ScreamingBox Technology & Business Rundown

Dave Erickson 00:03
As a developer, you may think AI is going to kill coding. So then what are you going to do to find work? On this ScreamingBox podcast, we're going to discuss how software development is changing and what developers must do to stay relevant. Please like our podcast and subscribe to our channel to get notified when the next podcast is released.

Dave Erickson 00:45
Artificial intelligence is rapidly transforming software development, automating tasks that once consumed hours of developers' time and is fundamentally changing how software is designed, built, tested, and maintained. So how are developers going to save their jobs from AI? Welcome to the Screaming Box Technology and Business Rundown Podcast. In this podcast I, Dave Erickson, and my technically savvy co-host, Botond Seres, are going to explore how software development is evolving in the age of AI.
And how technology professionals can adapt in order to remain relevant, with James Edgel, founder of SoftExit. James is a technology entrepreneur, software developer, and a business leader who has spent years helping organizations build, scale, and modernize software solutions while navigating the constantly changing technology landscape. He is the founder of SoftExit, a configurable business application platform for building CRM systems, databases, dashboards, reports, workflows, and custom business applications without traditional software development. Throughout his career, James has witnessed multiple waves of technological disruption, from traditional software engineering through cloud computing, automation, and now the rise of AI assisted development. His expertise spans software architecture, application development, business strategy, developer production, and technology transformation,
giving him a unique perspective on how the role of software developer is changing and where new opportunities are emerging. James, welcome to the podcast.

James Edgel 02:18
Thanks Dave, Thanks for having me here.

Dave Erickson 02:21
To begin with, your life could have gone in any direction. What led you in a technology direction?

James Edgel 04:21
Well, it was a movie actually. it's kinda funny. We were talking earlier when we met about the MSI. I saw a movie called War Games and I was intrigued by the whole entire technology aspect of it, you know, the the coolness of it. It was you know, when that movie came out, it it was a different time. None of this was available and I thought that was literally the coolest thing in the world.I was on a path to become a pilot and that took me off the path completely. I made a complete shift in the technology. I thought, Wow, that is frickin' awesome. Went out, bought a computer, and it's been, you know, downhill ever since, so.

Dave Erickson 03:14
Yeah, we might have to introduce your computers on the podcast as well.

James Edgel 03:19
Yeah, yeah, my wife is like, I'm like, so excited to have these. She goes, What the hell do you use those for? They're old. And I'm like, You don't understand.

Dave Erickson 03:30
There are some pieces of software that I really love that will only run on an old computer. So, you know, there you go.

James Edgel 03:37
I have the urge to play Stop the Bit, you know. There's nothing else we'll do.

Dave Erickson 03:44
You know, AI has definitely come a long way from war games. and I think, you know, I remember seeing that movie when I was young as well. And I can't, I never envisioned it being where it is now. Maybe you can talk a little tiny bit about your experience with AI, how it kind of began and what you've done a little bit with AI, kind of a background as to where you're at.

James Edgel 04:12
So yeah, sure. Well to tap on what we were talking about earlier, was war games the first warning of AI, I think? It was. Didn't that predate Terminator?

Dave Erickson 04:20
It did, yeah.

Botond Seresl 04:21
It did, I'm pretty sure. By a decade at least.

James Edgel 04:25
Yeah, AI, I got involved in it a little bit because I saw something on YouTube that said Google is dead, right? And this is the technology that's gonna replace it and I and I'm thinking, no way. So I watched a video and it introduced me to ChatGPT and I became intrigued. And I started to play around with it and it was an amusing little toy at the time. It could answer a lot of questions and I didn't see at that time any of the value of what it does now. I, in fact I think even OpenAI admitted that they were surprised by the interest in their product that they didn't think it, it did a lot when they first introduced it and then people got their hands on it and started to figure out how to apply it. And then, you know, the rest is history, right? We've, we've learned that it is quite capable of doing a lot of things. And over time it has improved. So I got more deep involved a couple of years ago when I learned it could code and I was intrigued by that and I started experimenting around with it and wasn't extremely impressed by its capabilities back then.

Dave Erickson 05:34
Well, you know, AI is definitely transforming the landscape and it's transforming software development probably faster than any other technology shift in the past. From where you're looking at it, you know, what are the biggest challenges that are happening right now that the developers kind of need to understand when it comes to software development?

James Edgel 05:58
Yeah, I think their world is completely changed and it really you know, as a coder and I still code and I love to do it and you know, when I come home I, I code as, as a hobby, right? Most people do sports or do other things. I'm a complete nerd, but ,I I gotta say that looking at the output of the code and working with it that I can't see going back to manual coding ever again. it is
enormously impressive on what it can do. Of course, now you gotta ask it the right questions. You gotta be the right person driving it. but I would say I would say seriously, coding, hand coding is over. It it really is. I hate it pains me to say that because I love coding. But I I've s I've seen what it can do. I'm a coder. I look at the output and I know there are a lot of people on YouTube that wanna grab on to the idea that it's a, it's a poor coder. I see videos of humans have won over because they're hiring us back. And I I think that's because of inefficient use of tokens and and how they're building software. But we build software and we keep the tokens at a reasonable cost because of the way we approach it. And I think that they're fooling themselves. I think the sooner that people grasp the fact that hand coding days are over and start trying to figure out how to survive this wave, the better off they'll be. They'll be ahead of the curve of everyone else when this thing steamrolls over everybody. I,I and again, it really pains me to say that because I love the code still to this day.

