How to Build AI Capability Without Chasing Every New Model, with Patrick McQuaid
Most organisations are asking which AI model they should use. Patrick McQuaid thinks that is the wrong conversation.
Patrick is Group Head of Data, Analytics & AI at NGM Group. He joins James MacDonald to explain how leaders can build practical AI capability without getting trapped by hype, vendor lock-in or an endless cycle of new tools.
They discuss how NGM approaches AI inside a regulated banking environment, where privacy, risk and customer trust make it impossible to turn a weekend prototype straight into production. Patrick explains why teams need both strong foundations and useful short-term wins, why the people closest to the work should own their technology costs, and where generative AI should not be used for decision-making.
The conversation also explores what AI means for technical careers. Patrick shares why he hires for attitude, curiosity and problem-solving ability, rather than waiting for candidates with years of experience in technology that has only just emerged. He and James discuss the continuing value of junior engineers, the rise of AI enablement and orchestration roles, and the human skills that remain difficult to automate.
If you are responsible for introducing AI into an established organisation, hiring a technical team or preparing your own career for the next wave of change, this episode offers a grounded alternative to the fear and hype.
About the Guest
Patrick McQuaid GAICD is Group Head of Data, Analytics & AI at NGM Group, one of Australia’s largest customer-owned banking groups. NGM Group operates the Greater Bank and Newcastle Permanent brands.
Building Tech Teams is produced by DayOne.
Listen to the episode
Transcript
Auto-transcribed with AI.
Patrick McQuaid [0:00]: If you’re arguing about what model to use, you’re having the wrong conversation. There’s a lot of people out there trying to instil a FOMO approach to this, right? If you’re on LinkedIn, they’ll be like, oh, you need to do this, you need to do that. You’re not going to get the full benefit of AI unless you do it now. But there’s also a lot of hyperbole. I would rather have a couple of core people with depth and pair them with people with the right mentality who are just keen to get in there and learn and soak it up and apply lots of energy and have a go. It’s not enough to be right. You’ve actually got to be engaging, you’ve got to find allies, you’ve got to build support, you’ve got to build relationships. One of the core skills in the future that is unassailable by AI is the ability to build relationships and talk to people and communicate.
James MacDonald [0:46]: That was Patrick McQuaid. He’s Head of AI, Data and Analytics at NGM Group. There’s so much hype around AI at the moment. Everyone’s vibe coding, building proof of concepts. Far fewer people are actually rolling out AI into enterprise. NGM Group — banking environment, highly regulated. And Patrick’s talking about how to get AI into production whilst also dealing with risk and privacy. If you’re facing the challenge of rolling out AI into your organisation, into your company, and doing it safely, this could be the episode for you.
Welcome to another episode of Building Tech Teams. On today’s episode, we have Patrick McQuaid, Group Head of Data, AI and Analytics at NGM Group. Welcome, Patrick.
Patrick McQuaid [1:25]: Thank you, James. Nice to be here.
James MacDonald [1:26]: That’s a big title — Group Head of Data, AI and Analytics. What set you up? Tell the audience a bit of your background that’s led to being in a role as big as this at a bank.
Patrick McQuaid [1:38]: Probably started when I was in consulting. I was promoted really quickly at one stage and I realised that I was quite out of my depth from a technical perspective. Up till that point I’d been a technical leader. And I think I just adapted and realised I was good at managing technical people, leading technical people — and we can get into how I’m good at it — but that’s formed a bit of a pattern over my career. So I’ve often jumped into new areas. I’ve only spent two or three, or maybe even four or five years there, but jumped across areas because I knew going into them, hey, I can grab a bunch of people who know a lot more than me about stuff and make them work well together.
James MacDonald [2:23]: Sounds like this will work well on a podcast titled Building Tech Teams. I’m going to get straight into that part right now.
There’s a lot of talk about AI at the moment. There’s a lot of talk about technology. There’s also a lot of talk about the changing nature of roles and how maybe managers might be the roles that are getting squeezed. How important is it for someone in a role like yourself to understand the technical nature of it? How technical do you need to remain to be successful?
Patrick McQuaid [2:50]: I don’t think I’m very technical, so that’s maybe a bias to my answer — although maybe some people might think I am. I think you’ve got to understand a lot of first principles and fundamentals. I lead data teams and I’ve never really done SQL in production or anything like that, but I know good questions to ask, I know how to prompt people, I know how to manage people and get experts together, like I said.
So a common mode that I’ve used for the last 15 years or so is, I ask people, what are the options? And I get people in a room together and say, give me the pros and cons, and I ask questions, and that’s when the answer falls out. Often they don’t need me to decide, which is really my ultimate goal. And then if I do decide, I know at least I’ve got pros and cons. So I’m not too into the technology, but some of my peers are. So I think you can be both. But I’m saying you don’t have to know the tech really deeply.
James MacDonald [3:53]: Perfect. But you did mention the knowledge of first principles.
Patrick McQuaid [3:56]: Yeah. I mean, look, one of the things you often try to ask when you’re making big decisions is, is this a two-way door or a one-way door? Can I reverse this decision later on, or am I making a decision now that locks us in? Those are the kinds of first principles I’m talking about. Some of them are not technical. But you also ask, how does this fit in with where we’re going from a cloud perspective, for example. So there are technical aspects to that as well. But generally a lot of what I ask is technology-agnostic, and I lead in that way, because at the end of the day, technology is a tool, right?
James MacDonald [4:36]: Yeah.
Patrick McQuaid [4:38]: I think it’s especially important in the age of AI, with new models coming out every three months. And I actually recently spoke to a group of leaders in the Hunter, and I sort of said, if you’re arguing about what model to use, you’re having the wrong conversation.
James MacDonald [4:55]: Correct.
Patrick McQuaid [4:55]: Because that model will change — six months, nine months, whatever, it’ll change. So you’ve got to ask more about, how am I deploying this? What are the pros and cons of the way I’m doing it? What are the things that lock me in versus give me freedom?
James MacDonald [5:12]: I think that freedom one’s a big one that a lot of people are overshooting at the moment. Being agnostic — whether it be model-agnostic, company-agnostic — what you’re actually tying yourself into, having that question is really important at the moment. Because as you said, we’re talking about token cost changes, we’re looking at model changes. Some models are going to be better than others. I think in the future we’ll be having more companies look at local models as well, just from a cost perspective and also a security side of things. So architecting things in the way that gives you flexibility down the track.
Patrick McQuaid [5:42]: Absolutely. And it’s not to say that the technical details are not important — I want to be really clear on that. I do care about the technical details. But I know that for me to get across the technical details in the amount of detail that would be required to make a good decision, it’s impossible, it’s volatile, it’s a waste of my time — given I’ve got people on the tools who are experts and whose job it is to be there. My job is to sit above that and to surf across it and to look ahead and look around. So it’s important, but it’s not my job.
James MacDonald [6:15]: Perfect. And I think it’s a different role being in an enterprise organisation like yourself versus if you’re a startup who’s building an AI-native or a fully agentic platform from the ground up. They’re two different lenses and ways of going about things. If you’re in that sort of secondary role, you need to be on top of the tools, learning how to build AI-native from the ground up. As opposed to, hey, we’re a big organisation, we’ve got a highly regulated environment, where can we use AI and data within the org?
