EPISODE SUMMARY
AI adoption is the number one priority for Australian IT leaders, and that is exactly why the unglamorous work of replacing what no longer works keeps sliding down the list. Meanwhile the risk is growing, because AI is now finding and exploiting legacy weaknesses faster than any human team can. The Telstra outage on 8 July, traced to a defect the vendor had flagged since 2022, is what happens when technical debt is left at the bottom of the pile.
About the hosts
Jono Staff is CRO at ASE Group and Executive Director of WinDC, with more than 20 years in cloud infrastructure, managed services and critical infrastructure. Naran McClung leads Australia’s leading Azure managed services practice as Executive, Azure at Macquarie Cloud Services, and thinks in cloud economics and AI. In this rant cast they go without a guest and pull apart the question the industry keeps ducking: why are we still so bad at replacing the old things that no longer work?
What this episode covers
- The Telstra outage on 8 July, traced back to a timekeeping software defect the vendor had warned about since 2022, and why that is a technical debt story rather than a one-off
- Why 93% of organisations carry technical debt yet only 24% feel confident they can track it, and what that gap costs
- How AI is changing the maths on legacy kit: the largest Patch Tuesday on record at 570 vulnerabilities, and end-of-life systems accruing 218 new vulnerabilities every six months
- Whether AI is now the threat, the defence, or both, and why a runaway agent is a tooling failure rather than Skynet
- Sovereignty and token economics: what happens when the frontier model you built on gets pulled or triples in price, and why sovereign is being redefined
- Data centre scarcity in Australia, with 14 companies bidding on the last three megawatts of air-cooled capacity
- The closing argument: make technical debt a board-issued KPI so it can never be quietly deprioritised again
Why now
Adapt’s latest survey of more than 2,400 Australian CIOs and IT leaders puts AI adoption as the number one priority. The risk is that everything else slides down the list while the attack surface grows. AI is now finding and exploiting legacy weaknesses faster than any human team can, which turns years of deferred maintenance into a live liability rather than a someday problem.
Who this episode is for
- CIOs, IT directors and heads of infrastructure balancing an AI mandate against ageing estate
- CISOs and security leaders watching vulnerability counts climb on end-of-life systems
- Boards and executives deciding how to fund and measure risk they cannot see
- MSP and cloud leaders advising customers on remediation, sovereignty and cost
Show notes
This episode is supported by NinjaOne.
Follow Naran McClung on LinkedIn here: https://www.linkedin.com/in/Naranmcclung/
Follow Jono Staff on LinkedIn here: https://www.linkedin.com/in/jonathanstaff/
And don’t forget to subscribe and rate the show! We love hearing your thoughts and answering questions.
For all enquiries, contact pru@thingsreasons.show
EPISODE TRANSCRIPT
But it’s not the sexy thing to do.
Yeah.
You say to your mate, “What have you been working on lately?” “Oh, technical debt. Oh, I’m so sorry.” Yeah. I’m so sorry.
Right.
Ladies and gentlemen, this is Things Reasons. We are the independent show for IT leaders in Australia. Thank you very much for joining us. I’m, of course, Naran McClung. You are? Jono Staff. Excellent. Now, we have no guest today because this is a rant cast.
It means we’re gonna have a moan. You’re gonna sit there and enjoy it, and we’re gonna sit here and love moaning about certain things. There are, of course, topics we’re gonna discuss. There was a very famous outage of a household name today in Australia. May or may not have been related to technical debt.
Let’s talk about that, Jono.
Why are we still bad at replacing old things that don’t work anymore? I think we inherently like old things. That’s where we’re going with that. I feel like it’s a- We’re gonna talk about it … it’s a choice. We’re gonna talk about technical debt. We’re gonna talk about vulnerabilities as well, and there’s a link here to AI.
I think most of our IT leader audience will have seen the mythos versus fable versus many patches for the biggest Patch Tuesday, uh, recently as well in history. Surely AI is going to make everything better. Well, listen, I feel like there’s a balanced view on that. So here we are. We’re gonna get stuck into it once again. What on earth have you been up to, Jono? Mate, the madness never stops. Right. Madness never stops. We’re full swing into Q1. Mm-hmm. I know all of our audience is as well. Yes. And, uh, and things are getting crazy.
