that makes it more of an agent. And so based on the prompts, now you can say, okay, go do this. and so now you're, you're enabling the same types of things that you're seeing people come out there with all of these different co-pilots in your own enterprise. And that's really where I think all the value chain is going to be.

Mike (23:28.014)
Do you see people writing APIs and things like that in real world shops through LLMs?

Dave Robertson (23:34.893)
A lot of it's out there, but yeah, so much of it is, it changes so fast. I think I started a proof of concept with some of these things and three weeks later they had breaking changes and I had to go change all my stuff, right? So it's just, it's happening so fast. A lot of it hasn't really settled yet, I don't think.

Mike (23:50.612)
It's so fast. It's so, it's so much to keep up with. If you're trying to keep up with this whole genre, it's, it's developing so fast that it's, it's, it's hard. It's hard to keep up with all of it. So, um, but it's neat. It's good stuff. Um, do you think, uh, we'll get to a point where

Dave Robertson (23:52.993)
So fast.

Dave Robertson (24:8.547)
for sure.
It is. It is.

Mike (24:19.520)
It'll be more of a BA in charge of a product or a BA in a product manager, maybe saying, Hey, I want a B and C.

Dave Robertson (24:32.099)
I'll answer that with a question for you, Mike. Do you remember Power Builder from the 90s?

Mike (24:34.638)
You

Mike (24:38.284)
Yeah. Yeah. Okay.

Dave Robertson (24:39.991)
Do you remember Microsoft light switch from like the 2018s, I think. Yeah, yeah, right. There's a lot of different things out there. So I call that the golden unicorn, right? Folks have wanted that for like, it seems like thousands of years yet for now, right? Yeah, can I kind of simplify things and just let a person who's not really technical kind of build some.

Mike (24:45.218)
Yeah, I get it. In front page, right?

Mike (24:57.486)
wrapper.

Dave Robertson (25:8.599)
build some stuff and put it together, right? Whether that's Visual Basic for applications and an old access database or whatever, right? You kind of put a friend in, you're right. You know, a lot of those tools have been around for a long time and some things they do pretty well, right? Yeah, right. So some of those use cases I think have been solved, but realistically, yeah.

Mike (25:16.608)
Yeah.

Mike (25:24.974)
Okay, all right.

Mike (25:32.056)
Yeah, like, like a basic storefront webpage, right? You can, you can do that in an hour without any technical expertise. Just, just hop on Wix and you can plug in your products and, make a good webpage. need graphics or whatever, probably. And, uh, you know, but that's maybe a good example of something that

Dave Robertson (25:37.091)
Yeah, you can go to Squarespace or whatever, right? Yeah. And you're done. Yeah.

Dave Robertson (25:47.063)
Yeah, Wix.

Mike (26:1.134)
in the early days in 1999 or so, right? Right. You had to have, that was a 10 man team probably. Right. And now it's not right. and I guess the devil's in the complexity, right? So, and you have an application that hits 10 other APIs.

Dave Robertson (26:3.894)
Yeah, right.

Dave Robertson (26:10.019)
It was non-trivial, right? Yeah, for sure.

Mike (26:28.522)
and 10 other web services and you have an authentication layer. You have multiple authentication layers. And. Right.

Dave Robertson (26:36.087)
Yeah. And how many devices, right? Are you going to distribute it to a watch? You're going to distribute it to a phone for your friend in? Yeah. It starts to get complicated pretty fast if you're talking like enterprise class, you know, stuff, right? Yeah.

Mike (26:49.110)
Enterprise, right? When you're talking enterprise, gets complicated. So that's probably one of the last things for testing too, is like mobile. You can't simplify all the diversity in mobile easily.

Dave Robertson (27:10.275)
a lot of diversity.

Mike (27:14.296)
There's, there's, is mobile still complicated and it's, and it's still hard to, to get right easily. So, uh, yeah. Okay. All right.

Dave Robertson (27:25.155)
Yeah, for sure. So back to your original question. Yeah, I think we'll see a lot of advancements in that. Like even today, Microsoft Co-Pilot can take a PDF or a data dump or something and generate a PowerPoint for me pretty well in just a matter of minutes, right? Whereas that might take me, you I'm not a PowerPoint guy, maybe six hours to do it, right? So I think we'll see some of those more tasks, simplified task-based things be optimized.
But I don't know, a lot of the AI and tools that I see out there, you you've got hallucination and all these kinds of things that you have to kind of watch out for if you're just getting into those things. It's more of an accelerator for the people that we already have as opposed to replacing a lot of the job titles and those people. Yeah.

Mike (28:15.510)
Yeah, but what's stopping it from becoming more than just an add-on? I would assume investment, right?

Dave Robertson (28:26.253)
Some of it, some of it. The other part I think is just time too, right? Yeah. It's going to keep progressing like you said. Yeah.

Mike (28:33.294)
Right. So, I mean, if you had a company that really wanted to put some money into it, right. Let's, mean, they put money into everything else in the world. They never seem to put enough money into SDLC, but they, put, you know, Uber gets a bazillion dollars, right. But if they, if, if you had a product company, I would assume that, that, I mean, had to put some Syria, $50 million in the creating.
an application, hundred million dollars into creating an application that allowed AI and but a diversity of technical compliments to come in at the same time. Is that going to happen? I can't think it hasn't happened ever in the past that we've had that much investment. Right. So it's been kind of

Dave Robertson (29:24.799)
Not yet, right? Yeah. Yeah.

Mike (29:32.366)
just bolt on stuff. okay. All right. So back to your book. I know we kind of went a long way, but on your what's your we talked about your first book. What's your second year? Is your second book the children's book?

Dave Robertson (29:48.631)
Yep, second book is the kids book. That was the second challenge I decided to take on with AI is the first one was generating words. Can the second one generate pictures, images? So it's a kids picture book.