Patrick McQuaid [6:45]: Yeah, for sure. I mean, look, I do some vibe coding at home for some personal things and I have fun with that. I didn’t know Python before I got into it — I had to do some stuff at work and had to learn it. So I actually dispute the fact that vibe coding means you don’t need to code, because to troubleshoot it, I had to. But the freedom to use those sorts of tools is not there when you’re dealing with people’s money, people’s personal data, livelihood, vulnerable customers — all these things that mean what we do has to adhere to a certain level and a certain standard. That means I can’t just whip something up over a weekend and say that is now in production. It just doesn’t work like that.
James MacDonald [7:31]: Yeah, I love the fact when I detect leaders that are playing with some tools themselves. It doesn’t mean, as you say, you’re building something for production, but you’ve got an understanding of what it’s capable of. Is there something that you think is important for all your other leaders, the people that are sitting under you, to at least be aware of — the different technologies, what’s available now, what can be done now? How important is that for you?
Patrick McQuaid [7:53]: It’s not critical, but I think it is important still. I pushed myself to do it because I’m not usually the sort of person who does that. I don’t like to be on a laptop if I’m not working. I like to be spending time with my sons, or out there, or exercising, or whatever. But I did feel that, particularly in the AI space, it’s important to get in and just play with it a little bit and understand what it’s good at, what it’s not good at. And I learned some really eye-opening things, or maybe even just reinforced stuff I’d read. So it’s not critical, but I’d say it’s important for a fast-moving tech to play with.
James MacDonald [8:31]: You mentioned that you’re playing with some tools at an individual level. I think a big question that’s coming up across enterprise at the moment is tooling and what you can actually give to your team to use. Otherwise you get this shadow IT that goes on and people are going to be using some form of tools. If you’re locked into a Google environment and everyone’s got access to Gemini, but they want to use Claude in their own time, they start putting some company IP into their personal Claude account. What’s your opinion? What’s your advice to other people that are starting to deal with these challenges around shadow AI?
Patrick McQuaid [9:06]: Well, for a start, I think you’ve got to give people some tools, something, right? And so we’re a Microsoft shop, so we use Copilot. The good thing about a lot of these enterprise platforms now is they do allow you to choose your own model at the back end if you want to. Now, we have disabled most other models because, A, we want to protect data — we’re not sure on some of these things where the data sovereignty is, where the boundary is — and B, we do want to control cost. Some of these other models, you can start to run up token costs and things like that, versus the standard where it’s all included. So you do need to provide some capability.
And our CISO and our CIO have both left it open to use other models should people have the need to. But that’s the key point — the need. Lots of people may have a preference, but I don’t think there’s a lot of need for different models for most corporate use cases.
James MacDonald [10:11]: I would agree. I’m going to go back one. You mentioned your vibe coding, you’ve learned some Python, you’ve had to learn some Python to understand the troubleshooting side of things. There’s talk of engineers becoming — there are multiple stories. A, that it’s the death of the junior engineer, because junior engineering is not needed anymore. Secondly, there’s other people saying junior engineers now have better tooling and can upskill quicker than any engineers before them. Others are saying senior engineers are now 10x better than what they used to be, because they understand those core principles, and then they’ve got better tooling, so they become even better and faster.
Again, the importance of that core understanding of software engineering, or good data practices, how to build a data pipeline, how to understand architecture — how important is that for the people that you’re hiring into your team, or the successful people that you’ve seen grow careers?
Patrick McQuaid [11:09]: Let me start with the reason I had to learn the code, which was I discovered it was faster. You can probably do it without loading code. But I’ll give you the example that prompted me to get into it. I was doing this on a Mac. The initial Python-generated code was all for Windows, so the paths were all wrong. It was faster for me to identify them — wait a second, why is that a forward slash than a backward slash? — than it was to go to the AI and say, why is this wrong, and for it to maybe prompt me and eventually find it out. So there’s a bit of, I’m not scared of code, so I’ll look at it.
In terms of your broader question, though, experience is still great, right? And so you always love to get somebody who’s experienced. But the reality is, if we all said we were looking for people with four to five years of AI engineering experience — the people who’ve really done it are pretty rare. B, they’re probably commanding larger salaries than I can pay in a regional bank, if I’m just brutally honest. And they’ll probably be at some of the big locations, and we do try and preference Newcastle hiring where we can.
So I’ve taken the approach recently of, a bit of experience is great, but I would rather have a couple of core people with depth and pair them with people with the right mentality. Which is — to be stereotypical — sort of younger people who are just keen to get in there and learn and soak it up and apply lots of energy and have a go. That mentality, I think, is probably more important these days than ever before.
James MacDonald [12:58]: I agree. And I think it doesn’t matter what stage of your career, whether you’re a junior or somebody that’s mid-senior that’s willing to upskill, willing to dive into the tools and try and find the answer. Where it ties back to at the end of the day is that problem-solving ability. And I think what you’ve mentioned there — attitude and problem solving — are the two real core fundamentals that if you have them, you go a long way.
Patrick McQuaid [13:21]: Yep, absolutely.
James MacDonald [13:23]: There’s a lot of talk at the moment — when you mentioned engineers just before — and there are various threads that people are pulling on, the death of the junior software engineer, those roles don’t exist anymore. To counteract that, we’ve got junior software engineers are going to come in better, more equipped, better tooling, learn faster because of the new tooling. And then you’ve got these seniors that have now got this great tooling, core fundamental understanding, better tools, now they’re 10x better. How do you go about looking for the people that are going to join your team? Obviously competing for those 10x engineers, AI-native engineers, is going to be very expensive and hard to hire for.
Patrick McQuaid [14:04]: Like you said, the reality of finding someone with as much depth of experience as you might say an experienced person should have, who is in a regional location, coming to work for a regional bank, is probably pretty low. We have found some great people, and I’ve actually been really impressed with the talent at NGM since I joined.
So the way we try and build that person: we’ve got people who understand banking — that’s a tick. We’ve got people who understand software engineering in a regulated industry — that’s a great tick as well. They might understand the latest tools, but then, not a lot of people do, and the space is moving so quickly. So I tend to think that we could bring in people who have the right attitude more than anything else. Hiring for attitude is what I call it. And you might say stereotypically that’s your young, hungry engineers. It could be an experienced person as well, who’s just never lost that ability to be curious and get in there and have a go. And that’s what I’m looking for. People who can demonstrate, I don’t know the answer, but I’ll get in, I’ll have a go, and by the end of the week I will. And that’s really priceless in this fast-moving area.
James MacDonald [15:19]: I agree, because as you said, tooling’s moving so quickly, and to say you’re fully across everything would be, for the vast majority of people, a lie. And things are going to continue to change over the coming days, weeks, months. So that attitude part — curiosity, that word — I think curiosity, problem solving, those are the sort of core skills of people I’m seeing jump in their careers as well.
Patrick McQuaid [15:42]: Yeah, absolutely.