Obviously- Mm-hmm … uh, I am neck deep in the middle of doing some interesting stuff with WinDC. Yes, you are. We’re at Adapt this week. Mm-hmm. And shout out to Adapt, that was a fantastic- Love Adapt. Let’s put the logo- … conference … up there. Ding. Fantastic conference. Had a good time at that. Um, I think, uh, everybody showed up really well.
Some really interesting stats coming out of that, actually. Oh, yeah? Which, uh, I think we should get into today. But no surprise, the number one priority from over 2,400 CIOs and IT leaders that were surveyed by Adapt is… Artificial intelligence. You guessed it. Okay. Excellent. So very interesting. But it does make me wonder, what are the things that have moved down the list that probably should be closer to the top?
Hmm. So if we focus on the shiny good thing that we love, does that mean that we forget about other things? Look, I, I don’t want to say too much about it, but maybe we could talk about that today. Okay. Very good. Well, look, um, pretty busy Q1 for me as well. Um, for those of you on LinkedIn, I apologize. I know I’ve been plastered all over it with new ventures.
Um, our cloud business expanding into government’s very exciting, so, uh, very much looking forward to that. That took the best part of six months to get that over the line, for good reasons, deliberative reasons. Our business is quite deliberative. As you know, there’s some interesting characters there that make you jump through hoops, for all the right reasons.
I’m not complaining. Not complaining. But we got it done. And so, uh, yeah, that’s, that’s interesting. I think government’s got their own challenges, but they’ve also got their own needs as well. And suffice to say, AI’s on the agenda for government as well. Look, I think everybody needs some help, Naran, and you’re very helpful.
Look, I try to be. Uh, you know who you’re helping. You’re bringing IT back to the working man. I love that you’re rocking the T-shirt today, and it’s a Bonds T-shirt. I have many of those. It’s Friday, and, uh, I, I’m really, uh, I’m just, uh, picking up the vibe. Look, I want you- to be comfortable on the podcast, Jono.
It’s important to be comfortable. I think so. Okay. All right? All right, good. I’m, I’m glad we pointed that out.
Right. So let’s talk about the Telstra outage. Now, we’re not here to bash Telstra. No, we’re not. That’s not what we’re about, all right? Now Telstra, some people love them, some people have different opinions.
That’s all good.
They had a major outage. Uh, it was fairly widespread. Uh, it was like 8th of July, 4:30 AM, uh, and it was a full outage nationwide. 90% of calls were restored 9:00 AM, full recovery by 4:00 PM. Now, I’m a Telstra customer. Right. I did notice it when I woke up. There was some mild panic. Mm.
Although it may come as a surprise to you and our audience that I’m not completely dependent on my phone. Right. Uh, I just connected to Wi-Fi- Yes … where I was instead. Well, look, and like you, Jono, if I lost mobile connectivity for a little while, it means I take less calls, uh, it means I get to relax a little bit.
However, we do need to be sensitive, obviously, to the 600-plus triple zero calls that were affected- Did, and it did impact people Well, that’s what we should talk about. Um, obviously voice calls are still really important. Yep. I think there was a bit of a mix of some data outages as well. Mm. That probably had more to do with congestion- Mm-hmm
uh, than anything. But, um, but this was actually pretty serious, and, uh, I know everyone was trying to find their own workarounds and what have you, but when you think about how reliant everybody in the country is on that network, there is a reason- Yep … that, uh, the telecommunications industry is, uh, is subject to the Security of Critical Infrastructure Act and certain standards, and, um, uh, because there’s big implications.
Of course. A- and we talk about that a lot, uh, in the work that we do with critical infrastructure and being in, in tech. Certainly you and I- Mm-hmm … and our experience with critical infrastructure. Yep. I think we had Vic, uh, from Ninja on the pod- We did. Shout out to Vic … the other week, and, um, and we talked about having a bit of a laugh at it, but also at the same time talking about how serious what the tech industry, uh, i- is involved with.