Botond Seres 07:42
Well, it it really depends on your definition of hand coding at, at this point. I I would argue that hand coding has been that for decades ever since we had IDEs with IntelliSense and some rudimentary cogeneration capabilities. Now obviously we have vastly advanced possibilities with ChatGPT and Gemini and Claude code and even Copilot. So, so James, where, where would you specifically draw the line between someone who hand codes anything, a couple hundred lines, or someone who does not?

James Edgel 08:25
I would say that I, I really do believe anyone who's hand coding is slowing down the organization as, as a leader now is the is because I look at it from both sides of the coin, right? When I first coded it was all about the coolest code, the coolest technology. When I got into leadership, I started to really understand a business perspective and they don't care about how, how clean the code is. They needed to do something, they need it to do reliable or needed to perform reliably and they don't care if it's spaghetti code as long as it works consistently, right? I don't think that's actually even the case. The code I'm looking at that is being generated from our request is very clean. But I think the understanding of coding is extremely important for asking. I think the new coding is prompting and it's gotta be the proper prompt. And if you don't know what you're doing as a programmer, you're not gonna be able to create the right prompts. So You do have to have knowledge. And it is like the catch twenty two situation, especially in ten years from now when you're not gonna h they're not gonna hire entry level programmers and then at some point these senior programmers are gonna move on into leadership or whatever and they're not gonna be doing the prompting. And they need to get replacements, right? But that's ten years from now. And I know businesses they can't even think past the quarter. And I have a huge criticism of that 'cause they have no long term game.

Because they're only beholden to the shareholders per quarter, but that's a whole different topic. But they don't care about ten years because they figure we're gonna be, you know, we'll have our exit, we'll be cashed in. I don't care about ten years. So that's a problem that's gonna happen. But right now, what the problem that's facing all of us is in our industry, they're learning really fast that coding is in you know, hand coding is in the past. And I think
I don't think this is gonna happen overnight because I don't think a lot of people are going to realize what I've come to realize from seeing it and working hands on because I am so close to the code and I know how to ask it to build things that are secure, that are clean and and all that. And I don't think there's a lot of people out there can even do that. Right. There's a lot of there's a lot of bad coders out there. We all know this, right? And and they're hired because it's it's a capacity thing. Anyway, I'll stop there 'cause I could go on, but I don't want to take over the show. But I you know, I don't I don't know how to explain it clearly that I truly believe hand coding is over and I know that's a terrible message to hear from a guy who from all of us programmers.

Botond Seres 11:07
I, I agree with you, James. I've been making a living coding by hand for quite a while, but even just five years ago I was generating most of my code. Even, even when these AI tools were just nowhere to be found. And with these AI tools, it went from generating it block by block, like a couple of words at a time, to tens or a hundred lines at a time. And if you, you are good at what you do, if you're good at coding by hand, you are gonna be amazing at using generative AI to code as well. I feel that the misunderstanding sometimes can lie on, on the other side of the spectrum where someone goes, Hey cool, build me a million dollar revenue SaaS application, make no mistakes. So

James Edgel 11:56
Yeah, you know, I feel your pain because being a CTO, you know, there's the stories you hear about, the CEO comes in and says, I just heard this guy on LinkedIn who said he wrote this application in like two days and why aren't we doing it? And I and I'm like, you know, first it's not peer reviewed, okay? So it's some guy trying to make a claim on LinkedIn, he's not make there's nowhere to be found anything that defines security or performance or scalability or anything.

It's a bunch of pretty screens, right? It's it it doesn't mean it works. And you're you're taking that and you're you're trying to pivot the entire organization based upon some guy's post who is trying to get attention. Right? And I and then I explain if the guy could really do this, I wouldn't even post that. I'd go make the million dollars, right? It's like why bother posting and bragging online? Because that's where I know it's BS. because

Is okay, where is their product in the market if this is so true?

Dave Erickson 12:58
Yeah, I see that all the time on YouTube videos and and LinkedIn videos where the guy says, You know, I've made a million dollars and I did it in this time I'm like, Well, if you've made a million dollars by doing this in in a day, why are you making a YouTube video to talk about it and try to get people to buy something from you? Why don't you just keep making a million dollars every day?

James Edgel 13:20
Why, why, why are they asking you to subscribe? At that point, I wouldn't give a damn, right? They're gonna get ten cents maybe on the subscribe, but you know, they've made a million dollars doing this. Yeah, it's I don't believe it. It's, it's seems unrealistic.

Dave Erickson 13:38
Well, we, we all know that even though AI tools help and and yes, maybe eliminate a lot of hand coding, if you've really worked with them, you also know there's a lot of work involved, right? It does, it doesn't make the work disappear. It's not a magic wand. It really does take real work to make them work correctly, but if you do it correctly, there is a productivity gain. Maybe you can talk a little bit about your experience when with that kind of balance between how much work it really is to set up these tools and work with these tools versus the impression that, you know, like you said, CEOs think it's a magic wand that you just wave and everything changes.

