E21: David Walker Group CTO Westpac Bank - How AI Will Transform Banking & Westpac's Vision for AI

WEBVTT
00:00:00.017 --> 00:00:03.557
Hey everyone, welcome to Power Time. Let's have a listen to what's on today's show.
00:00:04.937 --> 00:00:09.117
The concept that we might even step into a world one day where every employee
00:00:09.117 --> 00:00:10.417
has their own personal agent.
00:00:13.437 --> 00:00:17.917
They all have these roadmaps and that balances keeping things evergreen,
00:00:18.257 --> 00:00:22.257
keeping the legacy running well with introducing new capabilities.
00:00:22.677 --> 00:00:24.217
And so that trade-off, that balance.
00:00:26.057 --> 00:00:31.777
Um broadly speaking the vision for westpac is that we want to make uh artificial
00:00:31.777 --> 00:00:36.597
intelligence a real important plank of our deepening our relationship with our customers,
00:00:39.397 --> 00:00:44.197
well we start with a big question get that over and that's going to happen to
00:00:44.197 --> 00:00:50.497
you it's power time get ready let's go.
00:01:03.697 --> 00:01:07.337
Hi, everyone. Welcome to Power Time. I'm joined with a really special guest
00:01:07.337 --> 00:01:09.937
today, the CTO from Westpac, David Walker.
00:01:10.397 --> 00:01:12.997
And it's an absolute pleasure to have you on the podcast, David.
00:01:13.177 --> 00:01:17.477
Maybe you can give us a few minutes about yourself and your journey at Westpac,
00:01:17.577 --> 00:01:22.017
and then we can get into some of the details around Westpac and what you're
00:01:22.017 --> 00:01:23.277
doing around the AI space.
00:01:23.417 --> 00:01:26.077
Hey, Scott, Matt. Great to be here. Good to see you again as well.
00:01:26.637 --> 00:01:30.537
So, yes, Group CTO at Westpac have been here for six years.
00:01:31.177 --> 00:01:35.057
My journey in technology and banking is 38 years long.
00:01:35.297 --> 00:01:39.537
A little bit of experience there in terms of navigating different parts of my career.
00:01:39.697 --> 00:01:43.677
Started off as a software engineer, assembly language, mainframes.
00:01:43.957 --> 00:01:47.817
Awesome. Went into the data business, Ramon does a data consultancy,
00:01:47.977 --> 00:01:49.397
which you now call data science.
00:01:49.497 --> 00:01:53.197
It wasn't called data science back then that became trendy later and then sort
00:01:53.197 --> 00:01:58.497
of entered into my executive career which was all been large companies banks
00:01:58.497 --> 00:02:03.577
both here in Australia and up in Asia and Westpac's all been about transforming
00:02:03.577 --> 00:02:06.997
a bank a great bank in Australia into becoming an even greater bank.
00:02:07.493 --> 00:02:11.473
Yeah, Westpac's really been on the forefront of AI.
00:02:11.873 --> 00:02:16.513
And, you know, when you have a look at AI and what's going on in Westpac, what's your vision?
00:02:16.573 --> 00:02:20.493
And how do you think AI is going to really reshape, I guess,
00:02:20.653 --> 00:02:23.693
financial services over the next few years moving forward? Yeah,
00:02:23.833 --> 00:02:25.233
big question there, Scott.
00:02:26.853 --> 00:02:30.673
Well, we start with a big question. Why not? Get that over and done with.
00:02:30.813 --> 00:02:33.613
What's going to happen to the industry? Look, I think, you know,
00:02:33.933 --> 00:02:38.553
no doubt everyone, unless you've been living under a rock, has been focused
00:02:38.553 --> 00:02:41.793
on, you know, thinking about how they apply artificial intelligence.
00:02:42.293 --> 00:02:46.593
We've been on that journey for a number of decades at Westpac,
00:02:46.613 --> 00:02:51.973
but more recently, the newer forms of AI that started off with the Transformer
00:02:51.973 --> 00:02:57.093
breakthrough and which became large language models and so forth, about three years now.
00:02:57.093 --> 00:03:00.973
So we were even pre-ChatGPT, we were working with some of that technology.
00:03:01.253 --> 00:03:05.433
And we could see right from the very first day the power of what that was going to do.
00:03:05.673 --> 00:03:10.813
And so we worked with a couple organizations, one being Microsoft,
00:03:11.193 --> 00:03:14.153
and started working on how do we start applying this?
00:03:14.333 --> 00:03:18.873
What's the use cases? How do we think about that? And we collected use cases
00:03:18.873 --> 00:03:22.773
from all around the organization, build up a repository of hundreds of these
00:03:22.773 --> 00:03:27.313
and started thinking about, well, we can't build a hundred of these from scratch.
00:03:27.313 --> 00:03:30.013
That's a lot of scale going from zero to that.
00:03:30.173 --> 00:03:33.053
So we started thinking about what do the patterns look like?
00:03:33.133 --> 00:03:34.453
How do we get repeatable patterns?
00:03:34.653 --> 00:03:39.413
And that has been a huge step forward for us. So we've been building these patterns
00:03:39.413 --> 00:03:41.113
of technology out, which has been great.
00:03:41.873 --> 00:03:46.613
Broadly speaking, the vision for Westpac is that we want to make artificial
00:03:46.613 --> 00:03:51.373
intelligence a real important plank of our deepening our relationship with our customers.
00:03:51.613 --> 00:03:56.853
We see how it's going to help us become better at supporting our customers through
00:03:56.853 --> 00:04:00.493
different experiences, through freeing up our bankers' time,
00:04:00.653 --> 00:04:06.333
through productivity gains, providing better technology through software engineering at a faster rate.
00:04:06.873 --> 00:04:11.073
Everywhere we go, everywhere, look, AI is going to help give our customers a
00:04:11.073 --> 00:04:13.013
better outcome through direct or indirectly.
00:04:13.453 --> 00:04:19.133
And so that's where we are really focused. I think the finance sector in general,
00:04:19.453 --> 00:04:22.453
this technology is still at the infancy.
00:04:22.673 --> 00:04:28.693
We're going to see breakthroughs continue at the rate that we've seen over the last two, three years.
00:04:28.913 --> 00:04:32.613
And so it's very hard to predict how soon and how quick.
00:04:32.613 --> 00:04:37.833
But I can see a day coming where artificial intelligence is going to be creating
00:04:37.833 --> 00:04:41.433
this idea of what I'm calling a My Digital Twin,
00:04:41.433 --> 00:04:48.273
this concept of a customer representation that lives inside our company on behalf of the customer,
00:04:48.733 --> 00:04:51.353
acting on their behalf, something that we can communicate with,
00:04:51.513 --> 00:04:54.133
which will interact with the customer and vice versa.
00:04:54.193 --> 00:04:58.613
And we think this is a really powerful idea that brings the technology to life
00:04:58.613 --> 00:05:01.793
for our customers and looks after our customers in a very unique way.
00:05:02.614 --> 00:05:05.514
Yeah, I think when we look at the financial services industry,
00:05:05.654 --> 00:05:08.554
and Matt and I talk about a lot on the podcast, that industry.
00:05:09.154 --> 00:05:14.074
Mainly because I've got like 20 years experience in it myself and my whole team's across it.
00:05:14.254 --> 00:05:19.774
So for us, we see a lot of use cases, a lot of potential when it comes to agents.