Yeah, and look- Like, the implications of getting it wrong … indeed, and look, to be fair to Telstra, I mean, you can almost set your clock. You know, you blame Telstra one day, you know, it could be Optus the next, right? And for Telstra, with obviously the biggest mobile coverage nationally, uh, it’s tough at the top, right?
And so, you know, they do come under scrutiny when these things happen. I think- I’m particularly interested though in the root cause on this one- Mm … which is obviously a key aspect of what we wanna talk about today, and that’s technical debt. And so when we talk about the root cause, I’m led to believe that we’ve got a software defect in timekeeping service, something that had been kicking around, and the manufacturer had warned Telstra since 2022 that it needed to be either updated or replaced.
This is not uncommon. Um, how many of our listeners out there have done their best over the years and over their careers to warn all sorts of people- Yep … in their organization about things that need to be resolved? Yep. And, uh, for whatever reason, they’re not able to get that up. Yep. Whether it’s because it costs money, or there’s other implications, or- Mm-hmm
it’s inconvenient- Mm-hmm … or there’s other priorities. Yep. Everybody in our industry does struggle- Yep … with, uh, not only identifying where some of the, these issues and, and risks lie, but actually mitigating them and r- eventually rectifying them properly. Well, look, yes, right? And I think everybody listening, I mean, look, Gartner’s gone on record to say that 93% of organizations experience, are experiencing a degree of technical debt.
24% of organizations say that they’re confident and consistent in their abilities to track technical debt. I mean, that seems like a pretty big problem to me. Yeah. Well, I think it’s something also, uh, it’s a, it’s a result of the pace of change in the tech industry. Right. Even if you just look at the last 20-odd years that, that you and I have been doing it.
Yep. Things actually change so fast, it’s not hard to accumulate technical debt or- Yep … um, or have to manage, uh, obsolescence. This is a constant thing we’re trying to manage. Yep. Even if you, even with your iPhone, it, it’s built in. Yep. Like m- like it just eventually gets slower until you get so angry, you buy a new one.
There’s nothing wrong with it. No, that’s right. But you have to buy a new one. So, um, this is, this is a- actually a, a, quite a big issue. It is a big issue. Look, I mean, curious to get your take on this, but the, the organizations that I’ve worked with were acutely aware of their technical debt. Um, they had a broad plan of what they needed to do.
Um, but- It was a question of prioritization, it was a question of cost, against also remaining competitive in market as well, right? And you’re sort of balancing staying, you know, uh, staying liquid, right? Staying positive, um, and earning money and meeting your customers’ needs, versus a project that’s probably months, if not more, and expensive and time-consuming, and it takes prioritization and resource away from other things.
I mean, I get it, right? I, I get why re- uh, organizations juggle this. Um, but having said that, I feel like the risks are increasing more and more if we think about the adoption of AI, right, and use of tools like Mythos, for example, to accelerate perhaps the exposure of these vulnerabilities. Uh, I couldn’t agree more.
Right. I think it is getting more and more difficult for an organization to get away with sweating these assets. Yep, exactly. More and more difficult. Um, we talk about vulnerability management driving this. It, it didn’t used to as much as it does now. Mm-hmm. Right? So your ability to be able to patch something- Yep
that’s obsolete disappears. Yep. And if you can’t patch it, at least if it’s, if, if you think you can get away with it still working, I mean, a Cisco router’ll run forever. Yes. We all know that. That doesn’t mean it should. That’s right. Um, and where you could get away with things before, you can’t now- Hmm
because of vulnerabilities and cyber attacks. Yep. That is only now accelerating. I think, uh, what was the stat where, was it Mythos? Yep. How many i- how many vulnerabilities were able to be identified? Like, the AI now is able to identify vulnerabilities and exploits at a rate that is, you know, extreme, extreme multiples- of a human being trying to do it. Well, so, uh, recent data points, so 14th of July, biggest Patch Tuesday ever for Microsoft. Shout-out to Microsoft. Um, 570 vulnerabilities, three times the previous record. And that is largely due to- Mythos Front… Okay, Mythos specifically- Yes … but, but frontier models identifying- I believe so
out there identifying vulnerabilities. Allegedly. That’s right. Right. Okay, and so look, and then add to that, uh, end of life systems accumulate on average 218 new vulnerabilities for every six months once the support stops. So this is a problem that compounds, right? I think just sit with that for a minute.