Mike (30:1.804)
Yeah, how'd that turn out? I didn't I didn't read that one. OK, is this you're telling me to go buy it?

Dave Robertson (30:5.387)
Oh, yeah, well, I'll get you a copy. I'll, no, no, I just, I'll, we'll meet up at the happy hour and exchange some stuff and yeah.

Mike (30:10.682)
All right.

Mike (30:14.691)
Are they AI generated? They look like your typical AI generated images that came out or do they look more cartoonish type images that came out? I guess AI generated images are kind of cartoonish images, but.

Dave Robertson (30:27.611)
So.

Dave Robertson (30:33.475)
Yes. Ironically, this was the harder of the two to do. The first book is, you know, 60, 14 something thousand words, 14,000 words, 60 something pages. The second one is like maybe 200 words and only 34 pages. And so because AI is so non-deterministic, that was actually harder to generate visuals that were consistent throughout 30 plus pages.

Mike (30:43.971)
Yeah.

Mike (31:0.866)
Insistent. Right. You got to get the same redheaded little girl, right? And and all the so yeah. So is there a way that you can do that? Get the same without manipulating it yourself? Is there this?

Dave Robertson (31:4.791)
Yeah, right.

Dave Robertson (31:16.119)
So, yeah, we just talked about maturation of tooling, right? I know there's things that are coming that are going to enable things to be better, but one of my goals for now was to be, what can I do for free, right? The tools that are out there, chat, GPT, and lot of these things. wasn't using a paid license or any really additional tools. So I was using graphical editor tools and all these kinds of things that are just kind of out there, open, free. I think I used Microsoft Designer to generate kind of the base images.
And so based on that, I generated the images, but then there was a ton of editing that had to go through. But yes, to your point, consistency, what I wound up doing for some things was reusing some of the same images and multiple pages, essentially. But then there was heavy, heavy kind of user intervention. I was doing some manual photo editing to make that fit the other page.

Mike (32:5.921)
Okay.

Dave Robertson (32:14.381)
Some cases it was just a flip of the image, you know, to make it look a little different. But in a lot of cases, I had to go scratch something out and then put something in its place to now satisfy, you know, page 30, whereas it was different on page 10.

Mike (32:14.382)
Yeah, you cheated.

Mike (32:28.268)
Yeah, that does seem a lot harder because even when I describe and even when I describe an image to be produced exactly the same description, I get a completely different image. All right.

Dave Robertson (32:45.985)
Right? AI is deterministic or non-deterministic by nature. Yeah.

Mike (32:50.902)
Yeah, it would be interesting. It would be neat if you could sort of save a state or something and then be able to re able to use a state type of a thing. But yeah, OK, that's so how long was that book? How long did that book end up being?

Dave Robertson (33:8.956)
I think it turned out being 34 pages in length. But whereas the first one I completed started to finish in a week, the second one took me almost eight weeks to get it

Mike (33:20.290)
Yeah. you were just using Dolly?

Dave Robertson (33:21.697)
get it in the store, all these things.

Dave Robertson (33:26.849)
Yep, using Microsoft Designer, which is, yeah, I think it uses Dolly 4. May have been Dolly 3 at the time. Yeah.

Mike (33:35.406)
Okay. So anything, any lessons learned on that, David?

Dave Robertson (33:41.343)
Oh, a ton of lessons learned. Yeah, be careful what you ask for, right? Sounds like a lot of fun, but it's a lot of work.

Mike (33:44.376)
Yeah.
Where do you, could you, have you found that you could, well, let me ask this when you're using GitHub co-pilot, can you incorporate images into that? Like, let's say I want this to be what my design looks like. Let's take it. This is a really bad example, probably, but let's say you take a screenshot of some other application and say, Hey, I want this application. Can you use something like that and GitHub co-pilot and.

Dave Robertson (34:1.022)
It's, yeah.

Dave Robertson (34:14.915)
start here.

Mike (34:19.033)
And as a starter.

Dave Robertson (34:21.547)
Not so much images. They have added features recently where you can upload content. So a lot of it tends to be more data, right? I can upload a PDF or I can upload an Excel spreadsheet or something and then ask it questions around that content. But yeah, I haven't seen a lot of good tooling that says, other than the PowerPoint example I mentioned, kind of start here, right?

Mike (34:47.553)
Yeah.

Dave Robertson (34:47.680)
a lot of it's more database. And especially for images, I haven't really seen anything that says, hey, I want you to use this as your base image and then curate it for whatever prompt I give it. I'm sure it's coming.

Mike (34:59.934)
I haven't seen it for enterprise. I've seen it for, um, MVP type stuff, you know, but I haven't seen it. I haven't seen it for, uh, like an enterprise place, right? Where, and it, it could make a lot of sense if you're, if you're just, well, but you would have this in code anyway, if you had an existing page, page one, and you wanted to make a page two and

Dave Robertson (35:2.188)
Yeah

Dave Robertson (35:8.631)
Yeah, right. There's a lot of those.

Mike (35:29.910)
But you would have page one's code to be able to do that off of you anyway. maybe not, maybe that wouldn't be beneficial at all anyway. So I don't know, man. All right. So, so what's the, what's the third challenge in your, in your third book?

Dave Robertson (35:47.619)
Oh, the third book is not really AI. It's just something I started about 20 years ago and I want to try and finish it. One great thing that AI is really good at is generating one-off images. And so I'm generating cover art for the book. I'm doing everything myself essentially. So that part is beneficial, right? I don't have to farm that out or go pay $300 for somebody else to...
craft something and just kind of experiment with it and get whatever I want. But yeah.

Mike (36:19.382)
That's kind of neat. So you're a published author now too, right? There you go.

Dave Robertson (36:23.299)
I so. I don't know. I don't think it counts. I use chat GBT, Mike.