00:05:19.814 --> 00:05:23.774
And, of course, we're recording this on the day that ChatGPT5 has just dropped
00:05:23.774 --> 00:05:28.374
this morning, which looks phenomenal already,
00:05:28.654 --> 00:05:32.354
straight out of the gate and supported by Microsoft on day zero,
00:05:32.354 --> 00:05:35.554
which is great and so you know i could see so
00:05:35.554 --> 00:05:38.354
many use cases i think moving forward as this kind of
00:05:38.354 --> 00:05:41.254
technology gets smarter and smarter over the next few
00:05:41.254 --> 00:05:44.694
years or actually i was talking about it last night with a few friends of ours
00:05:44.694 --> 00:05:48.974
and you know we have this concept of we might even step into a world one day
00:05:48.974 --> 00:05:53.754
where every employee has their own personal agent and and so your agent communicates
00:05:53.754 --> 00:05:58.694
with their agent and and handles a lot of that mundane kind of admin stuff that we Well,
00:05:58.834 --> 00:06:02.214
that's the fundamental idea behind the My Digital Twin piece.
00:06:02.494 --> 00:06:08.014
It's essentially, and I think it goes, you know, the technology to make that
00:06:08.014 --> 00:06:10.054
happen has to go beyond just agents.
00:06:10.314 --> 00:06:15.754
You know, it's the culmination of things like knowledge graphs and the ability
00:06:15.754 --> 00:06:21.154
to understand who you are in context, but in context over a long period of time.
00:06:21.154 --> 00:06:27.314
So you can see how technologies over the last decade have given us this position
00:06:27.314 --> 00:06:32.994
we're in where we can reimagine a whole host of new capabilities to represent, as you said.
00:06:33.394 --> 00:06:35.134
Your vision, your digital version
00:06:35.134 --> 00:06:39.314
of yourself operating inside in multiple companies acting on your behalf.
00:06:39.314 --> 00:06:43.034
And, you know, my digital twin will talk to your digital twin and off they go
00:06:43.034 --> 00:06:46.914
and they'll go sort out some stuff where that's everything from travel to I
00:06:46.914 --> 00:06:51.274
need to make a payment or I want to create a new bank account to operate in this way.
00:06:51.634 --> 00:06:55.554
And you can just see the power of that coming to life over the next few years.
00:06:56.114 --> 00:07:01.374
Yeah, 100%. Okay, so the question for me, David, is how do you prioritize between
00:07:01.374 --> 00:07:04.794
sort of modernizing that legacy stack that you alluded to before and then being
00:07:04.794 --> 00:07:07.994
at the forefront of all these new initiatives like you alluded to before with that?
00:07:08.154 --> 00:07:10.694
How do you strike that balance about with that?
00:07:11.394 --> 00:07:16.354
Yeah, look, that balance is always there, whether it's artificial intelligence
00:07:16.354 --> 00:07:18.414
or any new forms of modernization.
00:07:18.414 --> 00:07:22.154
You've got to keep the current technology stack alive and going well.
00:07:22.154 --> 00:07:26.974
And, you know, the first and foremost job of anyone in an organization like
00:07:26.974 --> 00:07:31.194
ours is to ensure that the lights are on well and truly, you know,
00:07:31.314 --> 00:07:33.914
it's safe and it's operating as expected.
00:07:34.294 --> 00:07:38.634
And that's tied up a lot in making sure the legacy, as you call it,
00:07:39.094 --> 00:07:41.874
is healthy, is running and evergreen.
00:07:41.874 --> 00:07:48.474
And so, we continue to put a huge amount of energy into that piece because you're
00:07:48.474 --> 00:07:53.934
not doing what you need to do for your customers if you're not operating your banks in that mode.
00:07:54.074 --> 00:08:00.114
So, that's almost job one to do is to ensure the safety and this reliability
00:08:00.114 --> 00:08:01.794
of your banking systems.
00:08:02.634 --> 00:08:08.774
Clearly, though, job two is, you know, striving to build out new capabilities to bring them on.
00:08:08.774 --> 00:08:11.974
And the flavor of the month right now and
00:08:11.974 --> 00:08:15.194
probably the flavor of the next few years will be how do you apply artificial
00:08:15.194 --> 00:08:18.934
intelligence in a sensible way and so it's always going to be about striking
00:08:18.934 --> 00:08:24.974
balance and any large organization like ours is is doing that all the time every
00:08:24.974 --> 00:08:30.694
day understanding when to move forward when not to move forward and and so we're very focused on,
00:08:31.286 --> 00:08:36.066
I'm building roadmaps right across our whole enterprise. We have 50-odd roadmaps.
00:08:36.286 --> 00:08:40.506
The business own those roadmaps. They're signed up to them. And that lays out
00:08:40.506 --> 00:08:46.006
a path for the mortgages business or the financial markets business or the HR
00:08:46.006 --> 00:08:48.766
systems in the back of the bank helping our people out.
00:08:49.086 --> 00:08:53.366
They all have these roadmaps. And that balances keeping things evergreen,
00:08:53.706 --> 00:08:57.446
keeping the legacy running well with introducing new capabilities.
00:08:57.446 --> 00:09:03.166
And so that trade-off, that balance mixed against cost, mixed against cost to
00:09:03.166 --> 00:09:06.066
serve, the risks that we've got in the systems,
00:09:06.666 --> 00:09:12.386
the competitive nature that we want in the marketplace, there's multiple levers
00:09:12.386 --> 00:09:15.646
that we're constantly looking at and adjusting to make those decisions.
00:09:17.006 --> 00:09:20.006
It's interesting because i know when whenever my team goes
00:09:20.006 --> 00:09:22.826
in to do a demo or to to any of
00:09:22.826 --> 00:09:25.706
the big four banks really the very next question is
00:09:25.706 --> 00:09:29.626
always by the architects that's great it looks good it looks good in greenfields
00:09:29.626 --> 00:09:32.746
but is it going to work in our environment where it's complex and we've got
00:09:32.746 --> 00:09:37.966
legacy systems and legacy security and all sorts of things so i think at the
00:09:37.966 --> 00:09:42.506
at the end of the day it's not just about being able to you know come up with
00:09:42.506 --> 00:09:44.006
a system that's going to work really well.
00:09:44.046 --> 00:09:46.906
It's got to work in a really complex environment most of the time,
00:09:47.026 --> 00:09:51.926
which, let's face it, in banking, there's a lot of complex environments that we have to navigate.
00:09:52.126 --> 00:09:53.866
Yeah, 100% Scott there.
00:09:55.086 --> 00:09:58.286
It's one of the things that I communicate to all our partners is,
00:09:58.526 --> 00:10:02.086
look, we'd love to hear your ideas, bring them to us because they do form a
00:10:02.086 --> 00:10:03.886
lot of our forward thinking.
00:10:04.166 --> 00:10:07.986
But when it comes time to make a decision of implementing it,
00:10:08.086 --> 00:10:09.266
it needs to be on the roadmap.
00:10:09.546 --> 00:10:13.846
You know, we can't just turn right and left. You know, we need to know when to do that.
00:10:13.966 --> 00:10:19.086
And don't waste your time or ours, you know, because it is, it's time consuming.
00:10:19.426 --> 00:10:22.966
So, it's about sequencing and making sure that alignment's there.
00:10:23.066 --> 00:10:25.686
And the best partners know how to do that really, really well with us.
00:10:25.966 --> 00:10:30.346
And I think that's a really key feature of working together in a collaborative way.
00:10:31.107 --> 00:10:35.907
Any tips for anyone that might be listening that's working in the FSI industry
00:10:35.907 --> 00:10:38.987
around, you know, how to partner better or work within those systems?
00:10:39.387 --> 00:10:44.307
Yeah, look, I think walk in our shoes and, you know, we try and walk in your
00:10:44.307 --> 00:10:46.747
shoes as well. It's always the key thing there.
00:10:46.987 --> 00:10:52.587
You know, I know when you're sitting sometimes in an office and you're thinking
00:10:52.587 --> 00:10:55.567
up, well, I've got a client out there and I've got this product.
00:10:55.707 --> 00:11:00.067
Let's go work out how best to get it in there. But the best ones always come
00:11:00.067 --> 00:11:03.547
and try and solve our problems, work out where it's on the roadmap,
00:11:04.147 --> 00:11:08.507
you know, when would that capability be taking an upgrade and things like that.