I- Right … and now scale that out over, what did we say, 94% of organizations- Uh-huh … have identified technical debt- Right … in their business. Mm-hmm. All right, and so, I mean, if you’ve got infrastructure that it- is exposed in any way, that’s one problem. It may be you try and ring-fence it, maybe you wrap more firewalls around it, you do all sorts of things to try and mitigate risk.
But let’s face it, if you get breached and there’s a lateral movement, uh, and there’s a bit of legacy kit kicking around, I mean, what are they gonna go for first? Right. Yeah, absolutely. Um, you see, it’s quite a hard thing for IT leaders. You know, I, I do, um, I do feel for the teams that would’ve been responsible- 100%
for this, uh, at Telstra. But I do think at the same time, a- and maybe, maybe it’s just network engineer Jono looking at this one, it does seem like a, an own goal. Yeah. It does. But d- just, I wonder how many of those are sitting out there, right, just ready to go pop on any given day. Yep. Well, I’m sure they’ll, um, be paying, paying more attention to it now- Mm
obviously. Yep. And, uh, gi- given, given the ramifications. Yep. I think, uh, I think if, if… Open question. Mm. If AI is getting better at exploiting vulnerable legacy environments- Yep … or equipment, uh, just a- as a general statement Can organizations leverage AI to combat this i- in, in some way, shape, or form? Yeah.
You know, it’s like only 24% of organizations know what to do about their technical debt. Yeah. Maybe that, maybe, maybe we… Maybe this is not a human problem to solve at scale. Yeah. Yeah, I think that’s valid, certainly. I mean, it, it’s… I guess there’s, there’s sort of two columns, isn’t there? There’s the good and the bad, right?
So let’s see. The tech is helping us to find the vulnerabilities at a record pace. Now the attackers can use the tech for the same reason, right? Now we obviously saw Fable pulled from the global market, um, which is interesting in its own right, right? I mean, if you built capabilities specifically off the back of Fable, you know, maybe if you built agentic systems that needed that faster time to first token type ca- capability, and that was all of a sudden pulled out from underneath you, I’d have to think you’d be pretty annoyed by that, right?
That’s point number one. But point number two, if you can use the tech, uh, maybe to help combat your own risks, maybe that’s a positive as well. I mean, it could go either way Yeah, it, it, are we gonna get to a point where, um, uh, AI agents are just, uh, de- defending and attacking and the humans will have nothing left to do?
Right. Exactly. I mean, we, we saw the thing in the news recently about, um, OpenAI’s little project, right? The, the rogue agent that got out of the lab and- I, I, I’ve seen, yeah, I’ve seen mixed reports on that. Uh, was it real? Well, you, you read the article. So it basically… I don’t have it in front of me, unfortunately.
Um, but it was basically, it was an agent that was built to achieve a certain task, and it had to go off and, and essentially, um, review or, or work with various frontier models to solve a problem. And it was… The article made it seem like the thing had, uh, designs for itself, right? Almost like it was self-aware, and it got on with it and decided to go off and hack something.
In reality, it was achieving its task, right? If anything, it’s, it reads more like a runaway process, where the agent actually was highly effective at doing the thing that it was specifically written to do. Right. Right? Absent some controls and some tooling around it. Okay, so Skynet’s not here yet. No, no, no.
And, like, you’ve heard me talk at length, and I could spend the next hour talking about why A- AGI’s not gonna arrive. In fact, speaking at my son’s school about this next week. The reality is that there’s a little bit more to intelligence, right, than just the correlation of data points and weights in a model.
All right? And so the thing didn’t have designs for itself, but it was highly effective at doing the job it was tasked to do, and it did so without any guardrails whatsoever. Whose fault is that? I, I’d say maybe the person that scripted it.