James Edgel 14:22
Yeah, that's a really great point, Dave. I think and then going back to what Botond was talking about earlier is, you don't ask it to build your Salesforce, right? It just can't do it. We've been working on our product and it's been a year and a half even with AI coding, right? So AI replaces the coding portion. There's a lot of portions it doesn't replace that people just don't have sight of. It's, first of all, you gotta host it, right. You still have to worry about security. You still have to deploy it. You still have to version control it. You still have to market it. You know it's, everybody seems to think AI is here and the money is going to come rolling in because we can shoot applications out. No, you've only solved like a quarter of the equation with the coding portion and you also got to be good at it. If you've got some bad developers who are asking the wrong prompts, you're gonna create a larger application faster that's total garbage. And it's that part of your equation isn't gonna change. It really comes down to the, the skill sets that you have for people. But yeah, and to put it in place, I think to put it in place is relatively simple and straightforward. Everyone has their own approach. I do more manually and and what I wanted to touch on with what Botan was talking about earlier is, it's better to build in chunks and control those chunks and make sure that they're secure. So you still have areas of focus. And I think if I would set up an AI developer team, I would have like an area for security, guys who just do security and guys who just do the the database stuff, right? And and these are these are people that are versed in those areas already with those skills. So I wouldn't turn my DBA into a UX designer, right? It's just it's not a good match. So you still need to match those skill sets. And this is where we were talking about earlier about where do we land if there's no coding, right? And and this is some of the things that people need to look at of their skill sets that they bring to the table. One thing I want to say is I know that when people hear coding is dead, it's, it's a natural feeling to think that,my God, my college education is, is worthless, and you know, my skills I built over the years are worthless. It's not true at all. So, first of all, you do have a college education that sets you apart from 80% of the United States at least, because it's roughly 20% of people have a college education. Second of all, you have technical skills that can be applied elsewhere. And I think those skills will be quite valuable, especially five years from now when they start to realize, my God, we didn't hire any entry level people. We're gonna have to compete with the, you know, the remaining skilled workers out there. And so I think it's gonna be a valley. We're gonna go through a very bad period in the beginning, and then later it's gonna be a heyday again, right? you only need to go far back as 2000. Remember COBOL was dead and then the Y2K bug came and COBOL programmers were making bank because they were the only guys who could fix that. I think that's gonna happen here.

Dave Erickson 17:49
Yeah, I, I remember learning COBOL, but at the time I was more interested in marketing than COBOL. so I didn't make bank at that time, but that's okay. I I I I'm happy with my life the way it is, so it it's all good.

James Edgel 18:02
It was a limited opportunity. Yes, yes.

Dave Erickson 18:05
You know, humans are really good at, at saying the wrong words to describe something. And I think that the, the sentence coding is dead is a very good opportunity to, to explore that because in my mind, coding has never been about solely sitting at a computer typing text lines in, coding lines, trying to make that code work. That's a small percentage of coding, right? For me, coding and when I went to school and got my computer science degree, a lot of the coding was really about systems architecture. How do the computers work? How does processing work? Yes, you're writing code, but you also need to learn about all these other things that fit into the code, including what would be DevOps and and other things. I, I think explore this or if somebody were to say to you, coding is dead. And how do I survive? I'm a developer. How am I gonna survive the death of coding? How would you kind of respond with that bigger view to them?

James Edgel 19:15
So it's kind of a paradox. I somewhat believe that statement. but you're right. It's not, it doesn't mean the individual skill sets are worthless overnight. I mean, they're still very skilled in the technical space. I think people should take an inventory of their skill sets. Now, people like us, now, I don't know your complete career background, but I suspect you're really technical. You always dive deep, right? And you always been on smaller teams that were more, you know, technically advanced and could move things faster and didn't want to work in a big corporation, right, with a thousand programmers. And I think that's where the thousand programmers should hear that coding is dead because they're not like us. We delve into technology because we loved it. We learned everything about it. We, you know, we would, I don't know about you, but I, I know I would have done the job even if I didn't get paid and then was surprised to find out I made an amazing amount of money, but I loved it as a hobby. That, that's how I got into this and started it. But that's not the majority of programmers out there. The majority of programmers out there work for a Google or a Meta or an IBM or whatever and they are cogs in the system. And they're the ones who don't know technology at our level, who don't eat it up and, and live it, right? They're programming for a job because it pays good. I think they're the ones that are in real trouble because they've only done the programming that they've been told to do. Here, build this widget, right? And then here's your inputs and outputs. Call me when you're done, and then we'll tell you what's wrong with it after you go through QA, and then you need to fix it. And then when you're done with that, here's another widget. And to me, I think

Dave Erickson 21:05
I don't know about you, but I would die in that position because I had no creative input. I would have you know, it it's like this is why I didn't work those kind of jobs and I got into the harder stuff that nobody wanted to do because it was fun, it was technically challenging, right? I wasn't it wasn't about the money. Although those positions oddly paid more than anything because they were so challenging. I think people like that and people like us will be able to navigate this quite well because, we'll be embracing AI and using it as a tool and using it to our will. But I think the majority of the programmers are out there need to recognize that if you've been in this cog programming position for years, you need to start thinking more like an architect and designing. You need to start thinking more like an entrepreneur and how to build something that doesn't exist or how to build something for less that's more efficient. You need to think completely different. Because you're not going to be writing the cogs anymore. Because they're coming for your job. And the reason why you're still programming now is because they haven't figured that out. Big companies move slow. I mean, look at the approach that they're taking now. Uber says everyone needs to, you know, max out on tokens. And then all of a sudden they spent their annual budget by April, which is hysterical. There's one guy who ran up what I think like twenty grand in tokens in a month.

I looked at my peers, I said they should fire that guy because that's not developing. That's somebody just gaming a system because they were dumb enough to make that statement of maxing out tokens. What they should have done is you know, and that's their li their leadership saying, Hey, we need to leverage AI with with no plan whatsoever. They just that I mean, that was it. They said we need to check the box because the CEO is yelling at me because he read something on LinkedIn that said

You know, this guy is doing this overnight and that's the domino effect. And I think because we're in this this early stage, this infancy stage of this, but I think companies eventually figure this out and they learn it, or they hire the right CTO who comes in and says, Guys, that's not the way it's done here's how we did it at this other organization. And once they figure that out, then all of these cog wheel jobs are gonna start drying up. The good thing is it won't happen overnight, so everybody still has time. Don't panic, right? Yeah, Douglas Adams, Hitchhiker's Guide to the Galaxy, don't panic.

