Deep Dive: The GenAI Build-it Trap

Olivia Carter

Alright, let’s start with something I think anyone in tech can relate to—the thrill of writing code. Whether it’s crafting it by hand or using GenAI, there’s just this... buzz, you know? You solve a problem, you see the fix, and you’re like, "Wow, I’m making progress."

Mark Putnam

Oh, absolutely. It’s creative, it’s satisfying. And let’s be honest, it feels good to see something tangible come together. You have this immediate sense of accomplishment, like a builder stepping back to admire a finished wall. But—and here’s the "but"—that feeling can be deceptive.

Olivia Carter

Deceptive? Like, in what way?

Mark Putnam

It’s the speed, Olivia. Quick solutions feel like momentum, but really, they can be laying down traps. Custom-built code, no matter how exciting it feels in the moment, has a tendency to boomerang back. It’s like, fast progress now, long-term headaches later.

Olivia Carter

This kinda reminds me of those "quick fixes" you regret because they make life worse after—like using duct tape on a leaky pipe. Can you give an example of how this plays out in real tech projects?

Mark Putnam

Sure. Look at locally maintained platforms. They’re great in the short term—tailored, flexible, specific to what you need right now. But over time, they become fragile. They rely on one or two people, and if those people leave, suddenly, nobody knows how it works. You’re stuck with messy, brittle systems that nobody wants to touch.

Olivia Carter

And that’s just the human side, right? What about the tech side?

Mark Putnam

Exactly. Platforms evolve, processes update, and suddenly what was cutting-edge becomes a roadblock. I’ve seen it time and time again—custom solutions that felt perfect at the time, but later, they clash with everything else. Maintaining compatibility becomes a full-time job.

Olivia Carter

Wait, so are we just... coding ourselves into a corner?

Mark Putnam

In a way, yes. The more custom layers you pile on, the more tangled and fragile things get. We actually have a term for it—Andy Kyte’s Gordian Knot. It’s this metaphorical snarl of technical debt, created by years of ad-hoc coding decisions.

Olivia Carter

That’s wild. And what’s worse is that every "quick fix" feels like you’re cutting through the knot, but really, you’re just tying another strand, right?

Mark Putnam

Exactly. That’s the illusion at play. Coding feels like control, like progress, but in the long run, these unwieldy solutions become the problem. GenAI, for instance, supercharges this dynamic—solutions appear in minutes, and it’s so tempting to lean in. But each of those quick wins adds to the tangle, making every future change harder to execute.

Olivia Carter

Honestly, I gotta say, it’s kind of depressing to hear how efficiency now equals headaches later. Doesn’t the industry ever... you know, learn from its mistakes?

Mark Putnam

Oh, we try, but the cycle is hard to break. It always comes down to balancing the thrill of speed with the discipline of foresight. We know where it leads, but, well, restraint is easier said than done.

Olivia Carter

You’re right, balancing that thrill of quick wins with the discipline of foresight feels like a constant struggle. So, when it comes to this Gordian Knot you mentioned, where do the biggest hidden costs start to pile up? What’s the fallout we don’t see right away?

Mark Putnam

Great question. Let’s start with maintainability. Custom systems often rely on a select few individuals—maybe even just one. If they leave, so does the knowledge to keep the system running. And over time, the technical debt snowballs, making it hard for anyone to step in and sort things out.

Olivia Carter

So, basically, you’re at the mercy of one person’s brain.

Mark Putnam

Exactly. And that’s only the start. Then there’s compatibility. Technology evolves—platforms update, APIs change—and those custom solutions? They don’t keep up. Suddenly, you’re stuck trying to retrofit an outdated system into something modern. Think of it like trying to connect a rotary phone to a 5G network.

Olivia Carter

Yikes. That’s painful. What else?

Mark Putnam

Efficiency, or rather, the lack of it. Custom code often duplicates what off-the-shelf solutions already provide. You’re pouring time and resources into reinventing the wheel, when vendors have already spent years perfecting it. It’s like building a car from scratch when you could just, well, buy one.

Olivia Carter

And I’m guessing these "DIY cars" don’t exactly win reliability awards either?

Mark Putnam

Exactly. Vendor platforms are tested at scale—they’re designed to handle pressure, to be reliable. Custom solutions? Not so much. They tend to break in the worst ways at the worst times. Like right when you’re relying on them most.

Olivia Carter

Okay, but here’s what I don’t get—if we know custom code creates all these problems, why do we keep falling into the same trap?

Mark Putnam

It’s the allure of control. Custom code gives you this false sense of ownership—you think, "We built this, so we know how to fix it." But over time, that control turns into constraint. Every new line adds complexity. And GenAI adds to this dynamic by making it so easy—solutions pop up so fast, it blinds us to the long-term implications.

Olivia Carter

It’s kinda wild to think about, right? Like, the very thing that makes GenAI powerful—speed and precision—is what makes it dangerous too. It’s like giving a kid the keys to a Ferrari without a single driving lesson.

Mark Putnam

That’s a great analogy. GenAI removes the friction of building, but that friction exists for a reason—it forces you to pause, to plan, to think through the consequences. Without that, it’s easy to create solutions that unravel the second they face real-world conditions.

Olivia Carter

So, we’re trading caution for convenience... and paying for it later.

Mark Putnam

Exactly. And what’s worse is that very few of these systems are designed with usability in mind. Developers tend to build for internal logic, not user needs. The result is clunky, unintuitive tools that no one wants to use—or worse, that no one can use without constant support.

Olivia Carter

And yet, people still think, "It’s faster to build our own."

Mark Putnam

They do. But speed now doesn’t erase the cost later. And that’s where strategy matters—knowing when to build and when to adopt. The challenge is convincing people to think long-term when the short-term fix is just... so easy.