This episode Is brought to you today by NinjaOne. Now, we said we were never gonna take anybody’s money, Naran. Here we are. And here we are. And look- … production value like this is not cheap. No. Money isn’t exactly growing on trees. No, it’s not. Right. But here’s the thing I like about Ninja, they were the only sponsor who was willing to front up and let us say whatever we want.
That says a lot about them. It says a lot about them. Mm. It, it’s about trust. It is about trust. It’s about trust. Mm-hmm. And look, for the next 30 seconds, I reckon you need to listen to this because you’re gonna love it. And you’re not doing anything anyway, all right? Yeah. What else have you got to do?
Nothing. Absolutely nothing. NinjaOne unifies IT to simplify work for nearly 40,000 customers in over 140 countries. The NinjaOne unified IT operations platform delivers endpoint management, autonomous patching, backup, and remote access in a single console to improve efficiency, increase resilience, and reduce spend.
And they’re also obsessed with customer success, and have held a 98% satisfaction score for more than five years. Try NinjaOne for free at ninjaone.com/freetrialform.
Now, speaking of Skynet- Yes … Terminator 2 is apparently playing at the cinemas in your Street.
Right. This is totally random. Okay. But, yeah. Why? Stay with us. Why is that interesting? I remember that was the first movie my dad ever took me to. Is that right? At the same cinema. Huh. Yeah. Right. I don’t know. There you go. Oh, geez, mate, that was… That wasn’t that long ago. What were you born before that?
Terminator 2. I think it’s… It was a long time ago. Not really. Terminator 2. Wasn’t that the ’90s? Yeah, that, that’s a long time ago now. Uh, first movie, right? the same, but you, you’re younger than me. First, not much by the way Who? The early ’90s. Well, the first movie I remember was, um, I saw NeverEnding Story in the cinema the day that my brother was born.
How about that? Wow. I wonder if everyone remembers the first movie they went to see. Yeah, I’m sure our- It’s weird that we do … IT lead audience are benefiting from this conversation. Okay? It’s Friday. That’s right. But look, no, look, Jono, I, look, I don’t think the Skynet scenario is likely anytime soon. I really, I used to really worry about this, as you know, right?
Because we’re actively involved in this stuff. I mean, there’s hyperscalers in data centres and whatnot. Uh, the last thing I want to do is feel somewhat responsible for 60 to 70 jobs being wiped off the market, hence I went deep on that. I don’t feel like that’s gonna happen. The tech is here to stay.
It’s augmenting human beings and, um, for the most part, helping us out, absent the odd rogue agent that goes off and does its job. Well, I think, yeah, everyone’s got, uh, mixed opinions on this. Mm. And, um, now the Australian government’s weighed in- Yes, they have … and our prime minister has made an announcement about a new special department- Mm
to deal with this. I see. Yeah. Talk me through that. Well, um, we’re talking about, this is the government’s AI policy, the Australia Digital Infrastructure Investment cycle. Is that what we’re talking about? Well, I think so, yeah. I think that’s the one. I think so. Right. And so we’re gonna have, we’re gonna have a dedicated government agency- Mm
to, uh, to bring together policy and industry and, and government around how should- Yes … Australia manage the implementation of- Mm … AI- Mm … into our economy and in- into our lives. Yep. Uh, quite interesting. It is interesting. I, I think he, he, sort of the analogy he used was the last time the government got involved in this was with aviation- Hmm
uh, at, at this kind of scale. Right. And, um, and I think there’s, there’s mixed opinions about what it’s gonna do. Is it gonna slow us down? Mm. Is it gonna keep us safer? Is it gonna make things better? Yes. For me, I think, uh, I think at the very least it’s important to be- Thinking about this from a government perspective when we talk about, uh, our sovereignty Yes
our sovereignty That’s a key word, isn’t it? The, the best… And, and what does that mean to… Means lots of things to different people, but- Mm … in this scenario, I look at practical examples. Mm. Uh, if you’ve, if you’ve built your business around a particular frontier model that is owned by, uh, another country- and, uh, and then it disappears- or they decide to triple the price. Yep. The, these things are important considerations. They are. Um, and what does it mean for an Australian organization to, to be able to leverage this tech and maintain some semblance of sovereignty- uh, how they use their data- Mm-hmm
how much they pay for these things- Yep … where the infrastructure is, the, uh, environmental impact of those things- Yep … if you’re interested in that- Yeah … uh, and I think you should be Um, it’s gonna be gonna be interesting to see how it plays out. Well, I think a few things are converging at the same time, right?