Dave Erickson 23:39
Yeah. The pendulum's swinging, right? It's moving a lot right now because now you got all those CEOs who said max out tokens realizing how much it really actually costs, saying, well, now hire some developers because tokens are too expensive. You know, it, it's finding the balance. You've obviously had to deal with that kind of equation. What's your thoughts on that equation of developer time cost versus tokentime cost, how are companies going to figure out what that balance is?

James Edgel 24:14
Well, I think one, stop making the goal tokens. That's not the measurable output. The measurable output is the output, right? It's I you know, they should be saying to them, try to figure out how to reduce our, our time and effort in coding without maxing out tokens. That should be the goal is, you know, sure, I can build a redundant system with a billion servers and spend a lot of money on AWS, but do I really need that? And I think there's two questions I've always asked in my career that has made me very successful. I don't like to listen to people's suggestions on technology. I'll take it in because it helps me understand. Because usually business people they now have a little knowledge and ask for something very specific.
and they say, hey, I need this to be a SharePoint site and I need to do this and this, right? Because they start throwing out technology. And people who are not as mature in the process will follow that directive because it's the path of least resistance. When the first question they should ask is, what problem are we trying to solve? And then let them talk. They're allowed to introduce technologies and I tell my staff this, let them tell you what technologies they think should be apply because that's their way of communicating. They don't know our space. We're experts in our space and we need to bring to the table what solution will solve their problem. But in that equation you must ask what's the problem we're trying to solve. Because they may tell you something and you say, hey, you know what? We, we might have actually already solved that problem. You just did, you're not aware that it exists in our system and here it is. So there's one way because then you don't have to build it at all. And another way is truly understanding what they need because they may not need something as large as what they're asking for. That's one way to keep your tokens down by reducing the scope of the solution and matching it to the problem. So I, I think that's really important is, is to understand that. And two is to be economical. Always try to think how, how could I build this and not build, you know, and not boil the ocean. And I, I think that's a really important part as well. I've had development teams who would go off and create this massive solution and I I would come back and say, Why would you do that? You, you only need to change this one parameter and it will work. It's you gotta think you, we're experts of our own tool. We need to act like that. We we need to be able to bring the knowledge of our system to the table and not just blindly take orders like a clerk. Consulting firms do this. They love to do it because it's buildable hours, right? They'll build you anything. They if they won't resist, right? They'll say, You want that? We'll build it. You want a hundred story building? We'll build it. And they don't ask why you need it. But if you're in a corporation, you should be asking those questions. What is, what problem are we trying to solve? Why do you need this? Why do you need it this way? I think those questions will solve a lot of this token maxing right there. Yeah and, and that's not AI, that's just you're good thinking, right?

Botond Seres 27:35
Yeah, I couldn't agree with you more on this, James, to be honest. personally, I I I didn't get into this career to to have fun. I got into it. I looked up which is the best paying position, and it was software engineer at the time, so I learned the trade, the craft. And then, as it happens, there were lots and lots of tasks at every company that everybody refused to do. And and those are the most interesting, the most fun ones, to me at least. People just tended to hate them because they weren't easy to solve, they couldn't be solved in five minutes with a bit of boilerplate. But in my experience, the the job of of programmers, 90 to 95% is figuring out what we actually want to code. It it's not to code the thing. That's the, the easy part. That's what AI is taking over right now. So I don't understand why everybody is so worried about this part of the job going away.

James Edgel 28:38
Well I, Yeah, I mean, it depends on the shop you work with, right? there's a there's one sector of the business that I refuse to go into and that's trading because it's insane. you're you're expected to write code within minutes, right? Within hours of something, and it's you don't have that luxury of sitting back and saying, Hey, you know, let me think this through and come up with the right way to do this. That you'll you'll be pushed out the door. They'll, they're like if you're not coding within ten minutes of me telling you the problem, you're out of a job, right? I mean, if you're working in a a larger corporation, you might have that luxury too. mid sized corporations, you might have some of that luxury, but not every position does. And I think I you know, I think that's that's kind of the challenge, right? The they need to start coding right away. A and and the AI is going to be rammed down their throats. So I think I think every position is a little different and like trading, trading stocks, trading floor, any of that software, that is in it it it is an insane environment. It's like the worst programming job you could ever have because it's everything's gotta be done like right now. And if the market shifts and it brings out a flaw in your algorithm, they're expecting you to have something in production within fifteen, twenty minutes. It's it's insane, I mean, I'd be like, the risk I'm like

Dave Erickson 30:11
Definitely yeah. An area of coding that is only for people who ,who love in adrenaline.

James Edgel 30:21
Or chaos.

Dave Erickson 30:22
Or chaos, yes.

James Edgel 30:23
I I mean I'm really good at something if if I'm giving enough time to to solve the problem. Right. So I’d that's why I don't do well on time tests and all that. And you know, some people consider me intelligent and if I do a time test, they're like, Why are you failing all this? I go 'cause you

Botond Seres 30:44
I’d say, Maybe I'm not as intelligent as you thought.

