Latest / Unstuck Pod / AI Creates New Technical Debt (Dennis Bruce and Jamie Schneiderman)

Transcript

AI Creates New Technical Debt (Dennis Bruce and Jamie Schneiderman) · Unstuck Pod

• Jamie: Hi everyone. Welcome to the unstuck pod. really excited today to have a Dennis Bruce here and, than me, you know, butchering, his background and his journey, I'd love, Dennis, if you could introduce yourself and share with everyone more about, know, your journey, what you're doing and let's go from there. Dennis Bruce: Yeah, sure. Thanks, Jamie. Thanks for having me. Appreciate it. I'm Dennis Bruce. I'm founder, CEO of Tanganyet Solutions. We're based in Atlanta and in Argentina where we have our technology teams. And Jamie mentioned something about my journey. My journey started in Boston, Massachusetts, ⁓ where I grew up, ⁓ lived until my mid-20s, worked initial technology. professional career was mostly, well, it started in Wang Labs back in the glory days of the mini computer. talking about early to mid 80s and worked as a QA analyst there integrating Wang and IBM systems. And then I got one of those phone calls from out of the blue. And at one point around 1986, I found myself working at the UN in Geneva, Switzerland, because I had a somewhat unique skill set that I knew how to integrate Wang systems with IBM systems. You know, this was in the day. ⁓ Pre-apis and all that sort of thing. So you had to cobble together a lot of a lot of different and platforms in order to get systems to talk to each other. So that's where I come from At the UN I worked there for several years integrating doing that integrating Different systems ⁓ mainly and a IBM for the purposes of field communication to sort of Let's say emulate a telix. Let's say by using an email front end. So that was an interesting one. Jamie: For those who don't know, sorry Dennis, I don't mean to interrupt you. For those who don't know what a Telex is, could you maybe share that part too? Dennis Bruce: Yeah, so don't even know if I know what a Telex is, to be honest with you. I saw that machine that they use in the telecommunications unit and they were like, well, what they were really doing was they would be retyping in these word processing documents that would be sent over for. Jamie: You Dennis Bruce: communication with field offices. This was a refugee organization. So there was some very highly sensitive, highly critical information that was going back and forth from UN headquarters to the field offices. And it just took an extraordinary amount of time because of the repetitive work that had to be done. The word processing document would be created, it'd be printed, it'd be sent into office envelope to the telecommunications section. And the telex machine is just this big thing that you type in, like it's I think the precursor to a fax you could say, know, it's for sort of digital communications, but very asynchronous, know, very, let's say, I remember they were using X25 networks back in the day, but basically it was a way of, many years ago of communicating effectively via digital means you could say, okay. So we streamlined that process, got it up and then, In the early 90s, found myself again through another one of those phone calls. I found myself working in Argentina in different development efforts. Firstly, privatizations, ⁓ which is basically a state-run company, in this telecoms, telecommunications. converting it to a private sector company, ⁓ a more, let's say, commercially based and viable company. And my role there was ⁓ ⁓ scale IT investments ⁓ as part of that transformation. So was that. I ⁓ later again in Argentina, or still in Argentina, in different turnarounds with a large retail group, retail and property management group. It was more of a technology turnaround than a business turnaround, just replacing all of the old and putting in the new. ⁓ Jamie: Mm-hmm. Dennis Bruce: And privatizations again with the airports, the airport privatization, they privatized the whole network. And again, was a lot of this recurring theme is just reinventing, helping companies reinvent themselves through the implementation of technology. Those that are undergoing transformations, is a transformation. In the case of the retail property management, was a transformation of sorts, technology and culture as well. the UN was a transformation also in that sense, you know, to get out of the old mindset of, you know, we have to do this via telex and we have to do a very cumbersome process where there are refugees sitting in a field, you know, somewhere that need very extensive and critical in many cases, ⁓ healthcare interventions that have to be airlifted out of ⁓ war zones and places where they could get that care. And a lot of times that these communications depended, it was a life and death situation. So minutes, seconds count, you know, and when you're talking about communication, it sometimes takes days from source to destination. we were able to fortunately make an impact there. So it was quite gratifying to see some of the improvements there. then beyond that, when I founded Tengrenet, was a lot of the same thing. In early 2000s or mid 2000s, was finding companies that were through transformation or startups or something that was mergers, something that was transformative effort that was undergoing ⁓ in the company. result of some transformational change in the company putting the technology pieces in place to accompany that transformation. And accompanying means to be ⁓ the for the transformation or sometimes the ⁓ leading of that transformation, but there's always some component of cultural change. of adaptation and transformation that goes just beyond the technology itself. The technology