And so, um, we all probably saw, uh, Apple’s announcements and Apple’s movement in this space, right? I mean, these Mac Minis and the things that you can do. I mean, shout out to one of my PCs, right? He knows who he is. Um, he’s built some incredible capability with a Mac Mini and just privately hosting a model and doing awesome stuff, right?
And so Mac- Apple and Mac are gonna continue to progress that as are other big hardware providers. I feel like the cost of that technology’s gonna come down as well considerably. Um, you might be able to run, you know, like a trillion parameter model, you know, in your own house doing its own thing. Um, so that’s interesting too.
Um, I can imagine a time, and this is similar to sort of Microsoft’s approach, right? In that you may well be, um, leveraging harness technology, right? That maybe routes jobs from your phone, from your Mac Mini, and if you need a frontier model for something complicated, maybe it’s a choice, maybe it’s automated and it goes off and uses it, right?
And so this sort of mix of local processing, local hardware, uh, uh, internet facing frontier models. But back to the question on sovereignty. If it’s about privacy, less so than it is about token economics, well, I think the, the definition of the word sovereign is gonna change. And our friend Dan Wright from Armada, shout out to Armada, right?
He went big on that, you might remember. He did, yeah. Right? I mean, usually when people hear the word sovereign, they think national, right? They think country sovereign. His view of the world was sovereign can be the trusted perimeter of the thing that you want for yourself, your version of sovereign, right?
And so I think that’s gonna play out. I think the hardware’s gonna be better. I think we’re gonna see more and more use case, um, for privately hosted models, particularly driven by token economics, and it’s all happening at the same time. That’s right. All at the same time. And they’ve consolidated it into…
Well, they will be consolidating it- Mm … rather, into a new, a new agency just to deal with this. Yep. One of the other things that’s, um, of interest and running in parallel is the copyright situation. Yes. And protecting Australian IP and artists and- Yeah … um, and, and our creatives. I think, uh, I think there’s a way through that.
Mm. I think there’s a way through it. There’s other countries have models that, that seem to work and make sure that people are rewarded in the right way for their IP and their content. Yep. Um, but I do think it’s right to, to pause and consider that carefully. 100%. Well, I mean- ‘Cause this is moving really quickly.
I mean, look, every organization, right? Is facing that right now, right? And so when you, you look at the fine print of your, your contract with, with Anthropic and with Claude, um, you know, there’s definitely ways to, uh, protect yourself from data retention and also ensure that your IP isn’t being used to be, to train the models.
That’s critical. Microsoft have got similar policy as well. Um- It is the right conversation to have, certainly. And people need to be aware of, you know, that classic thing where, like, if you don’t… I mean, we are, we are paying for it, but maybe if you’re using the free versions, if you don’t pay for it, you are the product, right?
Oh, yeah. So- unwitting people out there using the tools, um, with their clever ideas, I mean, those are being harvested, let’s face it. Yeah, for sure. Mm. For sure. Um, I mean, you gotta train it on, on intelligence. Right. Exactly. So, um, I think, uh, I think, I think it’s, it… Look, broadly speaking, I think it’s a good thing.
I do hope, though, that it doesn’t slow us down. Yeah. I think, uh, I think it’s incredibly important for our, just our economic prosperity in general that- Mm … that we keep up here. Mm. Yeah. And, and that we endeavor to keep up. Um, and that is probably Uh, probably no surprises there why the, the research, uh, coming out of Adapt this week said that the number one priority- Yep
for Australian CIOs- Mm … is IT, uh, uh, AI adoption- Yes … into every facet of IT in their business.