James Edgel 30:47
I'm not intelligent at that speed. It's like I'm not intelligent when I'm drunk either. So it's like it's kind of like that. Okay. You know, you gotta put me in the right environment and then I shine, right? so some environments I don't work well in. But you know, Botond, I wanna talk about something you talked about earlier. You know, you talk about easy to work with. I think that's another important differentiator of what some of these coders need to do. Speaking from a management standpoint, I've worked with a lot of personalities and there are, there are some cubby holes that I could put a lot of people in, right? You start to see a pattern. You've got your your programmer who thinks he's the best and isn't. You got your programmer is the best who thinks he isn't, right? He's always worried that he's not doing well enough and he's always trying to improve. You've got your legacy guys that are so knowledgeable and so arrogant, they're so difficult to work with, right? You've got those. I think it's really important that people change their attitude if they're difficult to work with. That they're gonna be able to have less developers. So they're gonna be able to pick and choose who stays once they start figuring this out. If if you're a resource that's difficult to work with, guess who's going?

Botond Seres 32:13
That's, that's something I heard a lot, is that programming is turning into a social club.

James Edgel 32:18
Well, I mean you gotta be I don't know

Dave Erickson 32:23
I I don't know if it's a social club, but definitely soft skills and, and social skills that developers didn't really care too much about are now becoming more important, right?

James Edgel 32:33
Yeah, I mean speaking from a leadership standpoint, so well, I mean I I should go back and say when I went from coding to management, I thought I was gonna be this great manager and 'cause I was technical, right? The first year was a disaster. It really was. I didn't know how to manage. I didn't know how to deal with the leadership. I didn't, you know i it just wasn't good at it, right? And luckily, I changed companies 'cause I was so frustrated with the situation that I worked for Motorola and this guy took me under his wing and said, you know, you're great technically. You're a good leader because he's a terrible manager and he taught me how to manage. And I think that programs don't realize what leadership faces in some of these rooms, especially like in a boardroom when, you know, you're trying to explain to them a situation and they say to you, Hey, you know what? I I'm I'm not big on excuses. And I'm like an an explanation is not an excuse, right? And they don't care. They just steamroll you in the yeah, in the room. They just like

Botond Seres 33:37
If I had a penny every time I heard that it would be rich. Rich

James Edgel 33:37
So it's like you learn you can't even teach them, you can't explain to them 'cause they don't want to hear it. 'Cause they've got, they've got their leaders that are jumping on them. So now this is where it comes into play of how easy it is to work with developers. So you go to the developers and who is easiest to work with? Because if you get a guy who's dug in and says, Well, I don't like that technology and I'm not going to use it and it's like that's beneath me.

Botond Seres 33:37
I’ll just tell you, I hate it, but I'll do it. It's, it's fine.

James Edgel 33:37
Well, I mean, you're different. I've experienced guys that dig in and just absolutely refuse, right? And these are the guys that are gonna be the first to go, because I need people I can work with because AI is gonna make things faster. I don't have time to negotiate with individuals over their, you know, their technical religious beliefs is what I would call it, right?

Dave Erickson 34:33
As someone who has dealt with thousands of developers, I, I completely understand I have my own balance. I don't want a doormat who basically doesn't say anything and doesn't put up any resistance. But at the same time, I've met the, the quote rock star developers who think that they are the world's greatest everything. And I usually, I if I work with them once, I usually don't work with them a second time, right? And it is my choice who I put on a project or who which developers I like. Cause, you know, especially if you're working with, okay, how are they going to treat the client? How are they going to treat other developers? It is a a critical thing that developers are going to have to start worrying about. And and I think that comes to a maybe a different kind of point in a sense. These quote cog developers or developers who are just, they love coding and that's what they want to do. Now they're gonna have to learn new skills. They're gonna have to become different developers in a sense because the manual coding part's gonna go away. So they're the ones who are with the question of, well, what do I do now? Or what, how do I become a developer who will be relevant and have value in this new AI world? What would you start recommending to them? If you, you know, what would you say, hey, here's the things you need to do in this order. To start making yourself relevant.

James Edgel 36:01
So yeah, one, yeah, as we were just talking about, rebranding yourself, right? Making yourself easier to work with helps. but if you still wanna code and you want that balance, I mean one choice is, you know, become an entrepreneur, start your own company and pick what you wanna hand code and and then pick what you wanna do with AI. I mean I suppose you could do that in a corporation too, if you wanna hand code you know, something that

calls different modules that you built with AI, I there's nothing there's nobody's gonna stop you. No one's gonna say, hey you can't do that. I I I I've seen you do this and you shouldn't be hand coding. I mean if they're doing that, I would recommend get the hell out of there. Because now you're being micromanaged to the level that's ridiculous. But I mean I I think but I really think they should start taking and embracing the fact that they're not going to be able to write code all day. That they definitely are going to I you know what and they should still maybe hand code a little bit. Keep their it but at the very least review the code. it's very easy to fall in the trap with AI is it starts to produce stuff well and you start to trust it and you stop looking at the code and I don't recommend that. Cause I've seen scenarios where well I'll take a perfect example. If if I'm building a dashboard and I got multiple components and the components are getting the business name and address from the database. And then the next component builds another table and stores the name and address, right? And then you change the name and address for one component and it doesn't reflect in the other because AI has taken it upon himself to to build another table and store data in two places and you don't have a single source of truth. Now if you're not paying attention to the code, that's gonna happen. And you're gonna have to go figure that out. And AI may not be looking at that. it'll be looking at your code and saying, well, this module's completely correct. It's pulling the data from the table the way it was designed to, right? But it doesn't have the bigger picture of, hey, there's another component that's pulling data from a different location. But it's the same attribute, it's the same address of a company. It's just being stored in two different places. And unless you're smart enough to pull the two components in for AI to look at it, it might not catch it. So