Mike (36:26.354)
So.
balance man. Do you have a what was it? A IBSN or there's a library Congress book number, right? Or something, right? Do you have that?

Dave Robertson (36:39.607)
Yeah, Balker. Yeah, I had to learn all of that in a week. Yeah, oh yeah. That was long, long nights that first week during the summer and trying to figure out like, oh my God, how do I publish a book? Yeah.

Mike (36:54.616)
So it's pretty amazing to say, I'd like to go to the Library of Congress sometimes. They're supposed to have every book ever made, right? That has one of those numbers in it. I would imagine that's a heck of a big library.

Dave Robertson (37:10.391)
You gonna take a road trip? It's probably all digital now, right? They just got a massive hard drive somewhere. Yeah.

Mike (37:13.503)
Especially with all this self publishing, right? So, I mean, people are creating stuff all the time. Dave's over here just creating images with... It's a book. Have you been able to sell any copies?

Dave Robertson (37:21.842)
Yeah. That's a bug.

Dave Robertson (37:31.940)
That's one of my current, it's a side hustle, right? It's not my day job. So fortunately, yeah, exactly. So I'm learning all of the marketing for all of this stuff. And that is just, it seems like the dark side, man. So yeah, I'm learning how to get better at it. But yeah, I've been selling a few books here and there. So mostly to family friends and I'll probably give away more than I sell. So there you go.

Mike (37:38.030)
It's just for fun, man. I get it.

Mike (37:58.412)
You got a copy in your office? Let me see. Show for the camera, man.

Dave Robertson (38:0.535)
See you!
Oh
Well.

Dave Robertson (38:11.000)
I got the ones that I've published so far. So I think you, yeah, that's the first one you have that one.

Mike (38:13.262)
Yeah, OK. Is that 2099? That's a really good image, David.

Dave Robertson (38:19.996)
This is the one that, yeah, it was pretty pricey. thought AI did fantastic at generating this cover, right? Then I had to go and do all the edits. Yeah, yeah, I thought so. No, no author pictures on this one. Yeah, yeah. You're right, I should get a turtleneck. That would make everything better. Yeah, of course, Bringing the 90s back.

Mike (38:27.328)
Yeah, that's a really good image, man. What's on the back? Let's see the back. Are you wearing a turtleneck on the back? David Robertson author.

Mike (38:42.353)
Black turtleneck, right?

Mike (38:48.043)
black and white photo, you know, about the author. You can make up a picture of what chat GPT looks like.

Dave Robertson (38:50.755)
the way you think, Yeah, I'm going to call you up for more tips.
Can you run my Instagram channel for these books? You seem like you're pretty good at it.

Mike (39:3.630)
That's.
Anyway, cool stuff, man.

Dave Robertson (39:8.471)
Yeah, well thanks man. It was fun.

Mike (39:10.424)
So, house church.

Dave Robertson (39:14.211)
Church is church. We do a lot of the stuff like you got here. Not so much on the podcast side, I think they do, but we do a lot of live stream. So I'm usually involved with a lot of that stuff. So I've got audio, video, running sound, making sure it sounds decent. I get to play guitar every now and then.

Mike (39:34.424)
Do you stay more on the AV side or do you stay more in the band?

Dave Robertson (39:40.320)
I'm a little, it's about half and half really. yeah, for better or for worse, I'm one of the IT support guys there. So usually, know, live stream is down or like the other day AWS had a problem, but so we couldn't run our live stream on Sunday morning. So we're, you know, sometimes doing tech support on for four hours or something on a Sunday morning. It just depends on the day.

Mike (39:40.610)
or both.
Yeah.

Mike (40:1.934)
You know, my father-in-law lives in a little city in Mississippi called new Hebrew in Mississippi. It's a really little city. It's a hundred people in the whole city. Right. And so, but his church is, uh, is live streamed and I, uh, I watched their church every Sunday. Right. Like who would have thought that we would be in an age where, uh,

Dave Robertson (40:8.695)
You okay? Yeah.

Dave Robertson (40:14.348)
Oh, it's tiny.

Dave Robertson (40:20.311)
Yeah. You tune in. Cool.

Mike (40:29.632)
You know, I mean, I'm watching this little, here we are in Dallas where there's mega churches everywhere. Right. And I'm watching every Sunday I'm watching service in a little town in Mississippi that has a hundred people in it. And it's a very, it's a Baptist, very traditional Baptist little community church. Right. So where everybody's singing, they're out of sync and they're, know, everybody's out of tune.

Dave Robertson (40:34.873)
Yeah. Right.

Dave Robertson (40:46.039)
Yeah.

Dave Robertson (40:51.363)
Okay, yeah. Yeah.

Dave Robertson (40:59.437)
Full experience, yeah.

Mike (41:1.999)
So, I, I watch it every week. Right. And so, uh, I think the life's the live streams. I don't, I don't know who started that or when that started. Uh, but it's such a, such a great idea to, have that and to implement that. So what do you, what are you playing in church? you playing a bass or drums?

Dave Robertson (41:19.747)
That's where technology shines for sure.

Mike (41:29.356)
or lead guitar.

Dave Robertson (41:33.569)
I am not that talented, Mike, thank you. But yeah, I'm usually not that guy. I would if I could, but yeah, I'm usually on the acoustic guitar. I keep it simple and I just kind of follow everybody else.

Mike (41:36.264)
You're playing old Petra songs, you know?

Mike (41:49.750)
Okay. Do you you just go to like once a week on Sundays or do you do like Wednesday services and stuff like that? Does the band always stay like for every every time they have a service or is that how does that is that consistent?

Dave Robertson (42:7.693)
Yeah, we just have a pool of volunteers and there's guitar players, there's usually a bunch of us floating around, so different people will play on different Sundays, but that's usually about it. We're not really involved in like a thousand activities during the week because we all have day jobs and things like that.