Olivia Carter

That really hits home—strategy over shortcuts. So, when we take this thinking to something bigger, like an "AI-infused university," what does that even mean in practice? What does thoughtful AI integration really look like?

Mark Putnam

It’s about weaving AI into the very fabric of the institution, Olivia. From academic research to administrative processes, AI needs to be both ubiquitous and responsible. And that’s the key—responsible implementation. It’s not just about throwing AI at every problem; it’s about determining where it actually adds value.

Olivia Carter

Right. So, basically, being smart about when to use it—and when not to.

Mark Putnam

Exactly. Responsible AI comes down to this framework—two categories: game-changing AI and everyday AI. And then within each category, you either focus internally on institutional capabilities or externally on user-facing improvements, like student services or faculty tools.

Olivia Carter

So, four buckets, basically?

Mark Putnam

Pretty much. And each of those buckets calls for a different approach. Take game-changing AI focused inward—this is your high-stakes research, scientific breakthroughs, that sort of thing. Here, custom coding is essential, because these are experimental, cutting-edge use cases. It’s all about pushing boundaries.

Olivia Carter

And what about game-changing AI facing outward? Like, for customers?

Mark Putnam

That’s where process automation comes in. Universities can automate workflows, simplify processes, and improve student services. For that, the best tools are often already baked into platforms like Oracle or Workday—or even third-party solutions that integrate across systems. Custom code might still have a role here, but it should be rare and strategic.

Olivia Carter

Rare, as in, only for really specific needs?

Mark Putnam

Exactly. Then there’s everyday AI—this is more routine stuff, like chatbots or faculty productivity tools. Customization here is totally possible, but again, it’s often smarter to leverage the low-code features built into systems like Microsoft Copilot or Google Gemini. The focus should be on adoption, not invention.

Olivia Carter

Okay, but how do institutions know when to build and when to adopt? Like, is there a rulebook for this?

Mark Putnam

That’s where maturity comes into play. A mature institution knows when to use custom solutions and when to trust vendor-supported platforms. It’s a balance, Olivia. Innovation doesn’t mean building everything yourself—it’s about knowing what’s already possible and applying sound judgment about where to invest your energy.

Olivia Carter

So, it’s about being picky, in a good way?

Mark Putnam

Exactly. A university that’s truly AI-infused isn’t the one coding like crazy. It’s the one that exercises restraint, strategically choosing when to build. Restraint, though, does take discipline—and sometimes, that’s harder than just going all in.

Olivia Carter

So, Mark, we’ve explored when restraint with custom code is actually a strength in leveraging AI. But what does this kind of disciplined decision-making mean for leadership? How do leaders really master this balancing act?

Mark Putnam

For leaders, Olivia, it’s really about shifting the mindset. In the AI era, success isn’t about how fast you build or how much you can innovate with new tools. It’s about clarity—knowing what’s actually necessary and resisting the urge to build just because you can. In many ways, it’s a reframing of what leadership looks like in tech.

Olivia Carter

So it’s like—less about speed, more about strategy?

Mark Putnam

Exactly. Leaders need to understand that rushing toward the latest shiny solution often creates more problems than it solves. The institutions that thrive in the AI era will be the ones that exercise discipline, prioritizing sustainability over short-term excitement. It’s about making intentional choices—choosing not to build can actually be the more transformative decision.

Olivia Carter

That’s interesting. Saying "no" isn’t usually how people think about innovation, right? We’re so used to equating progress with motion.

Mark Putnam

True. But restraint, especially in tech, is strategic. Take custom code, for instance—there’s a time and place for it. If it supports research breakthroughs or strategic integrations, it can genuinely move the mission forward. But when it’s slapped onto ERP systems or duplicated across departments, it traps institutions in that Gordian Knot—layers of complexity and fragility that hold everyone back.

Olivia Carter

And that knot never really gets untangled, does it?

Mark Putnam

Rarely. It just tightens over time. That’s why the ability to lead with restraint is so important. Leaders need to step back and ask, "Does this build really serve our mission? Is it sustainable? Or are we just solving the symptoms of a problem rather than the core issue?"

Olivia Carter

It’s almost like the real power in leadership isn’t in doing more—it’s in doing less, intentionally.

Mark Putnam

Exactly. The AI-infused university of the future will succeed not by eliminating custom code entirely, but by sharply defining its role. The key is using it selectively, in ways that maximize efficiency, security, and adaptability. Leaders must think long-term, which requires discipline and sometimes stepping back from the instant gratification these tools offer.

Olivia Carter

It’s kind of ironic, isn’t it? The tech that promises to make everything simpler ends up making leadership more complicated.

Mark Putnam

It does seem that way. But, honestly, leadership has always been about navigating complexity. The challenge now is to thrive in this moment of possibility without being consumed by it. Leaders need to embrace restraint as a strategy—not as hesitation, but as thoughtful, deliberate action.

Olivia Carter

That’s such a powerful point. You know, Mark, this whole conversation has really been about reframing how we think about tech—not as a race to build, but as a tool to strategically advance what matters most. And that requires seeing the bigger picture.

Mark Putnam

Absolutely. The institutions and leaders that thrive in this era will be those that understand where their real value lies—not in the lines of code they write, but in the clarity they bring to deciding where code belongs. At the end of the day, knowing what not to do is just as important as knowing what to build.

Olivia Carter

And on that note, I think that’s all for today. Thanks for such a thoughtful conversation, Mark—it’s been a deep dive for sure.

Mark Putnam

Always a pleasure, Olivia. Let’s keep asking the right questions—it’s the best way to navigate the future.