Technical debt

Well, Dana, can we go back to the lab? Many mission systems supporting national laboratory research have been in place for decades, and they remain tightly integrated with ongoing programs.

What practical strategies are working for extracting technical debt from that scenario while maintaining continuity for active research and operational missions?

Yeah, I think for us, you know, tight integration with mission and ongoing programs is important.

just hands down part of our core values.

And in my mind, we cannot and really probably should not take on extracting technical debt from our operational missions as our top priority.

The focus for us has to be on building the future capacity and all of that mentioned value as part of that capacity.

We'll leverage the core capabilities that currently make our institution so unique.

So this is the reason for like a national investment.

that humans have to understand to make sure it's safe and secure, test, maintain, and remediate.

All we've done is we're using AI to create tomorrow's technical debt faster.

I advocate there shouldn't be any technical debt in the future if you use models on what you build software.

That's

my view.

I had a phone call just earlier today with a, a group here in the Western US that Is it about 30-ish sites and they're in process to get to between s- you know, 70 or 80, and they're actually requiring standardization for, uh, as, as a condition of the, the acquisition.

And the properties they're talking about acquiring are a little bit distressed and, and interestingly in the past, as you know, you know, changing PMS and changing is, you know, that part of your platform to fix that tech debt you're talking about is change management and it's, there's a lot of pushback and it's tough and people don't like it.

In this case, the indu- the customer I was speaking to basically said they realize this has to happen.

They, they were saying, "Sign us up, we're ready to do it." Like they- they're seeing the exact same problem that you just described.

They have their data in order.

And almost all of them have consolidated on a tech stack.

They don't have that tech debt that you just, you just talked about.

And so there's, there's several areas where we could talk about that.

Enjoy.

we were carrying tech debt, a crazy amount of tech that we couldn't get out of it.

Everyone- I love that you

call it that as well.

Tech debt is such a good way of describing it, yeah.

Everyone for the last like 10 years up to that point had heard, that's a great idea, but we can't do it.

That's a great idea, we can't do it.

Das sagt Thomas Elsner, Partner bei McKinsey und Experte für große IT-Transformation.

Thomas beschäftigt sich seit Jahren mit, mit einem Thema, das viele CEOs seit nachts wach hält: IT-Kosten oder neudeutsch Tech Debt.

Und ich bin Philipp Hügle, Host dieses Podcasts.

20 % der Unternehmen haben in den letzten sechs Monaten ihr KI-Budget überschritten.

Das heißt, du musst bezahlen, je nachdem, wie viel du nutzt.

Ähm, und das verschiebt dann natürlich die, die, den Kostenpunkt auch noch mal, auch noch mal dramatisch.

Kann man denn sagen, dass, also ich mein, dein Thema der letzten Jahre oder eines deiner großen Themen der letzten Jahre, ähm, du hast viele Unternehmen dabei begleitet, Tech Debt, sagen wir mal, aufzulösen, zumindest zu verkleinern.

Also sprich, IT-Kosten, die irgendwie aus Legacy IT, ähm, herrühren.

And then, you know, Justin, as you know, there's startups and others that are really bringing hope to that space as well.

You know, it's interesting to bring up the, let's call it the legacy or the tech debt problem, for lack of a better word.

How do we, you know, fix the mainframe, the deck of axes of the world? Those are the safest.

I mean, they are.

Yeah, to get one under for research is quite expensive.

You know, but you had your whole, you know, identity authentication authorization layer in one environment, making it massively easy to secure or easier.

But, you know, I'm sure there's a chock full of issues, you know, but this tech debt problem.

And Ian, I mean, Cisco's been around a bit longer than Salesforce, but I assume that tech debt challenges still alive and well in your environment.

Yeah.

Well, it scratches a different question, which is somewhere back in, I don't know, whenever TaskRabbit started 15, 18 years ago, someone created some tech debt.

They said, you know what, let's just put this in a YAML file.

And here we are 15 years later paying back that loan.

Okay.

I think it's super funny.

I see this with my AI agents all the time, where they will trade off tech debt intentionally.

They'll say in their discussion, I got these agents and they're doing code reviews of each other's code.

So before we go to the break, my little take, my little two cents.

The real problem that we are facing is the fact that technical debt that every institution wants to resolve is not about a certain infrastructure or a certain way of coding, which of course can be modernized.

But it's about the fact that the monolithic code contains a lot of business logic that lost causality because people went from 10 lines, 400 lines, 1,000, 10,000, 10 billion lines, rest of code.

So the people that did the coding are leaving or they're not even capable of understanding it anymore.

And so now with agentic AI and with AI, you can extract that business logic, abstract it right in a way that it can own it.

And then we can reposition it somewhere else where more automation agentic AI can power it up.

Of course, as I think Paul mentioned before, making sure that you don't create a new technical debt because you need to have the competencies to manage those environments.

But the point is that why did they lose causality? Because the business, they don't have any causality any longer.