James MacDonald [15:43]: You mentioned back before — it was another point I wanted to pull out — the models, and why you’re looking at models from a costings perspective as well. I think it might be a standalone role in the future. I know it happened with cloud many moons ago, where everyone was about, we have to be fully cloud-native, and then people were seeing their AWS costs or their Azure costs or GCP costs go through the roof and, oh yeah, this is more expensive than on-prem, and then people started to reverse out. How do you see that? Is that something that sits within the data and AI team, or is it a finance role, responsibility? How do you see that play out ongoing? And how have you tackled it at NGM Group?
Patrick McQuaid [16:31]: Yeah, look, I’m a big believer — whatever the org model is — that the team needs to control the cost, because then you can make the right trade-offs. I think where you get disconnects is when, like we’ve had in the news recently, there’s a management push to token-max or something like that. Right? That’s crazy.
So I think, even if it’s finance that is the watchdog, or some FinOps team in the cloud engineering team — and that’s the model we have at NGM, we have a good FinOps team that manages the overall cost — I have, from the very start, instructed my team to worry about cost. Or to worry about tracking, sorry — it’s not always about cost, because it doesn’t come out as cost sometimes. But we’re tracking, and to work with the FinOps team. And that’s just a matter of, oftentimes depending on what platforms you have, building the dashboards, building the warnings, building the monitoring.
But also — and this is why I think it belongs in the team that’s using it — you can start to build lead measurements. You can start to realise that these sorts of queries, or this sort of thing, will lead to this level of cost. And once you start doing that, you can start to program a bit of perspective into the way your teams use certain tools.
James MacDonald [17:53]: Do you think we’ll have a standalone role, like a token optimisation person or a token costing person, in the future?
Patrick McQuaid [18:01]: I think it’ll be part of somebody’s role more than anything else. Probably part of that is because I work in a relatively small organisation, given other organisations I’ve worked at. And their role will be not so much to monitor and track it, but to set in place how to monitor and track it, and make sure teams are doing it. Because in the end, the accountability should sit with the team, should sit with the leader, to justify any cost and to make the trade-offs.
James MacDonald [18:33]: So for a smaller business than yours, to go about this and be on the front foot, it would be just looking at setting up some form of reporting, dashboards, QA, lead indicators, alarm bells I guess of sorts, to give yourself some form of —
Patrick McQuaid [18:49]: Yeah, I think you’ve got to work with your provider to know what tools they have. So for example, we work with one of our platform providers and they’re very helpful, because they know if we got burned by cost, it’ll actually lead to a whiplash the other way, right? So you’ve got to work with them to set at least thresholds and probably some warnings as well. And then over time, I think you develop the lead indicators yourself. You kind of have to, because the provider doesn’t know what you’re doing until it hits them. So you’ve got to figure out, actually, when we do this thing that touches these four systems or this much data, that’s a big-cost one. So let’s be careful about that.
James MacDonald [19:35]: Enterprises have different challenges to SMBs, but I think one of the bigger questions going around at the moment is, hey, we’re a business or an organisation, we want to do something. Do I grow the skill set internally, do I hire the skill set externally, or do I go and use a partner? Or is it a combination? Is there no one right and wrong answer? Can you just talk to me about how you envisage those three parts coming together?
Patrick McQuaid [20:01]: I always think it’s a combination. And again, everyone’s situation is different. But my perspective has always been, I don’t want to rely only on externals. And this is part of — you go back to what we said before about hiring people who are keen to learn. You pair a person who’s keen to learn with an expert that you can source, maybe beyond the budget you can pay, and all of a sudden you have a really nice combination, hopefully.
So I think it is a combination. I would use proper consulting-slash-accelerating partners for when you have a real pinch point that you need to accelerate, that’s high value. But other than that, I would always try and partner with contractors who understand the things, and then have them work with a team with a specific goal of upskilling the team as well. So it is all three.
James MacDonald [20:50]: I believe there are some challenges out there working with external providers. They come and build something flashy, beautiful, it works, then the agreement ends and you’re left to carry the maintenance of it. And I think there are a lot of missed conversations at the moment around the ongoing maintenance, QA, performance optimisation of AI within organisations. How do you approach that? How do you approach not only a build cost, but ongoing maintenance? How does this play out longer term?
Patrick McQuaid [21:21]: So I would say, in our industry, we’re a long way — well, we should be a long way. I’m sure there are people who are experimenting who may be further forward than us, but we’re a long way away from using generative AI type models or agents in really critical processes. And the reason for that is because they’re critical processes, and I cannot afford to have a probabilistic outcome which is vulnerable to model drift or bias right now. You will have heard about people saying they’re using models in decisioning, credit decisioning, but almost always those are proper deterministic, predictive machine learning models that we’ve had for 30, 40 years.
James MacDonald [22:07]: And I think that is the key under the hood for a lot of organisations. It’s traditional engineering, machine learning, automations. And that’s where the vast majority of benefit is really coming — with AI as a flashy tool, flashy headlines, LLMs actually being a much smaller part of the majority of actual AI projects.
Patrick McQuaid [22:33]: Yeah, I think that’s the case when you talk about core business processes. Where generative AI can help with core business processes is usually because your inputs are unstructured. So it helps with interpreting, organising, those sorts of things. A good example: we do some process documentation work right now. You could spend a lot of person-power getting things in the right format to feed them into a process documentation engine — or you can use an AI. And that’s been very successful. Really sort of trivial use case, but very high benefit, very low risk.
James MacDonald [23:14]: And I really like that point. That’s where a lot of the conversations I’ve had at enterprise level — it is actually using AI for something in a small use case, which might be seen as quite trivial, but actually has a lot of benefit to the organisation. Because if you’re talking about labour costs there, not only are you saving labour costs, but if you’re actually reapplying those humans to better tasks, or higher-value tasks, or tasks they actually enjoy — quality of the job for the human does better, more accuracy with the actual AI. The cost-benefit for the organisation is phenomenal. And those use cases in enterprise, just because of the sheer volume, actually play out really significantly, right?
Patrick McQuaid [23:53]: Yeah. And look, the use cases where you see direct-to-customer, or large-scale direct-to-customer, are generally retail. Apart from — I know Commbank has a triage agent and things like that. But the Commbank example, they are there to help find a human or a policy faster. The ones that are generally doing proper generative AI facing customers, they’re in the retail space, and the consequences of getting that wrong are usually quite low. Air Canada examples notwithstanding, where AI agents said somebody was entitled to a refund and they weren’t. But if it’s the wrong colour of shoes or the wrong size of shoes, those sorts of things are easily rectified. Financial advice, financial savings, what to do with your money — those sorts of things are far too important to leave to that at the moment.
James MacDonald [24:54]: Yeah. And I think the human in the loop is what we’re talking about.
Patrick McQuaid [24:57]: Absolutely. And there’s various examples of that. You’ve got human on the loop, you’ve got expert in the loop, and other things like that. But a lot of times it comes down to, this is too important to leave it to chance. Our regulators are also saying that sampling is no longer enough. You actually have to have more stringent in-process controls. And plus — if I can spruik NGM for a second — we’re a community bank, right? There’s a lot about being in the community, being personal, that is core to our brand as well.