00:11:08.627 --> 00:11:12.087
It doesn't mean we're not, as I said, interested to hear new ideas and when
00:11:12.087 --> 00:11:14.667
new things come about because that's how we learn as well.
00:11:14.767 --> 00:11:20.287
But at the same time, actually understanding how to fit into our environment is really key.
00:11:20.447 --> 00:11:25.667
And that's not just when and what capability, but it's also being very aware
00:11:25.667 --> 00:11:27.427
that we're a regulated organisation.
00:11:28.247 --> 00:11:33.607
We have very high bar when it comes to security and managing the quality of
00:11:33.607 --> 00:11:37.087
our operational systems to make sure there's resilient, reliable.
00:11:37.347 --> 00:11:40.647
So come to us with products that are mature enough to fit into those environments.
00:11:41.327 --> 00:11:44.667
Those are things. So again, it all comes back to understanding us and understanding
00:11:44.667 --> 00:11:47.427
what it takes and not just imagining I've got this great product,
00:11:47.587 --> 00:11:52.227
surely everyone will want to buy it, more how will it fit and what's the right
00:11:52.227 --> 00:11:53.527
fit for us at the right time.
00:11:54.747 --> 00:11:59.647
I feel like we partner the best when we focus on business problems that you've
00:11:59.647 --> 00:12:04.707
got within your business unit and can we address that and then look at the tech kind of secondary.
00:12:05.147 --> 00:12:08.527
And as long as we're talking about enterprise platforms, we should be able to
00:12:08.527 --> 00:12:12.027
tackle the integration and the security and the compliance and things like that.
00:12:12.127 --> 00:12:16.607
But when we center around that business process or how we can really help your
00:12:16.607 --> 00:12:20.667
employees work better, I feel like that's where we're in the pocket with that
00:12:20.667 --> 00:12:21.987
kind of the most successful. Yeah.
00:12:22.707 --> 00:12:26.747
Well, technology, you know, banks are largely a technology company.
00:12:27.087 --> 00:12:31.807
You know, most of the products we sell and the services we provide to our customers are technology.
00:12:32.027 --> 00:12:36.327
So, you know, we do love working with companies that are really technical and
00:12:36.327 --> 00:12:38.207
capable of bringing technology to us.
00:12:38.307 --> 00:12:43.127
But actually, what it is, is about solving a problem for our customers or solving
00:12:43.127 --> 00:12:46.327
a problem for our business in a way that makes sense.
00:12:47.187 --> 00:12:50.747
The technology is really interesting and, you know, as technologists,
00:12:50.947 --> 00:12:54.887
we're fascinated about how to make that better and more efficient and safer and secure.
00:12:55.007 --> 00:12:59.507
But at the end of the day, we're solving business problems and people that understand
00:12:59.507 --> 00:13:03.847
that and appreciate that because that's what we're doing as technologists.
00:13:04.007 --> 00:13:05.087
We're solving business problems.
00:13:05.327 --> 00:13:08.607
They help us solve those business problems is the key. Yeah,
00:13:08.707 --> 00:13:09.447
yeah. Yeah, at the same point.
00:13:10.032 --> 00:13:13.092
Definitely. You mentioned before partnerships and tech vendors and the like,
00:13:13.232 --> 00:13:15.112
I guess, Westpac's a heavy Microsoft house.
00:13:15.132 --> 00:13:18.812
I guess, can you talk about the way that Westpac is leveraging sort of the Microsoft
00:13:18.812 --> 00:13:23.252
ecosystem, you know, the co-pilot products, Azure, all those types of things? Yeah, absolutely.
00:13:24.252 --> 00:13:28.192
Microsoft has been a wonderful partner over many, many years,
00:13:28.232 --> 00:13:33.472
and we certainly couldn't be where we are without the support of Microsoft,
00:13:33.692 --> 00:13:37.972
both from the products and the services that you provide across a whole range
00:13:37.972 --> 00:13:40.392
of things. we could probably spend a few hours covering them all.
00:13:40.592 --> 00:13:47.732
But, you know, just Azure is obviously where we, Azure is our biggest cloud partner at the moment.
00:13:47.932 --> 00:13:51.732
You know, we're deeply entrenched there. We have all our data safely and securely
00:13:51.732 --> 00:13:53.832
managed in the Azure environment.
00:13:53.992 --> 00:13:57.072
We're building out on Cloud Foundry there.
00:13:57.272 --> 00:14:01.372
So Azure Foundry, which has been a great work and we've created some experimentation
00:14:01.372 --> 00:14:03.012
zones just recently on there.
00:14:03.332 --> 00:14:08.052
We're adopting ML Studio at great lengths. We've had that up and operational now.
00:14:08.292 --> 00:14:13.172
Around our AI and data platforms for some time since it first came out.
00:14:13.612 --> 00:14:17.332
We've been a Power Platforms user, Copilot, Studio.
00:14:17.612 --> 00:14:21.832
And I think probably if I do a bit of joint bragging, we've done some really
00:14:21.832 --> 00:14:28.832
great work in partnership with Microsoft and your EY partner or our EY partner together.
00:14:28.952 --> 00:14:34.572
So the three of us have got together and rolled out Microsoft 365 Copilot across the organization.
00:14:35.392 --> 00:14:39.232
I've been looking at some of the results in terms of the adoption rates.
00:14:39.472 --> 00:14:44.372
And I think we're benchmarked pretty much right at the top of the global benchmarks
00:14:44.372 --> 00:14:48.612
in terms of adoption across all customers that Microsoft have there.
00:14:49.152 --> 00:14:52.932
I think, you know, which is outstanding work, which means, you know,
00:14:53.012 --> 00:14:56.372
at the end of the day, talking about the business problem, means our staff,
00:14:56.852 --> 00:15:01.012
thousands and thousands of our staff are now using that and getting value out of it.
00:15:01.612 --> 00:15:07.252
They're actively using it and the benchmark data is showing that consistently, which is great.
00:15:08.072 --> 00:15:12.032
It's interesting when you look at Copilot Studio, because originally it started
00:15:12.032 --> 00:15:15.332
off as Power Virtual Agent, and then we introduced, obviously,
00:15:15.832 --> 00:15:19.732
our language models and AI into that and rebranded to Copilot Studio at Microsoft.
00:15:20.172 --> 00:15:24.012
But, you know, it really wasn't a thing two years ago. And if it wasn't for
00:15:24.012 --> 00:15:29.572
the foundational work around governance, clients, most of our FSI customers
00:15:29.572 --> 00:15:30.912
wouldn't be able to move that fast.
00:15:30.912 --> 00:15:35.372
So it's great that it kind of just fitted into the ecosystem and allowed for
00:15:35.372 --> 00:15:39.452
you to be able to really just turn it on and know that you're going to be able
00:15:39.452 --> 00:15:41.472
to comply with all the resources.
00:15:42.306 --> 00:15:49.646
Yeah, and one of the other big call-outs is GitHub, Copilot.
00:15:49.786 --> 00:15:55.606
We were strong early adopters of that, and our software engineers have been loving using that.
00:15:55.746 --> 00:16:00.666
And we certainly as an organization have been loving it too because they're
00:16:00.666 --> 00:16:04.846
getting to write more code, better quality code, and they love working with it.
00:16:04.946 --> 00:16:09.866
So I think right across the whole range, we've had some really good successes
00:16:09.866 --> 00:16:12.126
with Microsoft over the years and will continue to.
00:16:12.986 --> 00:16:16.806
So pivoting a bit from the technology, David, let's talk a little bit about the culture.
00:16:17.086 --> 00:16:21.906
How do you kind of balance, you know, innovation within Westpac as well as encourage,
00:16:22.066 --> 00:16:26.386
you know, that innovation to thrive, but also deal with compliance and things like that?