Botond Seres 38:30
The data structures are a whole different level to just

James Edgel 38:36
Yeah, that's the one place where I tell people do not take your eyes off the target on this one. There's two things that are really important. UI, which is obvious, it's in your face, but data structures is not. And data structures dictate the entire architecture. And I am real particular about how data is stored and where it's stored, and I'm really big on not building all these tables. I, I take great pride in the fact that I built a social network that runs on less than 15 tables and it's completely dynamic for groups and users and all this stuff. And you know, and I use that as a showcase when I go to developers who want to build a table for everything. So I'll give you a perfect example. worked on a project where developers had a drop-down list box and they wanted to sort in a table. Well they had a table for let's say ice cream flavors. Right. Then they created another table for t shirt sizes. And I'm like, they're both drop down list boxes. They both have a key. They both have a value. Why are you creating two tables? They're like, Well, they're storing two different values. I said, But they're storing the same concept. So why don't you have another identifier that's the ID of the table list, right? One's for the ice cream flavors, one's for t shirts, and then you only have one table. Does that make sense to you guys what, what I'm describing? Yeah.

Botond Seres 40:04
Why don't you just create an innumerable if it's like t-shirt size and flavors? Like you don't even need a table at that point.

James Edgel 40:13
Well, for something like that, yeah, I would agree. But I'm simplifying it, right? There are other things that would require someone to manually update those, like locations. Okay, maybe that would be a better example. So I'm just pr you know, we could get into really technical stuff and and and go down to keep it simple. But I think, I think developers are not paying attention to the data structures and this can cost you in design. So now you have dozens and dozens of tables. I've walked into organizations where you'd look at their data structure and they take great pride that they printed and taped together like fifty pages that were printed out that shows their database structure. And I go, God, a frickin' nightmare. You know, I to change this field and I've got to go on a on an expedition to find out what table that you put this value in. So I think starting with you know. Keeping in control of your data is a massive help to developing with AI because you can always fix crap code, but you cannot fix a bad database design. You have to live with it.
Fixed state.

Dave Erickson 41:25
Yeah. Well that’s …

Botond Seres 41:26
Not even the design, the, the data is also something you have a very hard time fixing.

Dave Erickson 41:33
This is where kind of the developer who wants to be more relevant in the future kind of comes in and, and we haven't really talked about it, but the, the concept that you're talking about kind of brings it up. And that is it's not just writing an AI prompt, there's also such a thing as AI governance and giving the AI kind of the guide rails and you know, basically if you're running a team of developers and you've been in the business a long time, you know. You would say to the developers, this is how I want you to write the code. This is how I want the quality level. This is how I want you to design a database. And people aren't realizing that you have to kind of do that with AI also. They just think, I'm just going to type it, please design me a database. And and they don't realize that, yeah, it's going to design you a database, but it's just randomly picking thing concepts of what it think it should be. You're allowed to say to it, hey, I want my databases to be like this, and I want to have this kind of connection. I want it to be based on this structure and this is what my definition of quality is. And, you know, that's a lot of the work that has to be done before you even write a prompt, right? Is letting the AI know what the governance is for what they do. How do you feel? I mean, i it for that category, how are companies responding or recognizing that's something that needs to be done and who are they finding to do it?

James Edgel 43:00
So that's a real interesting point, Dave, and it's, it's one of the reasons why I've been extremely successful in my career is I'm still technical and hands on. I sit in the daily meetings with developers and I give them those exact instructions. And that's why I haven't had a failed project in my career. I, I always read the statistics. Fifty percent of all technical projects fail. I'm like, how do these people still have jobs? I would have been fired. Right? So I think, I think that directive is missing in companies to this day. And I, I don't think AI is going to fix that. In fact it's gonna make it worse. Because they're gonna be able to get to the bad software faster than ever before nd and build

Botond Seres 41:25
We will see single digits of success today.

James Edgel 43:50
And you know, and then they're gonna turn around, and it's the funny thing is those are the groups that will turn around and blame AI and then they embrace that. but there is the competing factor. There other companies that will figure it out will bury those companies and, and they'll be in deep trouble. So the importance of having those technical skills is extremely important. So what can a developer do now? They need to start thinking more like a technical manager and kind of make the shift that I did. when I, when I started using AI, I realized that I was very comfortable in the role because it's my my techniques haven't changed. I've always told developers here's the structure I want. I want minimum tables. I want you to use these technologies. I don't want if you're gonna introduce another technology, you're gonna have to prove to me why we embrace it, because I'm not big on embracing the latest technology just because it looks like fun and it, it's fun in a resume. It's gonna make economical sense. That's the balance I've had to make, make from going from technical to leadership. You gotta, you gotta play that balance. You can't just embrace every single tech that comes along. you also will be creating further problems down the line when let's say you embrace every single new tech that comes on. Now these people leave. Now you gotta go hire all those people. It's not and, and, and the developers there are very siloed. So you may get a guy who knows three technologies and then he leaves, then you end up having to hire three people because no one learned all three of those technologies as an individual. So that's why I like to resist incorporating as much new technology as possible. It's gotta make sense. So I I think that's what people need to start really looking at is they need to understand the tech. They need to keep it simple. That'll make their lives easier. And they need to start thinking more like a manager. And, and that's that's the approach, right is you're, you're gonna have to advise people. Like senior developers will do well. They really will do well because they're already coaching. They're already explaining to developers that, hey, you know, you don't do polling, you do interrupt, right? It's, you don't poll a database every ten minutes to see if a record's changed. You create a trigger. And if you're not careful, AI's gonna do the same thing. It's gonna pull instead of doing a trigger.