James MacDonald [25:36]: I agree. NGM Group — let’s open the door, mate. You started there in a temporary role to start with, and you’ve been there for a little while longer now. But I’d love to understand — you started in a temporary role. What did you see at the start? What was the environment that you walked into? Let’s start with that, and then we can expand.
Patrick McQuaid [25:56]: I’d worked with the CIO before, and he knew I was available, and this was my area of specialty, and I’m not afraid to travel — I’ve done it for most of my career. So he asked if I could come and help them out. And I got here and it was a fairly tumultuous time. The data re-platforming project was stalling, and there were two teams that were coming together. So there was a lot of org change in there. A big job, complex job, even though the organisation’s smaller than I’m used to.
And what I found when I got here, which actually surprised me — and I must admit, hand on heart, I thought I would find not the level of person that I’m used to dealing with in some of the bigger organisations. And I was arrogant in that respect, and I was wrong. Great people here, and that was the first thing. Great people.
I also felt like I could make a difference. I felt like my style was going to work here. And one of the things that everybody likes to know is that they’re going to add value. And within a month, I think, of being here, I could see I was going to add value. I could see great people here that I really liked, and a really encouraging environment, from the executive as well. And so, like I said, within a month, six weeks, I went to my boss and my other boss and said, I like this place, I’d like to stay and see what I can do here, if we can make things work out nicely.
James MacDonald [27:37]: And what does it — it is a big title, Head of Data, AI and Analytics. What is your remit? What’s the environment that you’ve walked into? You gave a bit of a mention of it. What’s the end goal? What does that roadmap look like for you?
Patrick McQuaid [27:52]: Yeah, so most head of data roles, or chief data officer roles, a large part of it is consolidation of data. Not just for the point of consolidation, but consolidation of data because that leads to better access, lower cost, better transparency and less risk. So that’s really the core of a lot of head of data roles — how do I get it onto one big platform, and then how do I expose it to my internal customers for value? And that’s the project that I walked into that was kind of stalling.
I’m really pleased to say we’ve almost finished that project now. So there’s a lot of data that’s been centralised in a really great way. You’re never done — absolutely never done — but I could say we’ve got 60 to 70% of the way there, and we’re really going to start to add value now. So now we’re moving to a phase of, how do I leverage this data better? I’m no longer just building the foundations. I’m sort of saying, well, who needs what, and how do we talk to people about what we can provide them with data?
So that’s the data side of things. And then very closely related to that is analytics — which, in this conversation, just think of as traditional analytics. Because once you have your data in a good spot, and you have it on a cloud platform, then the thoughts immediately turn to, how do I derive value? And that is basically analytics in a lot of cases.
James MacDonald [29:21]: How do you manage — and it’d probably be from a business perspective as well as a personal perspective — the dual nature of, we can’t do anything with AI automation unless the data is in a good spot and we have clean access to quality data. Finding that drive balanced with, hey, we have to sort of take two steps back to get it in the right spot before we can do anything new or flashy with it. How do you both individually deal with that, but also at a company level deal with that? Where we have a drive and a desire to be moving forward and not missing the boat, let’s call it, whilst also knowing that to do that we actually have to — can you do both things at once? Do you have to take the step back to go forward? I’d love to get your opinion on that.
Patrick McQuaid [30:07]: Maybe I’ve been lucky here — I’ve never had to completely take a step back. So I’ve always approached, in the last 20 years anyway, sort of every executive or leadership role, saying I need a two-speed approach. I love building foundations and big, complex, cool things, but they take time, and the organisation doesn’t have the patience for those long times. So the first speed is, build the foundations and look for some big wins — big hairy audacious goals, they used to call them. But also, what are the value increments you can do along the way?
Now, we — having said that, last year in NGM we probably focused pretty much on just the big foundation, but we also found ways to add little bits of value along the way, like streaming data to places and things like that. But it was mostly focused on building that. And I’m thankful to the organisation, they had the patience to do that. But yeah, almost always I think you have to approach it with two speeds. One is, find those quick wins, the value drops you can do along the way, while you are building out your platform.
James MacDonald [31:28]: I like that advice. I think that’s a challenge that a lot of — whether it be a CTO in an SMB-type business where they don’t have a head of data, or even head-of-data-type roles — where you have that challenge.
Patrick McQuaid [31:42]: Yeah. And look, big organisations often get this wrong, I think, as well. I have worked at some orgs where the goal was measured by how much data was in the target platform, and I think that’s the wrong measure. I think you should be looking at how much of a business unit’s data usage is in the platform. The awful thing is, you can have 80% of your data in the platform, but if every business unit only has 70% or 60% of the data they need, you have no value.
So that’s another way of looking at it — pick your target area. Usually it’s marketing or customer engagement, because that’s the fastest value cycle for data a lot of the time. And say, how do I get you up and running on my platform as quickly as possible, while I’m building out all the foundations and all the things to bring everybody along the journey?
James MacDonald [32:42]: I like that approach. I think it’s a common problem at the moment, and something that will be ongoing. And it’s very different for enterprise, or a traditional organisation, that challenge, versus building AI-native from the ground up. There’s a lot of hype at the moment with AI-native, and I think if you are building a company from the ground up, significantly different problems and challenges — also some things that are a lot easier — rather than trying to re-architect a solution that’s been around for tens of years, and you’ve got tens of thousands, hundreds of thousands, millions of records sitting in there, and then facing that challenge.
Patrick McQuaid [33:17]: I do think — and I’m not in an AI-native business, but from what I’ve read and from what I know about AI — there are options available to you. Again thinking about that two-speed approach, right? Not all AI usage today, because of the generative AI nature of it, requires big data sets and high-quality data sets. Like I said, one of our niche, most successful use cases is just processing documents. Now, you still have to have a bit of hygiene — you have to make sure it processes the right documents and they’re in the right place — but there’s not a lot of database work and building-platform work you have to do to make that a valuable use case. So I do think generative AI opens up some faster value use cases than having to build the traditional big “put all your data in one place so you can run machine learning models across it”. They’re still valuable — absolutely, and that’s probably where the big value is. But you can find ways now through the new technology.
James MacDonald [34:21]: How far in advance do you build a plan? Obviously you’ve got bigger, grander visions as well with the journey at NGM Group, and you’re still quite early on in your journey. How far into the future are you building the end state, with — as you said — the acknowledgement that things are going to continue to change?
Patrick McQuaid [34:40]: Depends what you call a plan, really — that’s a weasel way of saying it. But I have these milestones or goals that I can see two-plus years out. Detailed delivery, we only do on a quarterly increment. At the moment we do quarterly planning, so I’m probably — that means I’m six months, because my job is not to look at the current quarter, but to look at the next quarter mostly. So we’re probably six months. But beyond that, it does get fuzzy. So I’ve got these ideas of, these are the things I need to do in the next two to three years, these are the things we’re working on right now, these are the things I’m going to get to next quarter, and then everything else is sort of in the fog — it’ll become clearer as I get closer.