00:16:26.486 --> 00:16:29.446
We've talked about, you know, it's a heavily regulated industry,
00:16:29.466 --> 00:16:33.926
as everyone knows, in FSI, but how do you really guide your teams around this?
00:16:34.646 --> 00:16:38.406
Yeah, you know, innovation is such a critical element.
00:16:39.256 --> 00:16:44.236
If you want to be a company that's going to transform, and quite frankly,
00:16:44.436 --> 00:16:45.856
every company has to transform.
00:16:46.256 --> 00:16:49.156
You have to continue to strive to be better.
00:16:49.536 --> 00:16:55.496
And if you haven't got innovation built in as a foundational DNA element within
00:16:55.496 --> 00:16:59.636
your company, then that's not going to allow you to sort of bring in that new
00:16:59.636 --> 00:17:02.196
capability all the time, those new ideas all the time.
00:17:02.296 --> 00:17:06.336
And so, I'm quite privileged to run a group called Westpac Growth Labs,
00:17:06.476 --> 00:17:08.456
which is our R&D unit for the bank.
00:17:08.989 --> 00:17:14.509
They're an amazing small group of people that have done some phenomenal work.
00:17:14.669 --> 00:17:19.029
And the key thing with that is they're not the innovation department.
00:17:19.029 --> 00:17:21.329
It's not the place you go to do innovation.
00:17:21.549 --> 00:17:24.769
What they are is they're the innovation enablers.
00:17:24.969 --> 00:17:28.089
They're the team that actually allow everyone else to innovate at scale.
00:17:28.249 --> 00:17:32.529
And so this is something that I learned in my work up in Singapore in DBS.
00:17:32.529 --> 00:17:38.049
You know, we changed our definition of innovation from being a place to go do
00:17:38.049 --> 00:17:42.109
the lab work, but to do the place that actually allows other people to learn
00:17:42.109 --> 00:17:45.909
and grow and become a whole company innovation unit.
00:17:46.089 --> 00:17:49.729
And so that's what we've done here at Westpac and Growth Labs has been able
00:17:49.729 --> 00:17:52.149
to help unlock people's mindset.
00:17:52.149 --> 00:17:58.049
One of the things that I would probably call out as an example is we're just
00:17:58.049 --> 00:18:04.849
right now running live this AI Shark Tank and we're doing that right across the whole company.
00:18:05.009 --> 00:18:11.469
So the 30,000 to 40,000 employees of Westpac now can work on the put ideas through
00:18:11.469 --> 00:18:15.589
for the AI Shark Tank and be competitive and that was all run by Growth Labs.
00:18:15.589 --> 00:18:20.609
And it started with how do you teach people right across the whole company, everyone,
00:18:20.909 --> 00:18:26.189
what is artificial intelligence, but also what is how to innovate,
00:18:26.389 --> 00:18:29.389
how to think innovative, how do you pitch your ideas, how to create ideas,
00:18:29.529 --> 00:18:33.269
how do you, you know, I can see problems. Well, how do you turn a problem into a pitch?
00:18:33.649 --> 00:18:37.229
So we've taught people how to do that. And so that's foundational,
00:18:37.529 --> 00:18:38.529
you know, creating innovation.
00:18:39.089 --> 00:18:43.209
And we're at the stage now where we've got thousands of ideas have come in.
00:18:43.209 --> 00:18:48.029
We've now whittled them down to the top 20 and they're going through technical
00:18:48.029 --> 00:18:51.709
labs right now to help develop their AI ideas into something.
00:18:51.929 --> 00:18:56.429
And then the final eight will go through to the sharks, which will include our
00:18:56.429 --> 00:19:03.289
CEO, who will pick a winner or some winners, which will then get developed into real live use cases.
00:19:03.449 --> 00:19:06.509
And so that's about creating innovation as a culture across the company.
00:19:07.008 --> 00:19:10.628
We love the ideas and we love the fact that it's AI, but it's actually about
00:19:10.628 --> 00:19:12.608
helping people unlock themselves.
00:19:12.968 --> 00:19:18.388
Now, you asked, how do you balance that with compliance and so forth around,
00:19:18.588 --> 00:19:20.348
you know, how do we manage that sort of things?
00:19:20.728 --> 00:19:25.968
Of course, we can't. So in the innovation world, we can't just have a free-for-all
00:19:25.968 --> 00:19:29.048
where you can just bring customer data down and use that as part of innovation.
00:19:29.048 --> 00:19:35.088
So, we very clearly have innovation environments and labs that lock down,
00:19:35.288 --> 00:19:42.008
but providing data and access to that data in a way that makes the data free and easy to use.
00:19:42.108 --> 00:19:46.688
So, we put a lot of work into creating labs that allow people to innovate using
00:19:46.688 --> 00:19:48.908
real data, but remain completely compliant.
00:19:48.908 --> 00:19:55.488
And I think we're pretty excited about how we can now scale those from ideas,
00:19:55.728 --> 00:20:02.908
innovations, right through to pilots, right through to live in production and scale.
00:20:02.908 --> 00:20:08.968
And so that whole pipeline from innovation through to live in production has
00:20:08.968 --> 00:20:11.768
to start with the compliance lens all the way along.
00:20:11.928 --> 00:20:15.448
And so it's got to have the security built in from day one. It's got to have
00:20:15.448 --> 00:20:19.208
the privacy considerations across the data from day one.
00:20:19.408 --> 00:20:24.308
It's got to have all the sort of normal other regulation things that we consider from day one.
00:20:24.448 --> 00:20:27.448
So even in an experiment environment, we're already starting to think about
00:20:27.448 --> 00:20:29.208
that to allow us to scale through.
00:20:30.048 --> 00:20:33.908
Yeah, and I really like that initiative. It's a great way for,
00:20:34.088 --> 00:20:37.248
I mean, we've spoken about this on the podcast before, you know,
00:20:37.268 --> 00:20:41.788
there might be just that one person who comes up with that one idea that makes
00:20:41.788 --> 00:20:43.288
a massive difference for the company.
00:20:43.428 --> 00:20:48.908
And it's about how can you give them the tools and the opportunity really to express that idea.
00:20:49.348 --> 00:20:52.328
So the Shark Tank sounds like a fantastic way to do that.
00:20:52.830 --> 00:20:56.690
Yeah, completely agree. Completely agree. I guess that role of innovation and
00:20:56.690 --> 00:20:59.290
driving and building a lot of these AI initiatives, I guess,
00:20:59.370 --> 00:21:02.750
how are you foreseeing the role of a software engineer in the world of AI?
00:21:02.910 --> 00:21:06.130
You mentioned before GitHub, Copilot and the like, like what's kind of your
00:21:06.130 --> 00:21:09.050
opinion, your direction that you're kind of foreseeing that your technical team
00:21:09.050 --> 00:21:10.250
is going to undertake moving forward?
00:21:10.250 --> 00:21:16.710
Yeah, look, software is going to probably one of the very early areas that gets
00:21:16.710 --> 00:21:19.090
the most amount of change going forward.
00:21:19.330 --> 00:21:26.110
If I go back a little bit in time, Matt, we were the very early adopters in
00:21:26.110 --> 00:21:33.370
the GitHub space and for Coparton, we ran, the Growth Labs ran an experiment for us.
00:21:33.370 --> 00:21:36.970
And as you know, you know, unlike, say, other areas of the bank,
00:21:36.990 --> 00:21:39.830
like operations, where you're doing repeating tasks, it's the same task,
00:21:39.910 --> 00:21:42.830
and you can sort of measure different pathways and see what's more optimum.
00:21:43.530 --> 00:21:46.650
You're not software engineering, you never typically write the same piece of
00:21:46.650 --> 00:21:48.250
code twice, right? Once it's written, it's written.
00:21:48.350 --> 00:21:52.210
So it's very hard to get that benchmark of how you're going faster or not.