Okay. And so le- let’s talk about, um, not wanting to be slowed down and demand. And so Jono, I mean, y- you know, we are in the data center game. You, you know, you’re in the data center game, uh, in a very unique way and novel way as well.
Um, a good mate of mine, hopefully he’s watching, he, he spoke to me about one of the, um, key data center providers, um, in Australia, um, uh, were down to their last three megawatts of air-cooled capacity recently. They’ll… And until such time as they refresh that in a couple of years’ time, I might add. The last three megawatts, there were 14 people bidding on that.
14 companies? 14 companies bidding on the last three megawatts, I won’t name the business, um, that they had available, and it’ll be two years until they have more capacity. Yeah. The, the supply and demand- Right … situ- the, the demand forecasts for just data center capacity, uh, to host AI- workloads is, is through the roof.
Absolutely outrageous. Um, and, and it is a global marketplace. Mm-hmm. I think the opportunity for Australia is to capture it here. Yes. Uh, f- for lots of different reasons, but mainly it, it’s not going away, so we, we do have to keep up- Yes … I think. No, indeed. Indeed. Uh, and do it in a responsible way, obviously.
Mm. Um, so it’ll be interesting to see how that plays out. But while everybody’s got their eye on that shiny ball- Mm … you go back to, uh, like where does dealing with technical debt sit on the list now? Well, exactly right. Yeah. I mean, look, if- ‘Cause that, ’cause you can’t do it all. You can’t do it all, and I feel like we’re gonna see more and more, um, press articles, unfortunately, about businesses coming afoul with this, right?
I just, it feels like the risks have increased markedly now. Well, anybody who wants to mess with your organization can use AI- to their advantage to exploit your weaknesses. Yep. And you’ll have lots of them if you are carrying a lot of technical debt. Yep Um, so you gotta deal with that. You gotta deal with probably tipping most of your budget into keeping ahead and figuring out how to do AI- Yeah
in your organization, and maybe blowing it all on shadow AI tokens, I don’t know. Well, I, I just, I’d love- ‘Cause that’s gonna be the new finops, is- like, how do you… How do you not blow all your coin on tokens? Look, I, I mean, I’ll speak candidly. I mean, I… Look, our business is going through… Now, we’re lucky, right?
Like, within the business that I work for, I mean, we’ve, we’ve got access to everything that we need. But good God, right? I mean, the costs can go through the roof. If you’re not careful. If you’re not careful, right? But there are also some clever ways to, um, manage your tokens, and I, I can… I shouldn’t say this probably out loud, but I can, I can foresee a situation where those of us with teams and P&L responsibility will be given budgets on this stuff, right?
And it’s, it’s down to us to manage within the tokens that we’ve been allocated, and there are absolutely strategies for doing that, right? And we’ve talked about, obviously, local processing should you have access to the kit, but that’s not free either, right? I mean, H100s and 200s cost a fortune, right?
Yeah, right. So it’s not like that’s an obvious answer. But I think token economics and responsibility, like using skills for example, right? So you don’t constantly fetch and, and ask the same question over and over again. So what is a CIO to do now? You, you… Like, my top priority, uh, if I’m with the cool kids, is to be, be doing AI and implementing that- Mm
and figuring out how to not spend too much and- Yeah … all the rest. But at the same time, we can’t afford to go down or- No … be hacked or be in the newspaper, and that box that’s got dust on it, we’re talking about, Do you reboot it or not? You hope it comes back. Like, how do you deal with all of this all at the same time?
I don’t know. I don’t know. But, um, to our audience out there, like, let us know, right? Like, what, what- How big a priority is technical debt for you now, and do you worry about it more than you did yesterday?
I, I, I, I do. Like, as a critical infrastructure guy, I’m- Mm … I- I’m always, I’m talking to customers about it.
We’re, we’re, we’re getting it on registers. You’re coming up with remediation plans. Yeah. You’re having a plan. You’re, you’re, you’re making an effort. Mm. Um,
but it’s not the sexy thing to do.
No, it’s not. It’s, it’s … And it only shows up when it breaks. Yeah. That, that’s the worst bit. That’s right. It’s all fine until it’s not.
Yeah.