James MacDonald [35:24]: I really like that approach. I think, as you said, things are moving quickly and things are changing, and the grand plans of, hey, here’s my 10-year vision — maybe the vision’s correct for 10 years, a big overarching vision, but detailed plans, I think a lot of people can get stuck in over-planning, especially with how quickly things are moving.
Patrick McQuaid [35:45]: Absolutely. And I used to be a project manager early in my career, so I’ve done some planning and detailed things like that. But I would say you’re better off being quite flexible now. And especially for middle-sized orgs like us, I think we’ve got a great advantage if we are open to possibilities.
A good example, if I can — agentic. The agentic wave hit when I was at Telstra, and we were talking about this. There was a whole lot of complexity — you were going to have to build to enable agents to work safely in the org. And then nine months later, Microsoft, Anthropic, other areas came out with fully fledged agent platforms where they’re almost — well, they’re low-code, no-code — and you’re like, wow, I’m so glad we didn’t build these crazy hairy layers to enable agents to execute safely, and we can leverage that from our partners. So staying open to things — like I said before, open doors versus closed doors, committing what you need to in the short term to make things happen, but always keeping your head up and being flexible is key.
James MacDonald [37:04]: I like this. I like it a lot. It opens the door to a question. You’re obviously not a product-based company, but there’s a lot of challenge at the moment for pure product plays and product startups in particular, where if you build something, trying to build a moat around that to make sure you’re long-term successful — and Anthropic or OpenAI can’t just come and build it and roll it out to all their customers within a day, a week, a month. That’s a very real challenge. How do you look at it from that product side of things? Because things are moving quick. I guess you’ve probably answered — be flexible and willing to see what’s out there and test and try it. But I’d love your opinion on that product side, and how do you look at vendors?
Patrick McQuaid [37:45]: It’s probably one reason why I’m not in products, because I don’t have really good answers. But there are products out there where I’m still curious as to how they’re delivering value. If I’m honest — so Harvey AI, the legal AI. I’ve heard various people describe it as just a wrapper around somebody else’s LLM, but it’s doing really well. So there’s something about being specialist, probably good customer service, good industry knowledge, which benefits you. And I think that is where the value actually is — knowing your target industry, knowing your target customers, means your product builds a moat around itself because of your ability to serve and engage. That’d be my advice.
James MacDonald [38:34]: Which is probably no different to NGM Group, right?
Patrick McQuaid [38:36]: Yeah. I mean, look, not being a product person, I don’t want to oversimplify, but yes, the big model companies are going after verticals, because that’s where they can see you can get layers of value rather than just being a commodity in the back end. So it is interesting. But I do think software bloat is still a real thing. Siebel was the Salesforce before Salesforce, and then it lost its way. These big companies, I think, get homogenised and bloated really quickly. So if you’re a nimble, specialist, industry- and customer-savvy focused product startup, I still think you can create value.
James MacDonald [39:19]: For non-software companies like yourself, how far would you go in building your own software to be used internally? You do have a software engineering team internally, you’ve got really strong capability there. How far would you go? You’re not going to rebuild the entire platform, but are there use cases internally for building unique pieces of software unique to your needs? How far would you take that, as opposed to using something off the shelf?
Patrick McQuaid [39:47]: This is my own opinion, so — I don’t think it makes a lot of sense to devote a significant amount of your own resource outside of your core competency. And our core competency is not building software. We do a good job and we have great engineers, but they build software in the service of our core competency, which is our customers and our communities and servicing their needs. So I can’t see us devoting the energy and resource to both building and, importantly, supporting things that are part of core processes.
But the answer is, on the outsides, maybe. And it’s just wherever that ring sits — you think about things that are non-core, maybe a ticket management system for IT. But even that, there’s a lot of complexity in a big enterprise organisation, and while it might seem on the surface to be something really easy to build, there are packages out there that do it, and do it well.
James MacDonald [40:59]: Yeah.
Patrick McQuaid [41:00]: So why would I do that? I think if you found something that is niche, that is a point of value or a point of difference in your core competency, then that is when I would build it. But banking, from a software and a behind-the-scenes model, is fairly commoditised, where the biggest difference is in customer service and your ability to service your customer and interact with them and build relationships and deepen those relationships.
James MacDonald [41:34]: On that point, how far does technology play a role in that for an organisation like NGM Group? Traditional banks 10 years ago — everyone’s got your outlets, you’d walk into a branch, you’d be spoken to by a human being. Branches across the board for all banks are limited in numbers these days. Technology plays a bigger part of a customer’s interaction with the bank, whether it be through an app, a phone, an agent, a chatbot. How do you look at technology from that side, with customer service obviously being one of the biggest differentiators and value-adds for you?
Patrick McQuaid [42:09]: Yeah, I mean, I’ve worked in banking technology since 2008, so I obviously think it’s really important, and that’s why I’m there. Even if you were to go to a model where all of our interactions are human-related, the reality is all the humans need technology to support them as well. Our branch staff, without good tech underneath them, are hampered. So we’re key to the staff-oriented or assisted model. We’re key to where most people interact with banking every day, which is their app and their internet as well. So I think technology is a key part.
We have to remember — there was a mantra going around a while ago in some of the big banks that banking is a technology industry. I don’t believe that. But I do believe technology is a core part of the banking value.
James MacDonald [43:15]: I agree. I was about to ask you that question, because I think you could go at the lesser side — technology is an enabler for banks — and at the greater side, banks are essentially technology companies. And there’s probably a continuum there between those two endpoints.
Patrick McQuaid [43:32]: Yeah. And look, I used to work for a while there — I sort of led NAB’s customer experience analytics and customer experience insights function. And the really annoying thing — annoying as a technologist — is there’s a lot of things where customers evaluate you to the best digital offering out there, not the best banking digital offering, right? So there was a lot of, we have to be as good as a Netflix or a Google or something like that. And that’s sort of true. But when you delve into customer experience insights, people might like a new digital interface for a day to a week, and then expectations get levelled there.
Moments of truth, moments of impact, and moments where loyalty is built, happen almost always where there’s a human involved. It’s when a crisis happens and a human helps, or when somebody just gets unexpectedly great service from a human. They feel that way more than if their app surprised them one day. So that is why I think technology is a core part of banking, but banks aren’t technology companies in the way that we thought they might have been go back five, ten years ago.
James MacDonald [44:50]: I think customer service has seen a significant fall-off in really quality customer service these days. So when you do get it, you really do notice. It might be the cafe that remembers your coffee order, or delivers it to you by name. It might be if you go to a physical shopping centre and the customer service person comes to you to ask if you need help, as opposed to sitting waiting in a line. I think customer service is a real opportunity for differentiation at the moment.
Patrick McQuaid [45:21]: Yeah, I totally agree. And from what I understand about NGM — and I’m not as close to the customer as I was — that is still a core differentiator in the way we service the markets here. And we are helping our frontline people be better through the provision of technology. But that’s where the differentiator happens.