00:21:52.210 --> 00:21:57.070
So actually in Growth Labs, what we did was we pulled 40 engineers off the line
00:21:57.070 --> 00:22:01.990
for a week and broke them into four teams and gave them, you know,
00:22:02.270 --> 00:22:08.730
gave three of those teams tools like Copilot and then one team just had to code by hand.
00:22:08.890 --> 00:22:13.470
And so we were able to measure a whole host of things during that week and they
00:22:13.470 --> 00:22:16.270
had to write exactly the same code, essentially, that just to get the same problem
00:22:16.270 --> 00:22:18.530
solved or a series of problems solved.
00:22:18.530 --> 00:22:22.410
And so, in that week, we were able to measure like-for-like comparisons,
00:22:22.410 --> 00:22:27.050
everything from how do they feel about using this, what's the adoption,
00:22:27.270 --> 00:22:30.730
and how do they feel about having use of this sort of technology right the way
00:22:30.730 --> 00:22:35.990
through to real genuine how fast they could do it, the productivity gains that they could get from it.
00:22:36.330 --> 00:22:40.350
And we were amazed at the results that we saw.
00:22:40.450 --> 00:22:44.450
And so, this was a real true go-back-to-university-style experiment.
00:22:45.470 --> 00:22:49.170
And from that, then we said, look, it's obvious. We need to take this.
00:22:49.310 --> 00:22:50.830
The gains were just so dramatic.
00:22:51.330 --> 00:22:56.830
The training, the education it gave the junior developers, the speed,
00:22:56.970 --> 00:22:59.590
the toilet took away from the more experienced engineers.
00:22:59.790 --> 00:23:06.210
And so we've been up and running and measuring that now through an external productivity tool.
00:23:06.370 --> 00:23:10.770
So we measure billable coding effort. And this billable coding effort gives
00:23:10.770 --> 00:23:15.390
us a number of hours of valuable code that's written a day by each engineer.
00:23:15.390 --> 00:23:19.850
And we can see those engineers that are pre-GitHub CodePilot,
00:23:19.870 --> 00:23:22.510
and then once they move and transition onto it, post.
00:23:22.710 --> 00:23:27.650
And so we're seeing anything up to 40% gains, which is 30, 40,
00:23:27.750 --> 00:23:31.430
50 minutes of extra code written a day of high value code.
00:23:32.305 --> 00:23:36.525
Now, that's just the start of things, right? That's pre-Ageos.
00:23:36.665 --> 00:23:41.445
So now we've got agents working inside our company. So we've got agents that
00:23:41.445 --> 00:23:45.485
helping migrate data from one place to another place.
00:23:45.665 --> 00:23:52.925
And so one example of that was we had a squad of four engineers that essentially
00:23:52.925 --> 00:23:54.685
were doing these repeat tasks.
00:23:54.845 --> 00:23:59.425
And they were taking about six weeks end to end as a cycle of work to take a
00:23:59.425 --> 00:24:01.985
data source and put it into the new location.
00:24:02.745 --> 00:24:08.285
We now have that same squad down to two people, two humans, along with four
00:24:08.285 --> 00:24:15.005
agents, software agents, and these agents now are doing a lot of that sort of scanning the code,
00:24:15.425 --> 00:24:19.565
scanning the source, writing the Python code, testing the Python code,
00:24:19.725 --> 00:24:22.465
evaluating that, and then productionizing it.
00:24:22.465 --> 00:24:28.565
And we're now seeing that six weeks now happen in less than four days.
00:24:28.785 --> 00:24:34.565
So, and then that's now, you know, we've been doing that for several months. And so, we can see that.
00:24:34.665 --> 00:24:37.665
And of course, that's just the starter as well, right?
00:24:38.145 --> 00:24:41.665
You know, we're now looking at the whole software delivery lifecycle,
00:24:41.925 --> 00:24:49.965
the SDLC, from writing requirements through to cycling in production and reimagining how to build software,
00:24:50.365 --> 00:24:54.285
you know, with giant coding coming in and, you know, ideations happening at
00:24:54.285 --> 00:24:57.685
a whole new different pace, which means you don't need a software engineer to
00:24:57.685 --> 00:25:00.565
ideate things, you know, that unlocks a whole host of things.
00:25:00.645 --> 00:25:04.925
So it's probably going to be, I think it's number two on my list,
00:25:05.045 --> 00:25:09.245
you know, if knowledge, global knowledge is number one, we've solved that through
00:25:09.245 --> 00:25:14.045
the likes of the technologies that have been with chat GPT-led and all the others that have come.
00:25:14.505 --> 00:25:18.705
Software is the second one that's going to be the biggest impact and we're pretty
00:25:18.705 --> 00:25:21.725
excited about what that means. We're going to be able to do a lot more with the engineers.
00:25:21.905 --> 00:25:25.365
We've got higher order functions, higher activity. So we're pretty excited.
00:25:26.145 --> 00:25:29.265
We're pretty excited about it as well. And I think a few years ago,
00:25:29.425 --> 00:25:32.265
probably about, I don't know, six or seven years ago, I remember going into
00:25:32.265 --> 00:25:34.385
Westpac and we were talking about the Power Platform.
00:25:34.965 --> 00:25:37.965
And we had this concept of Microsoft called citizen development,
00:25:38.005 --> 00:25:40.605
and nearly everyone that was in FSI was.
00:25:40.957 --> 00:25:43.557
Told me off for saying it, right? We're not going to have someone in the contact
00:25:43.557 --> 00:25:45.617
center who's building apps, right? It's just not going to happen.
00:25:45.917 --> 00:25:47.637
It's like, yeah, but we've got all this governance.
00:25:48.057 --> 00:25:52.357
Yeah, yeah, yeah. And just the message just didn't land. And I think here we
00:25:52.357 --> 00:25:56.397
are fast forwarding six years, you know, into the future from that point.
00:25:56.717 --> 00:25:59.877
And now all of a sudden it's like, oh, hang on a minute. We've now,
00:26:00.057 --> 00:26:02.397
we've really sorted out all the governance and controls.
00:26:02.817 --> 00:26:06.197
We've now got this vibe coding and this ability because natural language is
00:26:06.197 --> 00:26:10.337
going to be the predominant programming language moving forward over the next few years.
00:26:10.337 --> 00:26:13.397
We've actually got now the ability to unlock that and and
00:26:13.397 --> 00:26:16.417
microsoft released a product called planner about 12
00:26:16.417 --> 00:26:19.977
months ago it just went ga a few weeks ago and it's
00:26:19.977 --> 00:26:23.817
phenomenal when you when you bring all this together and the ability to be able
00:26:23.817 --> 00:26:29.297
to you know just prompt to build an app build a data model give it a couple
00:26:29.297 --> 00:26:33.557
of images and it'll build a ui i think we're really onto something here even
00:26:33.557 --> 00:26:37.897
if it's just inside of a you know a large organization like westpac even if
00:26:37.897 --> 00:26:39.537
it's just there to prototype apps,
00:26:39.797 --> 00:26:43.757
it's still a massive acceleration on innovation, just that product in itself.
00:26:44.037 --> 00:26:47.657
So I think, David, it's going to be a really interesting world over the next
00:26:47.657 --> 00:26:55.017
sort of 12 months, two years, as the AI models get more mature and the tooling gets more mature.
00:26:55.497 --> 00:26:59.377
Pretty much anyone can be a developer inside the bank at some point in time
00:26:59.377 --> 00:27:02.717
if they've got access to the right tools and they've got the right idea and
00:27:02.717 --> 00:27:04.677
the right business process that they want to go and solve.
00:27:04.877 --> 00:27:07.837
So I'm looking forward to see what happens over the next few years.
00:27:07.837 --> 00:27:10.717
Have you resisted going back and telling those folks, told you so?
00:27:13.497 --> 00:27:14.457
That's next statement.