Mike (42:29.390)
Okay. So do they, do they end up just taking whoever is at service? I, I saw, I've never been in the band in church before. Right. So I don't play anything. I'm like, I don't even play, I don't even play piano very well, but I have a keyboard, but I don't even play piano very well. But, um, but anyway, I've always kind of wondered because it's a volunteer.

Dave Robertson (42:38.583)
You should join the band, Mike. Come on, I'll get you some egg shakers, Get you the egg shakers. Yeah, there you go.

Mike (42:57.698)
thing. And so the downside with volunteers is, you know, it's, uh, it's volunteer. So you never really know who's going to show up until they sometimes show up, I guess. And I guess maybe, you know, that for Sundays because, you know, people have routines, but I don't know. I was just wondering.

Dave Robertson (43:16.247)
Yeah, it's kind of important. Just like everybody else, we use technology. There's a platform out there called Planning Center online where we can kind of schedule people in advance. And you can kind of, you know, if you're going to be out of town that week, you know, give courtesy, let them know ahead of time. yeah, there is some structure to it, but you're right, it is all volunteer. It's not our day job. So, you know, I may get to play guitar five minutes a week. Yeah.

Mike (43:41.594)
I didn't even think about that. I didn't even think about the structure that they would have to keep everybody in sync with this is what we're going to play this week and this is going to be what's on the agenda. I'm sorry, man. I know we delved into church maybe, but I was interested.

Dave Robertson (43:48.963)
Yeah. Yeah. Kind of have to these days, yeah. Yeah. I'm not that good to just show up and play something.

Dave Robertson (44:4.419)
It is, man. Yeah.

Mike (44:6.902)
I, I, I, I, being someone that's been in church all, all their life, you know, that's, it's also interesting that like, well, I have a really good, really good friend and I grew up in Louisiana, Louisiana's 99.8 % Catholic. Right. So, uh, I have a really good friend of mine that's a Catholic in Dallas and, he always feeds me, uh, stuff from his service a lot.

Dave Robertson (44:18.893)
Yeah. Right.

Dave Robertson (44:24.055)
Yeah. Yeah.

Dave Robertson (44:30.913)
Mm-hmm.

Mike (44:36.214)
I'm Baptist, obviously, but I don't care. It's just interesting to see the dynamic and how the dynamics are different. And and all of that. So anyway, all right, man.

Dave Robertson (44:46.241)
Oh yeah. Yeah. Yeah. All good things.

Mike (44:53.166)
All right. Anything else going on?

Dave Robertson (44:56.459)
Well, I wanted to ask you some questions about automated testing. So you're the guy who knows all the stuff. I just kind of dabble in it. yeah, what does that look like these days? Because I know there's a ton of tools out there. And some people are still doing a lot of the manual stuff. Other folks are automating all these things. How do you see that maturing?

Mike (44:59.862)
Oh, sure, sorry.

Mike (45:17.794)
Yeah.

Mike (45:21.667)
Well, first of all, do you guys have a QA team at XTier?

Dave Robertson (45:26.657)
Yeah, we have two actually. there's a large set that do the manual testing and there's a very small set that do the automated stuff.

Mike (45:29.110)
Hey, all right

Mike (45:35.156)
So what do you all do at Xter?

Dave Robertson (45:38.323)
It's a lot like drive financial sent in there back in the day. Folks are just kind of split off onto different products and requirement chain, all this kind of stuff. They create test cases, which are usually manual, and then they execute.

Mike (45:42.285)
Yeah.

Mike (45:49.410)
Well, I guess let me ask a little bit more pointed. you do do you do mobile testing?

Dave Robertson (45:57.729)
Not really. We do more mobile compatibility testing, but not so much the device testing. So they're checking viewports and stuff like that, right, for kind of mobile scaling and compatibility. But no, they don't. I think they looked at browser stack and a few of those things. I forget the other one. The big popular one starts with an S, but yeah, no real devices. Yeah, yeah, yeah. They don't really tie into.

Mike (45:59.424)
Okay. Is there a need?

Mike (46:9.740)
but not on real devices.

Mike (46:22.040)
Soft Labs.

Dave Robertson (46:27.309)
real devices.

Mike (46:30.518)
Well, some think they can.

Dave Robertson (46:33.953)
No, mean, yeah, I'm saying Exeter doesn't, so we don't license any of that stuff.

Mike (46:37.864)
Oh, OK. OK. So we need for that.
I'm assuming not since someone hasn't made a fuss about it.

Dave Robertson (46:48.063)
So we don't even have a mobile app for one. No, there's no app. We have a website that's built using Angular and Ionic, so it's mobile friendly. I took that and added more Ionic to it so that the plan was we would build once and deploy many. And I got that to work, but nobody really progressed with it.

Mike (46:51.182)
Do you have an iOS app? Does. OK.

Dave Robertson (47:14.019)
I think part of it is because it's a very, very simple application. You can basically log on, see your loan status, and click a button to make a payment. And then there's not much else you can do. It doesn't have a whole lot of features, right? So I guess maybe it's not important to them just yet.

Mike (47:30.604)
Well, an app, an exterior app where you can log in and make a payment seems like a legit thing to do. It seems like an app, like a web app that you would want to have. I don't know who your CIO is, but it seems like that would be a good, it seems like that's something your business would really be attracted to. Um, uh, and that doesn't seem that hard to do to be quite frank, but, uh,
But yeah, then you'd have to incorporate mobile testing and you probably have some kind of testing that just does just test the web page right now. Right. Not really much more than that web page on a mobile device kind of testing. I bet you have to use an emulator probably right now. But I don't know. But OK, so.

Dave Robertson (48:19.159)
Yeah.

Dave Robertson (48:23.239)
Um, yeah, I don't know. I don't know much either. Yeah, they do it, but not much.