is only one piece. And many times it's the easiest piece. Jamie: Yeah. So, so actually I don't want to interrupt you, but I, I feel like it's a great segue into what is transformation that is occurring on a massive scale, like way, way beyond just the technology any one company and kind of tech transformation that is Pacting every company or every sorry, every, so I think every company, but certainly I think about SaaS companies and that's AI and where AI fits into this. And what is that doing for you? Cause you've been on the leading edge of being part of this in a variety of ways. What's happening in your business or what are you seeing related to that now? Dennis Bruce: We're seeing a lot of things, Jamie. It's a loaded question, you know? you know, from our business, I'll start with our business perspective, but, you what we're seeing is, we're seeing major disruption in terms of the services that we provide. And Part of our business, about 40 % of our business is software development, custom software development for ⁓ our clients. And that's undergone some major shifts in terms of, first of all, lot of clients can do this stuff themselves because the productivity enhancements with tools such as Claude Code and Opus. There's such that, you know, the traditional bottlenecks and competing priorities and limited skills or limited say, internal limitations of their teams, because there are just so many priorities, so many projects going on, has now been let's say, reduced to a extent, thanks to these tools, you know, now that, and we have clients that tell us that, right? So ⁓ we know because they tell us that. And we know that that you know, from our own of these tools that you know, we can do a lot better, faster, if it's done right and there are ways to do it right and there are ways to do it wrong. ⁓ So that's one thing, the business aspect of it is that we are seeing disruption, are trying to, where the new value prop is evolving, right? So we have to evolve with that. We have to see, like where do we provide value in this new age of and AI assisted software development? So that's like we're right in the middle of that storm right now, repositioning And then I think from what we've witnessed, from what we've seen, we've seen a little bit of everything. Our teams, for example, we've seen code that we have to integrate as part of a project, a client project that comes from somebody else, it comes to us, we have to do the integration piece, but we see that what we're integrating into has not been, let's say, it's been, I would call it, allowing the inmates to run the asylum in the sense that if we rely 100 % on cloud code and you don't do the proper code reviews, if you don't have the proper engineering governance around the software development, then it can go off the rails. And it Doesn't matter that it compiles correctly. Is it fit for purpose, functional? And we've seen that, yeah, it all looks right. It's structured right. It compiles correctly, but functionally speaking, it's not fit for purpose. You might as well just throw it away. You've just created a load of technical debt, which is another aspect of this. You can create what we've discovered, what we've seen and witnessed and read about with other case studies from other, looking at some of the industry topics on this issue is that you can create massive amounts of technology debt. if you don't govern the AI development properly. And we position our services to help companies overcome technical debt, we're also seeing that technical debt is being recreated and on a massive scale even. it's a problem. And one of the things that we're looking at, believe it or not, is we're looking at positioning some of our services to say, well, we can help you overcome the massive technical debt you've created. based on Cloud Code to run, let's say, to run the shop, ⁓ run the show, you know? We're not going to put it in those terms, but I mean, there is, this problem is starting to appear. So, are several, you know, those are some ⁓ aspects it, you know, that we're seeing, creation of technical debt, the lack of guardrails around... the development process, the lack of engineering, let's say the human in the loop type of thing, which I think is important, especially for the more complex initiatives. But there's also been some really cool stuff we've helped develop, know, some of the smaller, like smaller language models that are more like, you know, think of a chat GPT or a cloud, but more concentrated within a company, you know, building that, that, that that model that is, let's say, the internal intelligence of a company. Very interesting project. in very sort of, let's say, in of productivity, it's just amazing ⁓ what be done. But it needs rules, needs governance, it needs a culture of understanding what the capabilities are on the one hand, but being able to ⁓ do properly with the right guardrails on the other. Jamie: Yeah. So when we start to talk about, there's technical debt, like ⁓ oftentimes when I, when I work with companies, you know, I get into conversations with them about, you ⁓ you've, you have an idea, you've got some stuff, you, things are going and you often run. Like you're often running. so from a business point of view, they haven't thought it through or things have changed so much over time that they haven't thought it through. And so they, they kind of jump ahead and there's a lot of inefficiency and friction. Okay. So that happens. what I hear from you is there's, hear something when you're describing this situation. Okay. So, so I hear this from a technical point of view, which could be part of it anyway, but even from a building product and what product means. I see a lot