00:27:16.917 --> 00:27:21.597
Write them on the podcast and then catch them, you know. Yeah, yeah, yeah.
00:27:22.297 --> 00:27:24.957
Because Westpac were one of the first adopters of the Power Platform.
00:27:25.157 --> 00:27:27.557
Yes, they were. Early in the day, so we'll kind of point it.
00:27:27.697 --> 00:27:30.737
Yeah, it was one of those things. Thank you for that. First of all,
00:27:30.877 --> 00:27:32.417
we used to foresee the vision.
00:27:32.577 --> 00:27:36.277
But you mentioned before, those numbers are staggering. six weeks into four
00:27:36.277 --> 00:27:38.717
days and working with agents and stuff like that.
00:27:39.217 --> 00:27:43.437
I'm still blown away at some of the ROI. The business case really stacks up
00:27:43.437 --> 00:27:45.837
for some of that stuff. So it's just an exciting time.
00:27:46.397 --> 00:27:52.157
Yeah, look, and we get asked what's going to happen to those engineers.
00:27:53.377 --> 00:27:58.717
Look, I'm on the camp that says clearly there's not enough software engineers
00:27:58.717 --> 00:28:03.437
in the world because today there's still a level of discipline around building proper systems.
00:28:03.437 --> 00:28:06.977
And I think their roles will change and I think their roles will continue to
00:28:06.977 --> 00:28:08.757
change more dramatically.
00:28:09.077 --> 00:28:14.857
But, you know, the idea of building a system still requires a mindset.
00:28:14.917 --> 00:28:19.177
And I think, you know, software engineers are really perfectly aimed at doing that.
00:28:19.737 --> 00:28:23.237
I think as long as the engineers are prepared to adapt and adopt it,
00:28:23.297 --> 00:28:26.057
I think the engineers that do are going to be very valuable,
00:28:26.297 --> 00:28:29.317
that are AI literate, as opposed to those that aren't.
00:28:29.317 --> 00:28:32.397
So that's my encouragement to all our engineers is get in there,
00:28:32.757 --> 00:28:35.837
understand it, embrace it because you've got to be far more valuable and,
00:28:36.357 --> 00:28:39.077
Right now, you're going to be able to deliver a lot more quality for us,
00:28:39.117 --> 00:28:42.517
and you're going to be able to help people that aren't really able to string
00:28:42.517 --> 00:28:45.637
a whole system together. You're going to be able to help them get to where we need to.
00:28:45.777 --> 00:28:49.597
So, I think there's still a really important role for software engineers in
00:28:49.597 --> 00:28:52.077
the world ahead, but their role will change for sure.
00:28:52.877 --> 00:28:55.397
And, of course, it doesn't matter what enterprise you go into.
00:28:55.837 --> 00:29:00.797
The backlog on apps traditionally has been tremendous, maybe 12 and 14 months worth of backlog.
00:29:00.797 --> 00:29:07.837
So I think it's going to be a more combined role with software engineers and
00:29:07.837 --> 00:29:09.237
the business together kind of,
00:29:09.397 --> 00:29:15.117
I don't know, if it's solving like Vibe kind of working together to solve the
00:29:15.117 --> 00:29:19.217
same business problem, but having tools to be able to be clear on how to get
00:29:19.217 --> 00:29:21.717
to an outcome faster. Yeah, I think you're right.
00:29:21.837 --> 00:29:24.777
And I think, you know, the world,
00:29:24.917 --> 00:29:29.277
this is a bit crystal ball-y and prediction-like, so it may or may not happen,
00:29:29.277 --> 00:29:34.837
but I can imagine that that list of what we would all traditionally call commodity
00:29:34.837 --> 00:29:38.437
code or commodity products start to shrink down.
00:29:38.437 --> 00:29:43.597
So there'll be a narrow list of things that we truly look at as commodity and we just get that in.
00:29:43.737 --> 00:29:46.697
And, you know, things like Office 365, we're not going to build that,
00:29:46.797 --> 00:29:49.377
right? It's always going to be something you want to bring in and that.
00:29:49.497 --> 00:29:54.697
But there'll be a fringe products today that may be considered you buy today,
00:29:54.697 --> 00:29:58.637
but actually we might want to code that ourselves or build that ourselves for
00:29:58.637 --> 00:30:02.377
us because we want to build it our way for us, for our use cases,
00:30:02.657 --> 00:30:03.417
and we might want to do that.
00:30:03.557 --> 00:30:06.557
We would have never done that before, but now we might have that ability.
00:30:06.557 --> 00:30:11.837
So, I can see it's building more and probably moving more into those product
00:30:11.837 --> 00:30:18.057
areas that may have been a little bit more semi-commodity and start creating them.
00:30:18.197 --> 00:30:22.157
So, then you build a more bespoke environment, unique for your customers.
00:30:22.457 --> 00:30:26.037
Unique for your employees, exactly what you want to differentiate on.
00:30:26.117 --> 00:30:31.017
And as I came back to the beginning, we want to create those deeper customer relations.
00:30:31.017 --> 00:30:34.577
Well, to do that, you really want to target exactly what you want.
00:30:34.577 --> 00:30:38.577
And so, to do that more and have more power to do that, to allow our engineers
00:30:38.577 --> 00:30:41.637
and our people across the bank to do that, that's pretty exciting.
00:30:42.100 --> 00:30:45.820
Yeah. And I think, you know, traditionally we've looked at a whole bunch of
00:30:45.820 --> 00:30:49.520
use cases in FSI and a lot of them might have been access databases or Excel
00:30:49.520 --> 00:30:54.600
spreadsheets, kind of turned into, you know, an application that was used by a lot of people.
00:30:54.820 --> 00:30:58.340
And so we want to turn that into an app, but we just can't get the business case.
00:30:58.560 --> 00:31:00.920
And then we would go in and go, well, let's look at load code.
00:31:01.060 --> 00:31:03.960
Maybe we can get a business case with Power Apps or something like that.
00:31:04.140 --> 00:31:08.020
And maybe we can't get a business case for that. But those use cases now,
00:31:08.140 --> 00:31:10.360
some of them are looking likely to be an agent.
00:31:10.900 --> 00:31:13.780
And an agent you know the cost to build and
00:31:13.780 --> 00:31:17.120
everything like this you know could be a lot less yeah so i
00:31:17.120 --> 00:31:19.860
can see a lot of use cases just going down down that
00:31:19.860 --> 00:31:22.880
path of you know an agentic kind
00:31:22.880 --> 00:31:27.620
of application maybe it's a combination of both but we will be able to solve
00:31:27.620 --> 00:31:30.200
those problems you just probably couldn't solve before you just couldn't get
00:31:30.200 --> 00:31:34.960
them yeah in any organization and you know you have a long tail of things that
00:31:34.960 --> 00:31:40.220
don't quite today make the grade for being important enough to develop and they
00:31:40.220 --> 00:31:41.720
sit in this funny, you know,
00:31:42.220 --> 00:31:46.340
environment that's in between, you know, full-fledged, oh, we can fund it as
00:31:46.340 --> 00:31:50.460
an application and we'll put dedicated people to it versus it's still important,
00:31:50.500 --> 00:31:55.620
but, you know, it's, and so I think that means that the, our tail will be smaller,
00:31:55.840 --> 00:31:58.960
you know, we'll be able to stretch further down, we'll either be able to vibe
00:31:58.960 --> 00:32:04.340
code and then, you know, possibly take that vibe code and productionize that even and allow,
00:32:04.600 --> 00:32:07.180
you know, much more agent based
00:32:07.180 --> 00:32:11.140
development to build out some of those capabilities that, as you said,
00:32:12.120 --> 00:32:16.540
traditionally were probably user developed rather than technology developed.