Mike (48:28.086)
So do you all have I mean, I assume you've got a ton of APIs that you're testing, right? You're testing webs, web apps. Do you have desktop apps that your loan originators are using or is that web?

Dave Robertson (48:44.545)
No, yeah, it's everything is weird. Yeah, we don't, I don't think, does anybody build desktop apps anymore? I don't know. Yeah.

Mike (48:47.368)
Oh, okay. Okay. I mean, I'll tell you one of my favorite gigs and I was only at drive for two years, but one of my favorite gigs was drive and it wasn't really the company. In fact, I thought the company was kind of crappy to be honest, but it was the people that were there. You and Stephen Bell and Paul and Shane.

Dave Robertson (49:5.690)
Not very exciting, right? Yeah.

Dave Robertson (49:15.031)
Yeah, we did some cool stuff together. Yeah, it was fun.

Mike (49:17.602)
You know, I, and I, I, I built a little application called bug smack or mini smack, right. Where, it just, it just a little application that's set on your desktop. And it was just an RSS reader that, that went out and ping the, the, the bug database and brought back, um, you know, I was, I was awesome back then. I still have it.

Dave Robertson (49:25.005)
Yeah, yeah, I remember that.

Dave Robertson (49:33.603)
Uh huh.

Dave Robertson (49:39.233)
That was fantastic. We all loved it. We cheered your names in the halls. Look what Mike has done.

Mike (49:47.438)
I have a little, cause I made a little image of many me, right? But, but it was my face, right? And, and that was, that was my image for the, like the little, the icon that I double clicked on it. But I, I loved, I loved that. I loved that gig. Not as much for the company. Um, but, uh, but the people that were there and, and, you and Steven, uh, Lori.

Dave Robertson (49:59.704)
Now you're a legend.
Yeah.

Dave Robertson (50:10.519)
Like you said, we had some fun times, yeah.

Dave Robertson (50:17.719)
Yep. Not a folks. Yep.

Mike (50:17.818)
And, uh, Shane and, uh, Garman was fun while he was there. So, uh, uh, but anyway, uh, Paul, I still talk to Paul these days. So he's running a casino now, man. Like if you're listening to this, Paul, who never gives me a free room when I go to Vegas.

Dave Robertson (50:25.911)
Mm-hmm.
for sure.
Yeah. He's a big time. Yeah. Sharp guy. Yeah. Crazy, right?

Mike (50:46.094)
Thanks.
So but yeah, that was was a fun time. It was a good it was a good gig. You're asking around QA and you know, there's a lot of stuff. There's a lot of stuff going on in the AI world since AI started coming out. You can take requirements automatically create test cases from requirements. A number of tools enable you to do that.

Dave Robertson (50:48.963)
Okay, good.

Dave Robertson (50:55.287)
Yeah.

Mike (51:17.496)
probably one of the most excuse me, one of the most used things that are out there. There are a couple of companies out there that have completely autonomous testing going on where a short version of that would be I don't basically spider your website.

Dave Robertson (51:23.085)
Okay. Yeah.

Dave Robertson (51:47.224)
Yeah.

Mike (51:47.642)
And, uh, and, and basically record that spider that they do basically. Right. And so they'll run every instance that you could possibly do. There's other companies out there that are monitoring. Yeah. There's other companies out there that are monitoring production and, and then recording scripts based off of what they're monitoring. And, and then they can use that for your regression. Right. Um,

Dave Robertson (51:55.235)
Okay.

Dave Robertson (51:59.747)
That's more.
Yeah, yeah.

Dave Robertson (52:12.597)
Okay, do some replays. Yeah.

Mike (52:17.300)
But a lot of that is still a lot of the really good stuff is really still in its infamously. It's still in its it's really early stages. But it's it's coming together. It's just coming together a little slowly. And it's it's really just the I think the biggest thing is just the VC money isn't about isn't there if they if they had

Dave Robertson (52:28.898)
Yeah.

Mike (52:45.352)
a decent amount of VC money to put into that. ⁓ You could write ⁓ like ⁓ when I think of an autonomous just. Basics of an autonomous SDLC. You have requirements ⁓ after you have requirements, can you code from your requirements? Could you make test cases from those requirements? Could you ⁓ automate from those test cases to ⁓
taught, you know, automation script and then on production, can you monitor production and make automation scripts from production? You can do that. It takes effort to do that. Uh, and then if you were, let's say you had a monitoring tool in production and you even had clients running in production, like let's say you were on the test team.
and had clients running yourself. So you're catching JavaScript errors and things like that. Could you feed all those errors back into your JIRA system or, you know, your and and then be able to throw them into sprints, right? And and put them in your backlog and throw them into your sprints. Like, we're not that far from all of that.

Dave Robertson (53:47.188)
Mm-hmm.

Mike (54:13.191)
You need a consistent, constant effort to be able to do that, but we're not far from that. The problem is the investment. Another problem is tools that are out there that serve a lot of the legacy. Like let's say you had a
$50 million, $100 million. Let's say you had a $100 million test product company. And you you're, you need to make products to keep that engine going. That engine's making money for you. Right. And, and you need to, you need to be able to innovate off of the stuff that you have, the stuff that has adoption right now that you.

Dave Robertson (54:52.941)
cash flow, right?

Mike (55:4.416)
You know, have your clients out there saying, Hey, I want to, I'll buy more of this. If you put a B and C in it. Right. You have to make that stuff. Right. And so that this other stuff is important and can be done, but it gets lower ranked on the, uh, the priority list because there's, know, you can have immediate adoption from.
these incremental smaller updates. So then your big game changers from a test product company would come from your smaller innovator startup companies. Right. But then you have to have the VC money to be able to be able to do that. Uh, and you need a decent amount. I, know, you don't need like some VC guys saying, okay, here's $2 million. Go change the world. Right. You may not throw another zero on that.