James MacDonald [45:45]: You’ve mentioned you’re not a product person, so this is going to be slightly left of centre for you. You mentioned Netflix before, and you look at customers and what their experiences are across other apps. From a banking environment — if I think about my app, as long as I can access my data, I can access my money, and it’s safe, that’s my core functionality, that’s what I need. Everything over and above that is great. But how do you look at other platforms — whether you look at competitors, or you look at a Netflix or somebody else — for inspiration on what your online experience looks like? Do you take inspiration from other verticals, other companies? How do you look at that from a product perspective?
Patrick McQuaid [46:27]: Yeah, look, I’ll speak to when I was at NAB, when I was part of their digital function and part of their marketing function. Absolutely — you look across industries, you’re getting ideas from everywhere. And by the way, you can’t avoid it, because consultants come and try and shove it down your throat anyway.
Where I think banking has changed — and this is along the lines of what I said before, we used to think banks were technology companies — same as, I feel a lot of bank digital apps used to say, we want to be the place where people stay, so let’s offer lots of services. And I remember being at a bank — not NAB, but another bank — where they were saying, let’s expose train schedules as a widget on our app. And I think that’s a really misguided approach, because people don’t want to spend time on banking. They’ll do it when they have to. You want to make it a great experience and frictionless. But it’s not where you just go when you’ve got a second — you go to YouTube, or somewhere like that.
So I think you’ve got to be very careful transferring experiences from apps that have a different perspective. And I feel there were some mistakes made 10 or so years ago in that vein, where the banks thought they had to make their apps really, really sticky. In reality, you want to make it frictionless, you want to make it really good, and you want to try and add value — absolutely. But it’s never going to be, oh, I want to spend half my day on my banking app.
James MacDonald [47:57]: Yeah. And I think, as you’ve mentioned a couple of times, it’s knowing who you are, it’s knowing who your customer is. Staying in your lane is the way I like to look at it. Know what you do, do that really well, and if you’re doing that, then your customer is going to have a great experience.
Patrick McQuaid [48:10]: Yep, absolutely. And it’s not to say things like advice and insights aren’t potential areas where you can add value. But we used to think about, well, look at how amazing Netflix’s recommendation engine is. I can tell you, most people don’t like Netflix’s recommendation engine — or, not most people, I shouldn’t say things like that — but I ignore Netflix’s recommendation engine now.
James MacDonald [48:34]: Yeah.
Patrick McQuaid [48:35]: Because so often it’s just filled with the stuff they want to push you onto. So I think there’s a lot to be learned in the way things have gone the last few years.
James MacDonald [48:44]: We’ve mentioned a lot about technology. A lot of the technology and what you can actually achieve doesn’t happen without the people in the door, the people in the building. Obviously in a regional you have different challenges. There is a changing nature of roles. There’s a lot of hype around the AI engineers and being able to access them. You’ve talked about hiring for skill and hiring for attitude, curiosity, problem solving. How do you see the nature of technology teams changing going forward? You’ve mentioned you did some project management before — project management’s changing in nature. Are you seeing just an evolution of roles? Do you think there’s anything black and white — we used to do things this way, it won’t be like that in the future? Or do you think this AI period in time is just another evolution, and we’ve always been through that — you know, old-school infrastructure engineers maybe learnt infrastructure-as-code and had to learn the code. Is that where we’re at? How do you see the changing nature of tech?
Patrick McQuaid [49:45]: Yeah, look, I hesitate to say this is like everything else, because AI itself I think is a more pervasive technology than we’ve possibly dealt with before, perhaps with the exception of the internet. But — go back two, three years ago, I’ve got young sons, I’ve got an older son who’s 21, I was worried about his future and work, and what was he going to do. But I think we’ve learned more and more that the whole human-in-the-loop concept — there’s still a lot of value, and there’s probably a thicker loop than we suspected around AI. Where it’s going to land, I’m not sure.
But you’re always going to have, I think, the need for experts, because you’ve got to find some way to ascertain and verify the AI is delivering the right thing. That comes from experts. You’re always going to have the need for storytelling — because again, AI is great at doing information and information presentation, but we’ve all seen AI slop, and everyone’s sick of AI-written presentations now. So there is definitely a human element that is needed there.
And I do think — stuff we’ve known about for years, and as a highly technical, data-driven person, one of the things I had to have beaten into me earlier in my career was, it’s not enough to be right. You’ve actually got to be engaging, you’ve got to find allies, you’ve got to build support, you’ve got to build relationships. That’s still a core skill.
So there are three areas that I don’t think change, and I think there’s a lot of value and a lot of space for humans to play in those spaces — even if we’ve got something able to find all the information at the click of a button, there’s still those three elements which are going to be critical.
James MacDonald [51:47]: I completely agree. I talk to my kids about those skill sets. I think communication, the ability to build relationships, those types of things, storytelling, become only more important as we go forward. And I think the more and more we deal with technology day in, day out, the more and more we also have this need and desire for human-to-human connection. And we’re seeing that come back as well, right?
From the changing nature of roles part — again, without having a definite answer — do you think longer term it becomes a positive thing, where we just do more of the things we enjoy doing, more of those higher-value tasks, and AI takes away all the repetitive, highly repetitive stuff we don’t want to do? Or do you think we get to a point where it’s taking out a bunch of white-collar roles, and it’s a negative for society?
Patrick McQuaid [52:33]: Yeah, gosh, that’s a bigger question than I think I’m really good at.
James MacDonald [52:36]: I don’t know if I should have even asked you that. I’m putting you under pressure.
Patrick McQuaid [52:39]: I mean, I’m not a Pollyanna about this. I think there’s going to be significant upheaval. And I still am worried about, what do my boys end up doing, and how do they find their place in the world?
Because — if I can tangent for a second — I’ve seen something like this before. Back when I was in consulting, early days, when I was actually still on the tools, the rise of the remote development centre happened. And not just offshoring, but before offshoring there was the remote development centre, where we had all of our junior coders and our analysts sitting in a room, not at the client site, because that saved a lot of money. And then we discovered you could also offshore it. And we were worried about, where are the roles in the future going to come from? What are they going to do? How do you grow people who are good at it? It’s probably turned out to be quite overblown. But we’ve been through that.
So I was very worried about that, because I prided myself on being a really good coder, a really good — what you’d call an engineer now — for a while there, and I was like, how are you going to grow more people like me? That makes me think it’s not all doom and gloom. There will be space for people doing higher-value work. But is there enough of that work? I don’t know. And I think that’s what scares me a little bit about what we’re looking at in the future. So yeah, that’s my opinion — I’m just uncertain. But I do know that you can’t deal with that uncertainty by ignoring it. You’ve just got to get in there and have a go, and see what you can do, and figure out what’s happening.
James MacDonald [54:34]: Just to stay down this teams route for a little bit — you mentioned the word remote there. How important have you found — there’s a big move back to sharing some time in the office. How have you found the team-building aspect? Growing a team, getting people’s storytelling on, this is a journey we’re on. How have you found that, remote-first versus in the office? You mentioned before, the start of your career, the need to bring people together, grow that next generation. Has your opinion changed over time? What’s your current experience look like?
Patrick McQuaid [55:05]: I don’t think my opinion’s changed over time, actually, because I’ve been in roles where I haven’t been with my team 100% of the time since 1998.
James MacDonald [55:17]: Right.