You say to your mate, “What have you been working on lately?” “Oh, technical debt. Oh, I’m so sorry.” Yeah. I’m so sorry.
I’ve gotta go down to the data center and unplug this old thing and- Yeah. It’s making some noises, smells bad. Yeah. Exactly. And so this is what, this is what everyone’s dealing with, I think.
Mm. Um, uh, yeah, but I, oh, look, I’m hopeful. I remain optimistic we could, uh, uh, the industry can leverage advancements in, in AI- Mm-hmm … to help identify some of these things- Right … and, and resolve them- Well, I feel like there’s in a smart way … there’s an opportunity here out there for the clever startups.
You’ve heard us talk about 93% of organizations have technical debt. 24% of those only feel comfortable with their plan to address it, so that feels like a business opportunity for someone. I reckon we should ask Claude. I think we should ask Claude. You’ve probably got more tokens. You could, Okay. I, tell you what, the day they take my tokens away, I’m gonna have a proper tantrum.
All right. All right. Well, that’s, Yeah look, there, there’s a bit going on there. I do, um- Well, what would I, what would I say to wrap it up? I, I’ve spoken before about technical debt. Yes. I, I think, I think it should be a KPI. I agree. Like, like a, a board-issued KPI that everybody buys into that says there’s a, there’s an ex- there is a certain line that we draw- Mm-hmm
that is an acceptable amount of technical debt- That’s right … to carry, and, and we cannot ignore it. Mm-hmm. That’s right. And I, I think the, this latest incident just highlights it again. It does highlight it, Jono. These, I mean, let maybe we can put it in the category of, like, what seemingly immovable objects, right?
We… Technical debt, it should be a KPI, board level driven. That way it could never be just ignored and deprioritized and, um, constantly put on the back burner. That’s one thing. We talked about native services in the same frame, you might remember, right? With our good friends at AWS, right? And so, uh, you know, if you embed yourself with a cloud native technology that’s specific to one hyperscaler, fine.
If there are existential threats to the business and you need to, say, shave the 20, 30% off your budget, how do you migrate it off that thing? It’s the same thing. right. It’s that, so you got that sort of technical debt KPI- Yeah … and that agility or portability KPI. Yep. Right? So you- Same thing
so you can maintain agency and sovereignty over your, over your tech- That’s right, Jono … uh, that, that drives your business. We’re here to help. We are. Yeah. We’re here to help. There’s, um, I’d be interested actually to hear from our audience as to how they’re dealing with this. Mm-hmm. And, uh, I know we’ve got our confessions page up.
Yeah. Um, maybe, maybe we’ll get a Telstra related confession. Oh, you know what we should ask? Who’s got the worst technical debt? It wasn’t that long ago one of my mates said to me they saw an NT4 server kicking around the other day. You know what I mean? Yeah. I, I ran into some 2010 things operating in the field.
All right. Um, the The organization shall rename, remain nameless. Yeah, all right. But, uh, th- they’re out there. All right. I reckon we should get some confessions on the worst technical debt. Give us the best ones, guys. We want to hear about technical debt. We’ll talk about it, and maybe, maybe if, if you submit it, we might, um, provide some advice as to how we might go about resolving it.
All right. We’re here to help. We are here to help. All right. Uh, great chat, Naran. Yep. Where can you find us? Where can our listeners find us? Thanks for listening again. Um, so look, you will find us everywhere you get your podcasts. That’s Apple Podcasts, it could be Spotify. Of course, we are on YouTube.
We’re absolutely everywhere. Thank you very much to all our subscribers as well. We really care about that. John and I talk about subscribers all the time. We love them, and so thank you for subscribing. It’s the best way to get the content. We are, of course, produced by Pru Loon. Our digital producer is Karina Aguilera, and you can reach out to us at pru@thingsreasons.show, and you can also visit the website, thingsreasons.show.
It’s pretty awesome. There’s good things there, especially the confessions page. Yep. Give us your best confessions. We’ll read them out on the show. Uh, that’s a wrap, guys. That’s a wrap. See you again next time. Thank you. Thanks for listening.