Dave Robertson (55:45.507)
for sure.

Mike (56:2.402)
right? And so it's a much bigger startup than just this little ad hoc. So there's a lot you could, there's a lot changing in the testing world. There's a lot staying the same, David. So, you know, I did a post on LinkedIn. Oh, it's probably been a month ago, but

Dave Robertson (56:2.603)
Yeah.

Mike (56:28.974)
You know, I just, I just queried the heck out of indeed. And, and went down every testing tool that I could think of, uh, that I had time to, and I probably put about 50 testing tools on there. there's probably about 400 testing tools out there, right? But I just queried about, uh, 50 or so, maybe more than that. Um, of those and.
I just went down the list and queried indeed on Sauce Labs, queried indeed on BrowserStack, queried indeed on Selenium. you know, the mass, the masses, I look at those job requests as a reflection kind of, of who's using what in the market. Right. So
Uh, maybe that's not a perfect way to do that, but I, I, I, I, I find that's probably more honest and reflective than like a GitHub stars or any of those kind of, kind of fake made up things, right? This is actual what people are hiring for. And if people are hiring, then people are using. Right. And anyway, like Selenium was way on top.

Dave Robertson (57:41.613)
Right.
Yeah.

Mike (57:56.948)
of those testing with selenium was like.
At that time, were like almost 3000 jobs open for Selenium. And everything else was like the, think the next closest was a couple of hundred. Right. And so, uh, you know, you, think of all these tools in these AI tools and these AI and these, these product companies and what they're coming out with. And just, just remember like Selenium doesn't really have any AI in it at all. Right.
Selenium still up here with huge adoption. And what we're talking about is such a small, small segment until you get that Uber like movement, right? Or like something kind of goes viral. You get something that works and then just goes, just starts hitting, you know? And so anyway.

Dave Robertson (58:55.048)
What are some of the catalysts though to bridge that gap for higher adoption? Right? Because all these tools cost money. That's usually what I hear, right? Where's the budget? Where's the people? How do you fit those two together? Yeah.

Mike (59:0.590)
Um, yeah. So I, yeah. Yeah. So, and, and, and not all these tools are crazy expensive. Uh, like the one company that I used to work for recently, their licenses aren't that expensive. Um, and probably they're good. They're tie in was.
that they had selenium on the backend and they made a low code front end for selenium. right. Uh, so, and I, I, so that was, yeah, yeah. You have to find your angle, I guess. Right. Like the sauce labs of the world. One really difficult thing and, and QA that hasn't
I haven't seen a way to out AI it yet. I haven't even thought of a way to out AI yet has been mobile devices. Like, I don't know how you out AI. You have to physically test on a mobile device. I don't, I don't know how you out AI that. So there are emulators out there, but people that want to test on physical devices and want to test on physical devices, they don't want to use an emulator. Right.

Dave Robertson (60:19.789)
Yeah, it's not the same. Yeah.

Mike (60:26.632)
And so I think that segment, like you mentioned, sauce, stack, Lambda test, Appli tools has a good cloud, uh, tool. There's a number of her force as, as a good cloud tool. Uh, there are a number of those, uh, but I think that's a good, strong segment. Now you have to have other, I think one of the things you have to have is a platform of a lot of all these products and you have to own not just
one of these silos, but you have to have a platform to do all of your stuff on. So that a testing company just has to make one deal with one platform guy and he has test cases and web automation, desktop automation, API automation, certain, you know, all your services automated, uh, your cloud testing automated. Uh, really there's probably about six tools that a tester, maybe seven.
that a tester, yeah, test data management, too. So maybe seven that a tester really needs. And you just need a good platform to be able to offer all of those. But anyway, I hope that did that sort of give you. I know I probably went way on probably.

Dave Robertson (61:43.096)
Yeah.
Yeah, no, that was good, but I didn't hear anything in there regarding security. I know probably almost a third of my job these days is dealing with something for security. And some of that is integrating some security testing tools into whether it's scanning code, you're scanning live APIs that are running, penetration testing, all these kinds of things.

Mike (62:8.214)
Yeah. So what do you all do for security? You can't talk about it because it's

Dave Robertson (62:11.925)
All of it. No, yeah, we, I tell everybody, know, finance organization, we're a high value target, right? So we tend to maybe do more than most people. I get a little paranoid some days, but yeah, I we scan all of our code every single week with an automated tool, whether it changed or not.

Mike (62:22.510)
Sure.

Dave Robertson (62:36.749)
There's usually something that comes up every single week. A lot of it's not our stuff. We're using some open source package or something that's been identified by the new flaw or something, right? So now we've to go upgrade the version. So, you know, it's all that kind of stuff, but.

Mike (62:46.488)
Okay. So you're, you're just, yeah. You see you're scanning for, um, uh, vulnerabilities, right?

Dave Robertson (62:55.451)
Yeah, that's more on the static analysis side, right? Yeah, it's just looking at the actual code that we write, some of the packages that we integrate with. But there's the other side of it too, the dynamic scans as well, right? It's running for, how is that resource, whether it's a virtual machine or a cloud resource, how is it configured? But what's it, based on known vulnerabilities and things like that, are you using authorization and authentication or is it just wide open to the world? Those kinds of things as well.

Mike (63:23.022)
Yeah, yeah, sure.

Dave Robertson (63:24.397)
Yeah, a lot of that.

Mike (63:26.239)
Yeah, have you? Is your test team doing that or developers doing that when you're?

Dave Robertson (63:33.997)
Well, a little bit of both. So yeah, I'm mostly the DevOps team. So it's me and a few other folks. We set up the integration. So yeah, it's all automated. It runs on a schedule. So that part's kind of nice for the static stuff. The dynamic side, our QA team actually manages that. And they have a tool where they point it at a URL, click a button. It does the spider thing that you were talking about, to kind of look for things. Yeah.