Patrick McQuaid [55:18]: I’ve had different roles where I’ve been flying around, and teams in different locations, so I can never be 100% with my teams. I don’t think 100% remote is great. I prioritise — I come to Newcastle every second week, and it’s a significant pain sometimes, but I do it because I think it’s important to show up and be around and to see people rather than be a face on the screen. And I think my career suffered during COVID, because I couldn’t build relationships as easily. It’s harder to do it over Zoom.
So my advice to anybody thinking about, what does this mean, do I take a 100% remote job — obviously, life stages, other things to factor in — but where possible, I think you should try and get in the office and build relationships. Because, like we said before, one of the core skills in the future that is unassailable by AI is the ability to build relationships and talk to people and communicate. That is enhanced by in-person. There’s no doubt about it.
James MacDonald [56:23]: I agree. And I think it goes one step further as well. If you have a look at a startup, or any company, the communication — understanding internal stakeholders, understanding your customer — that comes down to communication. If you can understand who your customer is, what their problems are, then you can actually build for that. So sales and marketing is communication. Internal — what product are we going to build, how do we build it well, how do we communicate this internally — communication. I think that skill only becomes more and more important, and as you said, I think it’s significantly easier to do in person than fully remote.
Patrick McQuaid [57:01]: It’s something you should practise too. And it’s something — again, I speak to my sons about, my oldest son — for a long time they led a predominantly online life. It’s really easy when you’re online to only hang out with people who agree with you, or if you hang out with people who don’t agree with you, to do the perfunctory, transactional elements. But getting into an office, working with someone who you don’t see eye to eye with on everything, and still forging something productive, is a real skill that has to be honed and trained, and it’s valuable. So I would encourage people to do it.
James MacDonald [57:37]: Especially with your LLMs at the moment, with how agreeable they are to you.
Patrick McQuaid [57:44]: Exactly. You’re never in the wrong, right? No, exactly. So you want to get into an office, you want to have that hot-flush feeling of disagreeing with somebody but still having to work with them, and having a good debate but still remaining respectful. Things that just don’t happen online, in my experience anyway.
James MacDonald [58:01]: You’ve worked at enterprise-level companies — NAB, Telstra — and NGM, still a mid-sized business. These companies have challenges where they can’t just roll out AI across the board. There are boards, there are general managers, CEOs talking about, we need to be doing something, we don’t want to miss the boat. How do you personally think about it, and what advice would you give other organisations, to approach an AI journey?
Patrick McQuaid [58:26]: Yeah, look, I think the first thing to maybe say is, there are a lot of people out there trying to instil a FOMO approach to this. If you’re on LinkedIn, they’ll be like, oh, you need to do this, you need to do that, you’re not going to get the full benefit of AI unless you do it now. And, like a lot of these things, there’s a kernel of truth to that. But there’s also a lot of hyperbole. Like — I’m not a believer in, there’s a theme going around at the moment that you’re not going to get the full benefit of AI unless you redesign all your processes around it. I don’t think that’s true. I think that might be an end state, where you can redesign and reimagine some processes with AI. But no one has 100% optimised processes, and AI provides a lot of ability to get productivity and efficiency in a different way without redesigning things.
And I’ll go back to the simple example I had before. Even doing that — stepping into these things in a light way, learning about them, still doing highly beneficial use cases, but learning about it without trying to break and toss up your whole system — is a great way to do it. And then along the way, you create the capacity, maybe, to think about bigger and better things.
So that would be one of my first pieces of advice: yes, do start, but don’t swallow the whole “I have to redesign my whole company to fit around AI”, because I don’t think that’s true. I think you can get a lot of value from doing it. But I do say, get going. And that’s a great thing, I think, for smaller enterprises as well — there’s a lot more power in your hands now than there maybe would have been previously. So get going, exploit it, have a look at it, but don’t think you have to throw everything up in the air to do it.
James MacDonald [1:00:17]: I think the other benefit for smaller organisations is that nimbleness, that ability to do something — as long as it’s not affecting the end customer significantly or negatively — learn from it, and if you get it wrong, adapt, iterate. And the speed at which we can adapt and iterate and build right now is quicker than ever before.
Patrick McQuaid [1:00:38]: Exactly. Look, I don’t think AI is different to a lot of other technologies in the past. You don’t immediately go to the guts of your company and toss it up and down and implement this new technology. You do it in lower-risk areas. You learn, you grow. So those are the key things I would emphasise to people. And look, I have operated most of my career in highly regulated industries, industries where we hold a lot of people’s data and a lot of personal information. So we have to be way more careful about how we do that.
James MacDonald [1:01:15]: On the AI journey, there are legal risks, financial risks around this, especially when you’re dealing with, or having any of your customer data accessible by, AI. There’s even more risk for a highly regulated environment like yourself. How do you advise others on doing something with AI — like you said, get an easy win or a quick win — but also looking at it from a risk framework, an architecture framework, a data privacy framework? How do you tie those two things together?
Patrick McQuaid [1:01:50]: Yeah, look, for a start, I’ll say there are some great resources out there now. There’s the National AI Centre, which is an initiative for the Australian government. Everybody in a small business should go and look at that, because it actually lays out some good frameworks and some good risks, which will help you. Because, what I said before — don’t fall for the FOMO, don’t fall for the hyperbole, but don’t ignore things either. So you take these frameworks and they’ll help you think through the risks. And then you have to engage with that and decide, do I have a legal risk here, or is it the same legal risk I’ve always had?
And that’s one other thing I have a big bugbear about, which is, some of the risks people are talking about, saying, because you use AI you have a risk of this — no, you’ve always had that risk. It’s just that the AI is the latest way that people sell to you that you’ve got that risk, and therefore you might need to pay their consulting services. So don’t ignore it, go in with your eyes open, use the resources.
But risk is — and I’ve done a data protection role at Telstra, and I did a risk role at Telstra, and most people who know me well laugh at that, because I’ve never been that sort of person. I’ve always been the forge-ahead, slightly-break-things — again, acknowledging I live in a regulated industry, so you can’t break a lot of things — but really push the boundaries and strive to innovate. But those risk roles gave me a new appreciation that I think is helping me a lot with AI right now. Because a lot of AI, I think, is about managing the risk. So in a smaller company or a medium company, find the areas where you’re not exposing or using customer data, find the areas where you have minimal legal risk, and deploy there to learn and get the benefits before you go into the higher-risk areas. But most of AI right now is about how do you manage and think about risk of using AI.
James MacDonald [1:04:13]: For a smaller organisation who probably don’t have a risk framework — probably, as far as they’ve ever gone is a privacy statement on their website, and maybe they got that templated from somewhere as well — how do organisations like that approach this? Because in a bigger organisation, highly regulated, you have to do it, you’re legally bound to do it anyway. Everyone’s got privacy laws they have to abide by. But most SMBs, I think, don’t spend a great deal of time actually looking at it, optimising for it. How do those businesses tackle that dual nature of wanting to do something and not being left behind, whilst also being more cautious in their approach?