Botond Seres 46:22
Oh, that, that’s going to be guaranteed. Afterall AI is…

James Edgel 46:26
I had AI do it to me last week. I was working with Zapier, right? And I was looking closer at it and it was AI built a thing that would detect a record change. And then I dove in and looked and I noticed that it was it, it reads ten records in a bunch around where the change might occur. And I'm like, This is ridiculous. So I asked AI I go, Isn't there a better way to do this? And it says, Yeah, you can use triggers. And I'm like, Why didn't you tell me that in the first place?

Dave Erickson 46:57
Well because it it's randomly picking from an assortment of solutions, it's just randomly picking it. It has people keep thinking AI has intelligence. It doesn't. It's just randomly picking things for whatever fits its context structure that it's you know, based on the query and the guidelines and everything. It's not thinking about it. I've had that many times where it's like it's, it's goes in this direction. Like, well, why are you going in that direction after it built everything? Why did you do that? Well, y can't you just do this? that's much easier. Yes, let's do that. You know you gotta you gotta do it in the beginning. There there's a bunch of hard work on the beginning of how do you want it to approach problems? How do you want it to choose solutions? I think that people are not focused on that 'cause they don't understand it yet. And definitely in the big corporations they don't understand it. maybe the smaller ones, they since they're having to do it faster and move faster, they, they do.

James Edgel 48:03
Yeah, you know what I recommend and this is how I started Softex at Studio. I originally wanted to create a dynamic database and and dynamic databases are hard to create. And I mean you can create them easily, right? But you making one efficient and fast is really difficult. And then with queries is difficult. So what I recommend is people do what I did when I first started this is I had design sessions with AI that no code. Let's just talk about a design. And then I'll challenge its design. I said, Well, what if I use this approach with a a hundred million records? Then it says, Well, you'd have a problem. I'll go, Well, let that let's not go that way. Let's let's figure this out. Right? You you need to ask it challenging questions because you're right, Dave. It's gonna give you the first answer that solves the problem. It's like it's like dealing with kids, you know, and it's they'll they'll give the the first answer to get out of doing a chore or something. They're just, you know, they're they're not looking at a better way.

Dave Erickson 48:49
Yeah, sometimes AI and my fourteen year old daughter are very same the same. My fourteen year old daughter can give the best excuses in the world. She can come up with a logical reason why she did something so quick. And sometimes AI is exactly like it. You just gotta have your parent detector on to detect, okay, that that that's that's some BS that's being run my way to to justify something. But yes, the there AI and children have more similarities than, than people may realize. Since you mentioned Soft Exit, just maybe you can talk a little bit about Soft Exit, what it is and and why you went there.
James Edgel 49:41
Yeah, Soft Exit came out of two desires. one, I was working with Salesforce at the company I was at and astonished at their arrogance and was really angry. And as I ran up and down the halls, you know, expressing dissatisfaction with them, I used to always say, you know, one day I'm gonna build something and I'm gonna take away their business because they're so arrogant and they charge so much money and they're so fat and it's so one day I actually started to look into this as just a way to entertain myself on a train ride home that was an hour long every day. And that's where these AI design discussions came with the database. But you know, I, I don't know. I've lost my train of thought here, sorry, but what was the original question?

Dave Erickson 50:28
What is soft exit and you develop it or why did you develop it?

James Edgel 50:33
Yeah, you should key off of that, right? Because that is my my core product. So I'm not here just trying to pitch it. But so anyway, so yeah, I built that and the other aspect of it was I saw a lot of businesses doing the same thing, right? They they all want to store internal data. They all want to have dashboards, they all want it to be easy. they can't wait for anything 'cause developers take too long. 'Cause it's you know, it takes time to build that stuff. So I thought, hey, you know, I I could take all the knowledge I've acquired over the years and apply it to something. The only thing that's stopping me is a dynamic database, is you know, adding fields, adding tables. I could code all this other stuff. I've been doing it for decades. But the limitation was the dynamic database. And then I dug in and started to work with AI and found a way to build something that's pretty efficient and fast. And it took it took like a couple of months of just design work just to kind of get that down. So this is why I think that people who ask for, hey, can you build me Salesforce? And they think it's going to be handed to them in ten minutes, it's not going to happen because designing stuff takes a lot of time. And that goes to what you guys are talking about. That coding is a smaller portion of it. I, I agree with you a hundred percent. That thought into something worthwhile, right? That's going to be useful and used throughout time requires better design. So then I built SoftExit, which is a no code platform that has dashboards that you can put components in, charts, you can put in grids with editing capability, it does the validation, it does all of the stuff. It's all in one place. And then to kind of give my nose up to Salesforce, I'm offering at a significantly less price.'Cause we're a smaller organization and we don't have a thousand you know, salespeople to feed. And I think that's why Soft Exit I mean I I think that's why Salesforce is so expensive. They have an army of people trying to sell you. Just talk to You talk to them once, they're gonna call you every week, they're gonna try and take you out to lunch, they're gonna you know, I'm like, Maybe you could do less of that and charge me less, you know, that would be nice. So this thing's really easy to use and then to embrace AI, we put a, basically an application generator inside of it. So you can describe that, hey, I want an inventory system that tracks POs and suppliers and all that. And then it churns and about ten minutes later, it literally builds the entire application for you on our platform. And we think the advantage here is is on the surface it looks like it generated an application like everyone else and everyone else says, Yeah, so what? I can do that in cloud code, right? But I'm like, well, on our system, you don't have to debug it because it's already on a platform that works. You're just configuring our platform. And you don't have to worry about deployment because it's already deployed. You don't have to worry about security because the security's already baked into it. We're not creating new code with new bugs and new vulnerabilities. We're configuring a system that's already been stabilized. And is is trustworthy and reliable. And I think that's the advantage we have. I think we design this specifically for AI to do that. And this is why you don't see this kind of a generator in Salesforce because Salesforce is so enormously complex. I don't even know how they would go about starting to do that. It's it's just requires. You'd have to rebuild it as a new product, basically. Exactly. And they won't. So this is my edge on sports, right? That, you know, hey, you can build anything. You you start with a CRM, it's free. We've got an app store that you can download like risk management or inventory or CRM. We have another application called D V I R, which is for the trucking industry where they have to do these mandatory checklists on trucks. That's already included in the app store, it's free. And they just start installing these. We're going to keep releasing these in the App Store. Everything in our App Store will be free. And then you just start installing what you need into your solution. And then you can it's not code that you can't touch. It's literally configuration. So if you say, Hey, I like this, but you know, it'd really be better with three more fields and and get rid of these four fields and you just go in and do that within five minutes and you're up and running in production within twenty minutes. 'Cause you know it's stable. You don't have to really test this out. It's it's just, you know, maybe your workflows you have to do a quick test, throw a record in, see if when you change this value that the next value changes, right? That kind of stuff. But you could be up and running the same day and then move on to something else. And for internal business stuff, this is not the exciting code that you want to work on. So you can, you know, have the company build their internal systems. This can receive input from landing pages or or data intake forms and then you get configure it up really quick and then you get on to the cool stuff that you want to work on instead of these everyday ask, you know, business tracking systems. They they track everything. You know, they love to track stuff and they love dashboards and this is a perfect system for them and it's inexpensive. So that's why we build it.