of companies doing that as well. I'm trying to understand as we talk about this is, you know, ⁓ know that all that friction and all those challenges from a business perspective are fixable. It's all fixable. It can all be corrected and it can be corrected relatively quickly. You're jumping in when someone has done some of this work ⁓ you can correct it to a point, right? So all of these things, what I want to know is, How far down the path have we gone or are we going toward transformation or adjustment or change or disruption, as you called it, that can't be fixed? You go too far, can't be fixed. It's going to be a real problem. So there's easy to fix. And then there's the bigger stuff. People are talking about things like SASpocalypse. It's like, is that the case or is it just about finding a happy place between the two is where I'm trying to go. Dennis Bruce: It's like. I think that it's a combination of things. You mentioned SaaSpocalypse, Jamie. We're seeing now that companies are doing it. We were recently at AI Week here in Atlanta a couple of weeks ago, and company had this use case of... a company, one of their clients had a very well-known CRM in place, they found that that well-known CRM, as functional as it was, didn't cover all the functionality that they need. And this company went in and they said, well, okay, do you know what you need? Do you have some user stories built or whatever? And they went in, they ⁓ reproduced what already existed in the current CRM ⁓ and they bolted on these new use cases from the company and essentially just rebuilt the system with these new use cases and built a new CRM as a result. In the matter of, don't know, I don't know if it was days or a couple of weeks or something like that, but you know, an incredibly short period of time. And think, you know, if you look at that, I think that what could happen out of that, it, if that's any indication of a trend, right, is that the SaaSpocalypse thing goes from one to many to maybe one to one or one to a few where you start building these very, very highly customized solutions for companies. that is what breaks the traditional SaaS model, let's call it, right? You know, you do that one to many, and you're know, everybody has their own version of a SaaS application. You know, we're going back to this, maybe you could even look at it as going back to, you know, the pre-SaaS days, you know, where we were building, you know, very specific products for maybe a specific market, but, you know, there was ⁓ maybe more customer or more customized solutions. know, now we have, know, before, you know, 30 years ago when we talked about ERPs, can count them on one hand. Now are multiple ERPs every vertical. So ⁓ it's that of thing, how that's proliferated over time. So are we back to a more ⁓ specialized way ⁓ of developing, thanks AI, that is very, very highly customized to a specific client and specific use cases that is, let's say, Jamie: Right. Dennis Bruce: much more than something of the box that comes from a SaaS provider. And I think maybe that's part of SaaS Poco. I think that's part of it. I think that there are certain areas where I think SaaS is going to live and thrive and, you Jamie: Thank Dennis Bruce: some very highly specialized cases. But I think there are other cases that there will be this sort of ⁓ in the SAS model, maybe not a SASpocalypse, something that's going to, let's say, alter that model in a way that's going to be very noteworthy. Yeah, you know, if that is an example, like I said, if that maybe is like a trend that's happening, I can really see that. I can really see people, you know, this buy versus build discussion that we've had for years, decades, know, do buy, do you build? You know, now it takes on a different meaning, you know, ⁓ takes on a sort of like... Jamie: Yeah. So I actually No, go on, go on, go no, I don't mean to interrupt you, Dennis. Dennis Bruce: No, go ahead, go ahead, I was finishing the talking. Jamie: I was going to say, but I was going to actually ask you about that is, there's, what's the, so, like to know you deal with CEOs of companies of these, of SaaS companies all the time. And so the question to me becomes, You know, what's their role in this? they, are they helping? they like, how, how did they view it? Because it's their baby. They've been doing things a certain way. They have a product. They have to evolve and go to this, these other places. also been given technology that is incredibly powerful, could open up more doors. So. What are you seeing from those folks in terms of the decision making and their approach are they helping or hurting or what do you see? ⁓ your observation? Dennis Bruce: Yeah, so a couple of things, you so for our existing clients we're seeing, you know, a very definitive move into saying, okay, using AI tools is a must. not impacting, let's say from, you our business. It's just, you know, we're doing that naturally. I to a CEO this morning ⁓ and told me, she said, Yeah, our team is using her team was using chat GPT internally. And I said, well, ours is using clog code. You know, we're doing the front end work. Our teams are in the backend work and then they come together and they integrate. I said, you know, maybe there's a, you know, something there that we should be, you know, let's say aligned, you know, and just sort of as maybe a best practice, let's use one tool, you know, for both, you know, and that of thing, you know, just as a means of just standardization, best practice, that type of thing. The other ones are, like mentioned earlier, there are some clients that they will say, look, we understand that you can do this in a three month project and 10 week sprints or whatever