00:32:17.780 --> 00:32:21.260
David, changing it up a little bit, have you got any advice for any aspiring
00:32:21.260 --> 00:32:25.140
technical leaders that might be listening into the podcast and thinking that
00:32:25.140 --> 00:32:29.020
they'd like to work in the industry and really make an impact?
00:32:29.786 --> 00:32:34.186
Yeah, look, I would say if you want to move into this space,
00:32:34.206 --> 00:32:37.266
I think the first thing is always be inquisitive.
00:32:37.486 --> 00:32:39.186
You know, never stop learning.
00:32:39.486 --> 00:32:42.506
Listen to some great podcasts. Do you know any that you'd recommend?
00:32:43.426 --> 00:32:44.706
I reckon we know one.
00:32:45.666 --> 00:32:49.486
That was a leading question there. But number one, you have to be learning.
00:32:50.306 --> 00:32:54.066
Love the plug. Always learning, good period. Question everything.
00:32:54.706 --> 00:32:57.986
Yeah, don't rely on the past, you know, continue to go that way.
00:32:57.986 --> 00:33:04.206
I think my second thing I'd say is, you know, always surround yourself by the smartest people.
00:33:04.706 --> 00:33:08.626
Whether that's being a leader or even joining the smartest teams in the bank,
00:33:08.786 --> 00:33:10.406
you'll learn from smart people.
00:33:10.606 --> 00:33:13.586
Be humble to realize that you're not the smartest person. In fact,
00:33:13.726 --> 00:33:17.126
it was good not to be the smartest person and do that.
00:33:17.386 --> 00:33:21.266
And I think the third thing, if I was going to offer you another one,
00:33:21.386 --> 00:33:25.146
would be make sure you get good at leading without power.
00:33:25.146 --> 00:33:28.346
You know, there's a lot of people that like to, you know, in the current day,
00:33:28.706 --> 00:33:33.146
a lot of people want to create a power and create that sort of empire.
00:33:33.266 --> 00:33:37.866
But I think the real power comes from influencing and communicating and being
00:33:37.866 --> 00:33:39.226
really, really good at that.
00:33:39.426 --> 00:33:43.086
And being able to get on a podcast and explain your story is one good start.
00:33:43.846 --> 00:33:47.946
Yeah, I think so. I agree with that. I think it was Jeff Bezos said that if
00:33:47.946 --> 00:33:50.626
you're the smartest person in the room, find another room.
00:33:51.326 --> 00:33:55.926
Yeah. I love that advice. I think it was. Someone will probably message me and
00:33:55.926 --> 00:33:57.686
go with someone else, but I'm pretty sure it was him.
00:33:57.846 --> 00:34:01.226
And, yeah, great advice. I'm sure you could find Bezos or Jobs or any of those
00:34:01.226 --> 00:34:02.626
guys would have a great idea. Oh, it would be one of them.
00:34:04.826 --> 00:34:08.366
We've heard that from a few people. I hope it wasn't Bill Gates. That's true.
00:34:09.386 --> 00:34:14.066
You'll have an email in the inbox. Your access is cut. Yeah, yeah, yeah, yeah, yeah.
00:34:16.486 --> 00:34:19.686
So what trends do you think? We've only got a few more minutes,
00:34:19.786 --> 00:34:22.426
David, so we're pretty interested in, you know, What trends do you think in
00:34:22.426 --> 00:34:25.046
the tech industry are really underestimated right now?
00:34:25.206 --> 00:34:28.906
Is there anything you could think of that we should be looking at?
00:34:28.906 --> 00:34:32.026
Look, it's one of those interesting things.
00:34:32.346 --> 00:34:38.406
Clearly, I'll just say, given we've spoken about AI pretty much the whole time today.
00:34:39.490 --> 00:34:43.250
I think there's a lot of people out there still saying artificial intelligence
00:34:43.250 --> 00:34:48.510
is overhyped and there's a risk around that because technology,
00:34:48.750 --> 00:34:51.690
all technology that's like this, that's new is always overhyped.
00:34:52.310 --> 00:34:57.190
My question is, you know, picking the ones that are going to be real.
00:34:57.290 --> 00:35:00.730
And I just say to everyone, there's no doubt that this is real.
00:35:00.890 --> 00:35:04.110
So don't underestimate that this is coming because I'm sure there'll be some
00:35:04.110 --> 00:35:08.150
people still sitting back thinking it's overhyped and why should I pay attention
00:35:08.150 --> 00:35:12.910
to it? So, I'd say this is definitely a trend that absolutely needs to be followed and pursued.
00:35:13.350 --> 00:35:20.150
Some of the others, though, I think we're at this unusual point where AI is
00:35:20.150 --> 00:35:24.830
going to unlock a lot of other capabilities that we just don't even know today, right?
00:35:24.830 --> 00:35:34.610
So, as ChatGPT5 and other capabilities come out that are closer to really super
00:35:34.610 --> 00:35:38.250
AI engines that can solve other problems,
00:35:38.490 --> 00:35:43.050
and so we're going to see a whole host of other things just appear because these
00:35:43.050 --> 00:35:47.690
technologies are inventing other ways of solving these really big problems.
00:35:47.690 --> 00:35:53.730
So I think, you know, so that's more unconventional left field things that are unlocked from AI.
00:35:53.990 --> 00:35:59.250
And so keep a close eye on these things that are maybe going to accelerate some breakthroughs.
00:36:00.270 --> 00:36:03.350
But if I give you something a bit more, one last one that maybe is a little
00:36:03.350 --> 00:36:07.290
bit more tangible that probably not enough people are talking about yet.
00:36:07.970 --> 00:36:11.950
We're all in the robotic space, particularly humanoid robots.
00:36:12.190 --> 00:36:17.490
I think we're going to probably see the acceleration of them.
00:36:17.490 --> 00:36:22.250
And, you know, there's enough, you know, NVIDIA and Elon Musk and others are
00:36:22.250 --> 00:36:23.930
out there talking a little bit about that.
00:36:24.050 --> 00:36:27.030
But it all seems a little bit science fiction right now. And it's probably not
00:36:27.030 --> 00:36:29.990
talked enough about in corporate world what that means.
00:36:30.270 --> 00:36:34.090
You know, a billion humanoid robots on the planet, you know,
00:36:34.230 --> 00:36:37.450
within the next 10 number of years, you know, two, three, four, five.
00:36:37.730 --> 00:36:40.930
What would that mean? How does that mean for every organization?
00:36:41.030 --> 00:36:42.250
What does that look like?
00:36:43.047 --> 00:36:48.787
And I think that's been those sorts of breakthroughs as we plant a brain that's
00:36:48.787 --> 00:36:52.847
driven by artificial intelligence inside these humanoid robots.
00:36:53.027 --> 00:36:58.327
That could create a whole host of new unlocks as well that we should all be thinking about.
00:36:58.407 --> 00:37:02.227
What does that mean across lots of different industries and sectors, I'd say?
00:37:03.287 --> 00:37:08.227
It's interesting because you could say the robots could come as the maturity
00:37:08.227 --> 00:37:11.527
of an agent or a personal assistant or something like that.
00:37:11.527 --> 00:37:14.607
But we were coming up with this use case last night around banking,
00:37:14.607 --> 00:37:19.167
which I thought was really interesting, is if we got to a point where we had
00:37:19.167 --> 00:37:21.767
ASI agents assigned to us,
00:37:22.227 --> 00:37:25.187
I could pretty much point at it all the big four banks and go,
00:37:25.287 --> 00:37:27.447
you know, can you renegotiate my mortgage every three months,
00:37:27.527 --> 00:37:28.547
whatever the best rate is?
00:37:28.727 --> 00:37:32.527
And it could connect with the Westpac agent and they could negotiate on my behalf,
00:37:33.287 --> 00:37:35.027
kind of like a broker-to-broker relationship.
00:37:35.127 --> 00:37:38.007
That would be really interesting too for consumers.