Patrick McQuaid [1:05:03]: Well, I mean, I go back to what I said before. Something like the National AI Centre framework is a good way, because they do list the AI risks, the categories that you can think of. But ironically — use AI. Even in my space, every now and then I’ll go onto our model and say, here’s my AI risk framework, tell me a bit more about it, or tell me where it’s weak, or poke some holes in it.
James MacDonald [1:05:03]: Yeah, exactly.
Patrick McQuaid [1:05:03]: So it’s a thought partner as well. And anybody can do that, anybody can use AI. And the great thing is, it will find those things for you better than a Google search engine, and it’ll give you more context. So I would genuinely — if you put into your favourite LLM chat window, “I am thinking of doing this in my industry, what are the risks?”, it will spit you out a pretty good list of the risks that you can start to consider.
James MacDonald [1:06:03]: And if you give it a more detailed prompt, with “here’s what I’m thinking”, on the back of that, poke some holes again — I think that’s a good approach, using it as that thinking partner or the sounding board. Because in a lot of these cases you can be like, oh, go and talk to a lawyer before you do that. And for a startup or a smaller business, they’re like, these legal costs are things that can make and break a business. And maybe if it’s just a new feature they want to pull out for their business, those legal issues, sitting down with a lawyer, is probably not worth that trade-off, right?
Patrick McQuaid [1:06:33]: Yeah — although I will say, anytime you get into legal issues, don’t rely just on AI, because there are plenty of instances right now of people being burned by that. And this goes back to the point you said before about the whole AI sycophancy problem. These chat engines are mostly designed to make you feel good, and also designed to sound authoritative. So they will give you a point of view and opinion which will largely agree with what you’ve tried to get it to say in the first place. So use it as a sounding board — but anytime you stray into something to do with the law, always get advice, or go to the source to double-check.
James MacDonald [1:07:21]: I definitely will not be giving out any legal advice on this podcast, and I would hope it’s not taken as such.
Patrick McQuaid [1:07:28]: Exactly. But this goes back to what we said before — the real advantage for small businesses with AI now is, all of a sudden, you can have this really innovative, powerful capability for dollars of usage, right? So that’s where I would start if you’re thinking about it.
James MacDonald [1:07:47]: Nice, mate. So we’ll wrap up. I think one of the key takeaways that you’ve provided for me today for technology professionals is to really lean into those core human skills — problem solving, the right attitude, having that go-getter attitude, willing to take something on, admit when you don’t know the answer but “I’ll find a way”, curiosity. Those are the key skill sets that you would be advising people to lean into for the rest of their careers. Any other takeaways that you would share with business owners, technology professionals, in how they’re going to attack their career over the next 12 to 18 months?
Patrick McQuaid [1:08:25]: This advice is probably applicable to anybody. I was going to say if you’re not a software engineer, but I think anybody — learn to use the tools. A lot of AI usage now — and if I think about my own space, I’ve got some data scientists and some engineers, and that’s great. Where I am lacking, though, is I’m lacking an expert who might lead the Copilot charge, or Claude or something like that. So I think there’s no downside to getting into and using the tools, and thinking about, how do I become an expert in this tool? Because more and more — talk about the Pareto principle — I think 80% of the usage of AI in organisations will be based around tools that are produced by these companies, that you can go and learn now.
James MacDonald [1:09:16]: Completely agree. It was going to be the last question, but I completely agree on that. I want to dig into that a tiny bit, because I actually believe one of the key roles for the future is going to be what I call an AI orchestrator. And essentially, you don’t even need to be an engineer — you can use Claude to build the integrations between systems. Back in the day you needed to be an actual engineer to hit an API, and then you needed to use make.com or n8n, but now you can just use your Claude to build these integrations between different systems. And it is that orchestration role of — here’s the problem within a business, getting a tool off the shelf, which can be quite cheap, and just making sure we’ve got the right integrations between them to actually make it work and add value to a business. I think that role of an AI orchestrator is going to become more and more important going forward, and I think most companies will have some version of that within their org. It might sit within a current role, or a current subsection of a role. But how do you feel about that? Non-technical — and I think, you have a look at a role like a BA, a business analyst, their core role is to understand a problem and then map it. That skill set for someone like that could be invaluable.
Patrick McQuaid [1:10:28]: Yeah, it could be. You’ll see it advertised a little bit now — AI enablement is out there. And some of those roles are more enterprise change. But if you had an AI enablement team in your org — and I’m talking big orgs here — you could easily see these roles that are specifically like, there could be an AI orchestrator whose job it is to knit together things using AI. There could be a Copilot guru, or a Claude guru, who coaches people on how to do prompts better, or has workshops with people with a very specific question. And they’re not that technical, those roles. Those roles are about getting in there and having a go and playing with the tool.
James MacDonald [1:11:36]: Problem solving.
Patrick McQuaid [1:11:36]: Exactly. But problem solving with the tool you’ve got. So we have it now in NGM — I’m organising some workshops with people who are just good at Copilot prompting, with teams who have a very specific problem. You’re saying, well, actually what you need is somebody to teach you to prompt better.
James MacDonald [1:11:36]: It’s also the reason there’s been a significant investment in consultancies to actually enable companies to roll out AI. The partnerships with McKinsey and the likes — these AI consultancies are just growing, because they understand the desire for people to use the tools, but the need to actually roll them out properly and get the value out of them. That need doesn’t necessarily need to come from a consultancy, if you can build that internally.
Patrick McQuaid [1:12:03]: And they’re not highly technical roles. They’re just people who’ve got in there and done it. And that’s the key.
James MacDonald [1:12:07]: And I think that just ties in and wraps up what you’ve talked about today, which is that attitude part — attitude and willingness to go in and learn a tool — can be super valuable. And I think that will be a role of the future, or make people more valuable in their current role going forward. And I appreciate your time today.
Patrick McQuaid [1:12:22]: Thank you.
James MacDonald [1:12:22]: Thank you for the time. Always like to talk about AI, always like to talk about how organisations are actually using it in enterprise. So really excited to continue to follow the NGM journey.
Patrick McQuaid [1:12:33]: Thank you very much. Thanks for the talk.
James MacDonald [1:12:36]: My key takeaway from the conversation with Patrick was not even around AI in your enterprise. It’s around the human connection. I like to talk about building technology teams and technology careers on this podcast, and in my day-to-day job as Managing Director of NTP. What Patrick mentioned around those core personality traits being the success to a technology career — not only now, but into the future, as the tech roles might change and the tech skills might change — the eagerness, the attitude to be a go-getter, to try to problem-solve, fix something even if it breaks or you get it wrong, learn from it. Problem solving, communication, that human element — those personality traits which can be taught are other things that I’d be leaning into if I was a technology professional trying to build out my career.
It’s an interesting time with AI and how things change. But as much as some things change, some things remain the same. That was my key takeaway, and I think it’s really key advice for anyone exploring a career in technology.
Hope you enjoyed this episode. This was the Building Tech Teams podcast. We release new episodes every Tuesday. Really keen for feedback on guests you’d like to see, more topics you’d like to see in depth. If you’d love to provide any form of feedback to me, find me on LinkedIn, forward slash James MacDonald AU. You won’t miss me. This episode was produced by Day One.