it is, we, think some of these green field projects, where somebody just something built from scratch and it lives outside of their core systems and that type of thing, that lends itself pretty well to, and that, requirements are very well defined, that lends itself very well to just an AI development effort that maybe before we would have done, but now they would do it internally, that type of thing. So have to pick our battles, we have to see where we add value. And then there's another case of ⁓ talking to a CEO last week, who we built a solution for, an AI specific solution for. he loves the solution. It's very innovative in terms of the small language model and having intelligence that's within ⁓ the borders a company. And he's like, you know, I love what we built, but I don't know what to do with it. It's a like sort of a product. said, do I take it to market? Do I just use it internally? It's given me enormous productivity gains in general. But is there a market for this? know, so there's a it's all over the map, Jamie, what we've seen, you know, from standardizing on tools to, you know, determining what to do from a market perspective to, like, let's say, do we use you ⁓ do we just do it ourselves? always been ⁓ question, but now ⁓ ⁓ the productivity enhancements, there is some things that are going in house these days. ⁓ yeah, we understand that and it so fast that I don't know, what's it gonna be? ⁓ Six ago, I would have given you different answers, I think. Six months from now, I may give you different answers again. I don't know. Jamie: Mm-hmm. No, and I think that's a totally fair way to look at it, talk about it, because it's right? It's changing so fast. So, okay, let's do this. So this is in the context ⁓ of where things headed and what you've seen. And you've seen stuff that has, you talk about... Dennis Bruce: Yep, I know. Yeah. Jamie: Systems we started off talking about telex like think about where we're where we're shifting where we've shifted to advice would you give to? CEOs of companies that are obviously where technologies core to what they're doing, but it's more about application not necessarily about building the technology terms of how they think about this, approach they should take and what is best in your mind for their business going forward. Dennis Bruce: Yeah, so insofar as AI, okay, here's what I will say. There is a huge demand, and I only say huge demand because just judging by the inquiries I get and the people I talk to, CEOs and others, that ask the question, know, do you, can you or do you know someone who can help us at our processes internally to see how and where we can apply AI effectively. That's a really good question. If CEOs are asking that question, that's the right question to ask because there are lots of examples. of AI solutions that just don't provide the results or based on the expectations that they had in their proof of concept, MVP or whatever. I think people are beginning to realize now that it's not just a question of throwing, know, cloud code at it. And, you know, people talk about, you know, creating 60,000 lines of code in 3.5 minutes. I say to that, who cares? It's irrelevant. I don't care how much code you can, you know, you may have just created those 60,000 Jamie: Yeah. Dennis Bruce: lines might be the equivalent of three years of technical debt. Who knows, right? I think we need to get away from the vanity talk about, you know, how much code you can create and all this sort of thing and start talking about, you know, where and how. I can look internally at my processes and see, today where can I get the most bang for the And start small, ⁓ start small but think big at the same time, right? ⁓ ⁓ important thing there is to understand the process. We talk about business analysts, these AI analysts, the ones that understand the capabilities of AI, but also have the understanding of how to map business processes, how to analyze that to determine, look, here is your opportunity and here is how we can define the value proposition and here is how we can define the return on investment. I think that's the way to look at it. Don't look at it, Don't get carried away by Claude Code or ChatGPT or Codex or Gemini or whatever. Don't get carried away by the vanity of how much code you can produce. my mind, that's nonsense. You know, Think about you know, where you know, internally you feel the most pain, you know, where, you know, where the processes are either broken or inefficient or, or whatever. And start thinking about you know, how can we best analyze it. Whether you do it yourself, because the people that are on the front lines of those processes know them the best, or whether you bring in somebody to help as, a, let's say a phase zero of the process, and then start thinking about the the technology solutions and so forth. ⁓ Jamie: All right, well, in the spirit of bringing in somebody who can help, ⁓ can people find you? Can you let them know where they can find you, reach you? Dennis Bruce: ⁓ yeah, can find me. Please connect with me on LinkedIn. think we'll sharing our contact information here. Jamie, if that's possible by email. ⁓ Jamie: Yeah, yeah, sure. Dennis Bruce: there's anything that we can help either through network, we're starting to build these relationships with consultants others who can help map processes because we see that as ⁓ an important zero of all this AI implementation. yeah, LinkedIn, email, phone, all of that contact information, Dennis Bruce, TangoNet Solutions. to talk and if we can help in any way, happy to do so. Jamie: Well, Dennis, thank you so much and to everybody who's watching, listening, appreciate it and until next time. Dennis Bruce: Thanks, Jamie.