Mike (63:36.406)
Or your DevOps team doing that, I guess.

Mike (63:59.788)
Are you all in sprints?

Dave Robertson (64:3.799)
They do sprints.

Mike (64:4.172)
Yes. So I imagine in like at the start of the sprint, they say, hey, I think we need to do some additional security testing on this, this and this. Like your your QA team says that.

Dave Robertson (64:15.395)
It's usually all at the end. Yeah. It's usually all at the end towards the latter stages. Yeah. It's the last thing to go, man. You know this. We're almost done. Scan it. Yeah. Yeah. It's in their release plan. Yeah, for sure.

Mike (64:20.462)
Yeah, yeah, of course.

Mike (64:25.160)
But, that's in the, that's in their scope is what I'm saying. Do they mention that as part of their scope? Yeah. Okay. Yeah. So, uh, you know, it's a point, you know, the vulnerability scans are, pretty basic and pretty common. Um, but beyond that, right. Like, mean,
That's a good point. Where is the investment on that side from a testing perspective? There's a lot to that if you wanted to dig into that. And yeah, there's not a lot of investment. That's not where the...
That probably should be, that would be a good segment to carve out. Now that you said that.

Dave Robertson (65:15.555)
I see it coming up more and more. mean, I'm constantly getting hit up with.

Mike (65:18.530)
Yeah, I think it's getting worse, right? Like, I think Microsoft said that they get like, uh, uh, what was it? Like 8,000, uh, security, uh, attempts a second, uh, a second. Right. So you're right. It is obviously their systems have to be pretty freaking good right now. Right. But

Dave Robertson (65:33.901)
Yeah. Right? Yeah.

Dave Robertson (65:44.387)
Oh yeah, and there's always a new security tool to plug into the mix. Like I never figured out, me as a developer would have to figure out what a application firewall is and how to configure it. But that's the thing.

Mike (65:54.072)
Yeah. And the risk really wouldn't be with Microsoft. It would be with your small to medium business, right? It would be with a small shop that doesn't have the heads, doesn't have the resources, just trying to get something out. Right. So.

Dave Robertson (66:5.559)
Yeah, right.
Yeah. Right. And like you said, that's where the investment innovation has to be there so that folks can, you know, buy the tools, plug them in relatively easily. Right. You don't need a full security team of, you know, 2500 people to secure your business.

Mike (66:24.876)
Yeah, that's a good observation, David. I'll talk, I'll talk to some people, see what I can do for you.

Dave Robertson (66:27.629)
Yeah, thanks man.
Okay, sweet. Mike knows some people.

Mike (66:34.894)
We will, I will have a part of this series. I will have a VC, a person coming in as part of this series. so we'll talk about investment areas and what they view is as important. And you know that I think the big thing, a lot of the VCs you'd be really surprised. A lot of the VCs just have no idea. Like they

Dave Robertson (66:40.172)
Okay.
Very nice.

Dave Robertson (66:48.951)
Yeah, sweet.
Right.

Mike (67:4.104)
Even the real basic stuff, like they don't understand a lot of that at all. And so like when I get in conversations with them, they ask me really basic stuff. Like, what is this? And like, how does this interact with the market? And would this be a good thing to have? know, and it's, it's, it's, it's interesting. And, but what's also interesting is to see.
the thought process that they make to when they decide to invest in something and really who they're talking to, to get their advice and their insight on what this market is doing or how this market's going to evolve and things like that. Because sometimes the people that you'd be really surprised at the people that they get advice from. It's not
Well, like, how did you come to that conclusion? anyway, but anyway, what else do you have for me, Dan?

Dave Robertson (68:5.299)
Yeah, sure.

Mike (68:14.318)
You took notes, man.

Dave Robertson (68:14.680)
Man, I don't know. I think I'm about tapped you out. I'm sure I have a ton of other questions, what's life like in the podcasting world?

Mike (68:24.056)
Alright man.

Mike (68:30.126)
Um, uh, so since Joe Rogan and Donald Trump got together and had a happy discussion since the election, a lot of people jump into podcast, uh, space, but I have to say it's, uh, you know, whatever you jump into this extracurricular, that's not your job, you know, whether it's writing books or

Dave Robertson (68:42.443)
Okay, is it coming back now?

Mike (68:58.774)
AI books or whatever, you have to really like what you do and really have to find fun in what you're doing or some, you have to get some kind of satisfaction that Selenium user group that I've done for 20 years, right? mean, if you did that for money, that would never, that's you, you would have been sorely
disappointed after the first year, even after 20 years, you'd be sorely disappointed. Um, but, uh, so whatever you'd sort of decide to do, you need to have, I guess, the passion to sort of, you have to have the, the want to do it. Right. It's like, it's like playing in the band and church. And I know that is not just the passion to do that, but that's a, I know that is a service aspect.
as well, right. But, uh, but, uh, it can't be something that you're, you're doing for, this is the new thing or this is, I'm going to make a million dollars doing this, right? Like you're not going to make anything doing anything. Right. So just let you know, might get my kids cost my, my kids college cost a hundred grand a year.

Dave Robertson (70:6.211)
Yeah. Obligation, right?

Dave Robertson (70:12.417)
that you did. You didn't make a million dollars.

Dave Robertson (70:18.413)
Just shy, just shy.

Mike (70:27.886)
And I drive a 2012 Cadillac Coupe with 220,000 miles on it.

Dave Robertson (70:36.803)
You're doing well, Mike, doing well.

Mike (70:38.960)
I'm not making shit.

Dave Robertson (70:40.961)
You

Mike (70:45.046)
Uh, thanks man. I could, I could use some, I could use some being some beer and wings, man. That would, that would go well. Y'all still have a crazy deployment, uh, things going on. It sounds like a lot of the drive team is at exterior now.