Dave Erickson 56:08
You also got a lot of experience using AI and applying AI for coding and, and that's what gave you a lot of experience. I do have a question since you are a coder and understand it. When you're looking at the various AI coding tools around, are there ones that you prefer? Are there ones that you would recommend that developers start looking at first to to to work with?

James Edgel 56:32
I don't know if I'm the right person to ask because I'm not embracing a lot of those tools. I'm doing a lot of this manually because I find that I'm closer to the code. I can see what it's developing. I don't give it access to my code. I do a lot of cut and pasting. Cause when I paste it in, I'll spot something. I go, that's crap. That needs to be fixed, right? the thought of somebody turning their entire GitHub over to a AI generator just scares the hell out of me. I mean, you know, I, I remember it's, it's gotten a lot better. but at the beginning of the year, one of the things I had the biggest complaint about was feature drift and it would drove me crazy. I'd spend weeks building a feature out, and then three months later I'd go to test it and it's gone. The AI decided, well, I haven't asked for it lately, so it must not be important, and yanked it out of the code. So
You know, so I'm like, well, that's not gonna happen anymore. I'm not gonna let it have access to my source code. I'm gonna make sure I know what it's changing. So I've taken a completely different approach. And now a lot of people will laugh at me and say, Wow, that's very manual of you, and that takes more time. I said, Well, yeah, but my application's very stable and things don't disappear from feature drift, and so I so maybe maybe I'm not the right person to ask her about the latest tools. And I also think that you know, cloud code, I mean Anthropic charges a lot of money for that. We've kept our our tokens down because we don't embrace a lot of that. And, you know, I see these things called skills, and skills basically look like the prompts that I paste in myself. And I'm thinking, they're paying an exorbitant amount of money 'cause they refuse to cut and paste. It's so

Dave Erickson 58:21
Yeah, I, I've actually, I use Codecs more than Claude Code because Codex is much more efficient on tokens and it it it it reads the governance stuff in really cheap. And if you write the governance stuff well, Codex will actually produce stuff that is competent and and clean and and reasonable.

James Edgel 58:41
That's good to know. I th I'll maybe I'll take a look at that.

Botond Seres 58:44
So James, in your opinion, what is the future of coding?

James Edgel 58:48
So the future of coding I think is more and and this is my opinion, right? It doesn't mean I'm right. It's it's just I'm my opinion's right for me. It's not right for everybody. I think coding is gonna become more design. You gotta become more of a consultant and and less of a guy hiding away in a dark room just pounding out code, right? You're 'cause there's gonna be, there's gonna be more light shed on the individual programmers. You're gonna need to get closer to the business and understand the needs. I think there's gonna be way less hand coding. And I think hand coding is just gonna be a a self indulgence, a luxury that people can pick and choose what they want to work on. and if they're not micromanaged, right? Some companies may say, you know, at some point in time that hand coding is inefficient. You shouldn't be doing it. I can actually envision that, that at some point, maybe five, maybe eight years from now, they literally will say, No no hand coding anymore.

Botond Seres 59:55
James, thank you so much for joining us and helping us figure out how developers can stay relevant in the age of AI development. Well, we are at the end of today's episode, and I'm sure you have plenty to think about after this conversation. But before you go, we want you to think about one important question.

Dave Erickson 01:00:15
What are you doing to stay relevant in the AI age?

For our listeners, please subscribe and click the notifications to join us for our next Screenbox Technology and Business Rundown Podcast. And until then, learn some new skills to stay current with fast paced technology. Thank you very much for taking this journey with us. Join us for our next exciting exploration of technology and business in the first week of every month. Please help us by subscribing, liking and following us on whichever platform you're listening to or watching us on. We hope you enjoyed this podcast and please let us know any subjects or topics you would like us to discuss in our next podcast by leaving a message for us in the comment sections or sending us a Twitter DM. Till next month, please stay happy and healthy.