Dave Robertson (70:45.545)
worry I'll spot you at the next happy hour. It's on me.

Dave Robertson (70:56.419)
Exactly.

Dave Robertson (71:7.099)
Yeah, yeah, it's a lot of the same folks. There's a mix of various other folks. The finance is huge, as you know, in the Dallas-Fort Worth area. So it's a mix of folks. yeah, a lot of the same folks, a lot of the same patterns, practices. Still a lot of night deployments and stuff like that, which I don't quite understand given the tools that we have now. But fortunately, I'm not involved in most of those at somebody else.

Mike (71:17.325)
Yeah.

Mike (71:33.686)
Yeah. So one of my things that I didn't really like at drive were those Friday night, every Friday. And we used to release like every Friday and it was every Friday night. It was like,

Dave Robertson (71:43.394)
Yeah.

Dave Robertson (71:47.523)
Crazy, right? Who does that? Or a holiday.

Mike (71:51.046)
stays up and they were late. I, I'm not going to mention some of the places that I remember that we used to go or do right. Waiting for deployments to be finished. But, but, uh, like we, there'd be like midnight on Friday night that we would have a deployment. And that was, that was kind of the, the common thing. And, uh,

Dave Robertson (72:1.219)
Yeah, we do.

Dave Robertson (72:13.281)
when you make a run to Water Burger and get some breakfast tacos and sit back and enjoy the show.

Mike (72:18.234)
Yeah, man, when I was younger, I just get really frustrating these days.

Dave Robertson (72:25.259)
Unfortunately, we have the tools and the technology, so we can at least do it from our own house these days. But like I said, I'm not a fan of those those nights either.

Mike (72:32.086)
Yeah. Yeah. You have quick builds. you still have quick, do you have quick bells these days? You should have quick builds. There's no more, you're not waiting for four hours, six hours, eight hours for a bill or not eight hours. That's a long time, but you're not waiting four hours. Even if you're running an automation and your CI for your bill, which sounds like you're doing some of maybe, but even if you're doing that, you could do that in parallel.

Dave Robertson (72:38.795)
Oh yeah, everything, everything I built is automated. yeah.

Dave Robertson (72:48.863)
It deploys itself.

Mike (73:0.878)
parallelization, right? And so that should be snappy, right?

Dave Robertson (73:7.329)
too old for the manual stuff anymore, so yeah, I've automated all of it.

Mike (73:10.764)
Yeah. Remember smoke testing the whole manual of everything that smoked us in the whole thing.

Dave Robertson (73:17.027)
Yeah, can't do that anymore.

Mike (73:18.626)
Yeah. Yeah. I miss, I miss some of that, man. I miss Steve Bell, you and Steve Bell. and, and I went into that job. Uh, one of the reasons why I left that job was because I was in consulting before I went to that job. And, uh, you know, I was, I was paid well back in those days. I was paid well for consulting.
But you learn pretty quick that you're not the same rate that you make in consulting isn't the same rate that you make in a normal gig. Right. And so I went back in those days. think I went from one hundred and seventy thousand a year as an independent consultant working just telecom billing systems.

Dave Robertson (74:11.235)
Big difference.

Dave Robertson (74:16.547)
You

Mike (74:17.550)
Went from that. Yes, I was doing QA, some development for that and went from that, it only to drive. And I like, I think it drive. I don't remember exactly how much I made, but it was under a hundred thousand a year. was like something like 80,000 or something. And so part of my issue with, while I was at drive, wasn't even drive related.
You know, drive didn't do anything. It was me dealing with crap.

Dave Robertson (74:54.529)
Right? Kids gotta eat.

Mike (74:59.094)
Right. And I'm just struggling. Well, how do I get somewhere close back to where I was, you know, from a financial perspective? But I learned, I still learned a lot at drive. Like Steve Bell used to come over and we'd talk about different ways to test. And, uh, he would, he would impart new things and, me I was like, Oh, I never tried that before. I never, I never tried this before.

Dave Robertson (75:5.699)
Yeah. Yeah. We all do that.

Mike (75:26.606)
And I used to upset people because I would give people would ask me, how long is this going to take to test or how long is that going to take to test? And I won't name any names, but I said, well, who's developing it?

Dave Robertson (75:44.621)
Yeah, right.

Mike (75:46.834)
And if it's this person over here that's developing it, it's going to take about a week longer. mean, if it's Dave or if it's Steve Bell, know, I may find, I may get lucky to find like one bug or two bugs and I'll have to really look to be able to find it, you know, but if there's another person out there that's developing it, it's going to take

Dave Robertson (75:57.507)
just being honest.

Mike (76:16.852)
a lot longer and people used to get upset with me and I didn't really know how to. Well, I think I did say that probably said that to you a couple of times on, well, this is why there's a variance, but I could tell you, couldn't, you know, it's not something you want to share with everybody, but anyway, man, that was a good time. I enjoyed it.

Dave Robertson (76:33.421)
Yeah. I like your honesty, It was. It was fun. It was fun. Yeah.

Mike (76:39.956)
All right, Dave. That's all I got, man. Thanks for hanging out.

Dave Robertson (76:44.259)
Cool. Fantastic. Thanks for the invite. You got a good setup. It's a pleasure. Yeah. I like it. It's on me, man. I appreciate it.

Mike (76:48.842)
And we'll meet up again and catch lunch. All right, brother.
All right. Hey, good luck with the book.

Dave Robertson (76:59.437)
Thanks. Thanks. I'll let you do all the marketing. Oh yeah. Hopefully in January.

Mike (77:2.414)
Well, let us certainly let us know when it comes out and and we'll we'll make sure we let people know about it. OK. Our brother talk to you later.

Dave Robertson (77:10.947)
Okay, sweet. I like it, man. Take care. Good talking to you. Yeah, ciao.