00:37:38.187 --> 00:37:40.647
Well, it's interesting you mentioned that. If it came to fruition.
00:37:40.647 --> 00:37:42.027
Very interesting you mentioned that.
00:37:42.727 --> 00:37:46.507
One of our growth labs experiments right now is looking at that sort of,
00:37:46.607 --> 00:37:51.727
you know, futuristic use case and story about what if we did get agents talking
00:37:51.727 --> 00:37:56.827
to us rather than humans and how do we interact with them?
00:37:56.907 --> 00:38:01.267
So we've actually got in the growth labs world right now, spinning up an experiment
00:38:01.267 --> 00:38:07.187
around exactly that scenario so that we understand if and when it comes, what we would do.
00:38:08.087 --> 00:38:13.007
And I guess it's probably not too far away. I mean, the new ChatGPT agents looks
00:38:13.007 --> 00:38:15.947
really good. The co-polliture agents is fantastic.
00:38:16.307 --> 00:38:19.267
The ability to be able to say, hey, can you just go and do this?
00:38:19.367 --> 00:38:24.827
And it connects to, you know, the Westpac agent and solves that problem or maybe
00:38:24.827 --> 00:38:27.407
there's an issue and it just goes and resolves it for you.
00:38:28.707 --> 00:38:34.467
That represents a fantastic customer service, I would say, because it's working
00:38:34.467 --> 00:38:37.747
this on your behalf while you're at work or doing what you're doing.
00:38:37.927 --> 00:38:40.767
100%. And that would be really interesting. And I think, you know,
00:38:40.867 --> 00:38:46.027
some of the other things that we as technologists and thought leaders in this
00:38:46.027 --> 00:38:49.287
space need to think about is what's going to happen to the internet,
00:38:49.287 --> 00:38:52.767
what's going to happen to social media, what's going to happen to the mobile phone,
00:38:52.947 --> 00:38:58.867
you know, all these things that are part of our every day today may morph into
00:38:58.867 --> 00:39:00.807
something very different in the future.
00:39:00.967 --> 00:39:05.407
You know, we've already seen on the internet traffic change dramatically in
00:39:05.407 --> 00:39:06.907
the last six months with...
00:39:07.686 --> 00:39:11.746
You know, the advent of generative AI, locking people into those ecosystems,
00:39:12.006 --> 00:39:16.166
solving those problems, which means traffic's not going to websites as much
00:39:16.166 --> 00:39:17.586
as they used to and things like that.
00:39:18.246 --> 00:39:22.326
You know, bots generating content on the internet now is,
00:39:22.466 --> 00:39:26.346
you know, last year it was over half the content that was generated on the internet
00:39:26.346 --> 00:39:30.166
was through bots and 30 odd percent of those were bad bots, you know,
00:39:30.226 --> 00:39:33.266
creating not particularly positive content for the world.
00:39:33.766 --> 00:39:37.066
And so when you start looking at those sorts of stats and thinking,
00:39:37.226 --> 00:39:39.806
well, you know, what does that mean for the internet?
00:39:39.986 --> 00:39:42.666
What was the thing that we're very familiar with? How will that look?
00:39:42.766 --> 00:39:46.526
And therefore, what's as a business, as a company that uses an internet as a
00:39:46.526 --> 00:39:49.046
part of the interface and a protocol for interacting,
00:39:49.566 --> 00:39:53.406
if that's not something that people are going to use as in the same way anymore,
00:39:53.846 --> 00:39:55.106
how do we prepare for that?
00:39:55.226 --> 00:39:59.346
And as I said, same with social media sites or same with mobile phone.
00:39:59.466 --> 00:40:02.866
And there's a whole long list of things that are probably going to change as a result of AI.
00:40:03.766 --> 00:40:07.006
And we're all going to have to get ready for that. We're going to have to have
00:40:07.006 --> 00:40:10.266
a position on that and we're going to have to start to think how do we get ready for that.
00:40:10.386 --> 00:40:14.066
Just to your example, we're going to have to get ready for an agent coming to
00:40:14.066 --> 00:40:18.826
talk to us and ask us to do things on someone's behalf. How do we validate that
00:40:18.826 --> 00:40:21.866
that's a person behind that, that it's really got, you know?
00:40:22.446 --> 00:40:25.406
How are you going to authorize that? What's the entitlement right?
00:40:25.506 --> 00:40:29.166
So there's a whole list of things to solve to make that use case work successfully
00:40:29.166 --> 00:40:35.126
and safely and securely at scale. and we all need to be getting ready for those
00:40:35.126 --> 00:40:37.006
and other things that are potentially coming.
00:40:37.826 --> 00:40:41.886
Yeah, and would you trust an agent to buy something on your behalf?
00:40:42.066 --> 00:40:45.706
That's something I don't think anyone's really thought about at the moment.
00:40:46.026 --> 00:40:49.486
Yeah, yeah, yeah. You'd almost need like an agent bank account in that case
00:40:49.486 --> 00:40:53.666
that just had limited funds just in case it went crazy and spent all your money.
00:40:53.806 --> 00:40:55.426
You wouldn't want to give it access to your bank account. Well,
00:40:55.546 --> 00:41:00.426
we do know some of these can hallucinate and can get confused on goals and all
00:41:00.426 --> 00:41:01.446
sorts of things like that.
00:41:01.446 --> 00:41:07.106
So, really the guardrails, and I know that's a very overused term and used to
00:41:07.106 --> 00:41:10.666
hide a whole host of things that you have to actually technically do,
00:41:10.786 --> 00:41:16.306
but we do need to ensure that the safety elements are built into these systems for the customers,
00:41:16.526 --> 00:41:21.066
for the employee, for the world, for society to operate successfully.
00:41:21.726 --> 00:41:25.346
Because you don't want too many early failures because confidence will drop, right?
00:41:25.506 --> 00:41:29.246
So we've got to make sure we hit the ground running when these things are ready
00:41:29.246 --> 00:41:32.186
and do them at the right scale but safely.
00:41:32.926 --> 00:41:35.986
I feel like there'll always be a human in the middle with that kind of thing,
00:41:36.086 --> 00:41:42.746
approving something or some kind of governance, two-factor or something when that comes to fruition.
00:41:43.326 --> 00:41:46.646
Yeah, I think we will need to keep those things very much that way.
00:41:48.186 --> 00:41:52.606
So last question, David, before we will let you go. It's been fantastic,
00:41:52.966 --> 00:41:55.666
mind you, getting your insights and having a chat.
00:41:55.846 --> 00:41:58.086
And thanks so much for coming on to the podcast.
00:41:58.506 --> 00:42:05.306
If anyone's aspiring to be a CTO in the banking industry, any top tips other
00:42:05.306 --> 00:42:06.506
than the tips you've already given us?
00:42:07.191 --> 00:42:14.331
Top tips of becoming a CTO. Well, those three that I gave before definitely are a CTO tips as well.
00:42:14.491 --> 00:42:19.791
But if I was going to add one more, I think be ready for change.
00:42:20.371 --> 00:42:25.111
You're going to be probably at the front and center of how explaining this sort
00:42:25.111 --> 00:42:28.351
of massive amount of change that's coming to the organization.
00:42:28.471 --> 00:42:33.851
So being able to understand it, appreciate it, and then help people see it.
00:42:33.851 --> 00:42:38.991
Because some of these changes are hard for people that are non-technology people
00:42:38.991 --> 00:42:41.791
to see what it means for them and how to apply it.
00:42:41.851 --> 00:42:48.871
And so a good CTO has to be great at being able to take those in and then communicate them back out.
00:42:48.991 --> 00:42:52.571
And so that would be the thing I'd leave with you. Yeah, awesome.
00:42:54.331 --> 00:42:57.511
Fantastic. Well, thanks, David. You're always welcome to come back on the podcast,
00:42:58.971 --> 00:43:01.151
whenever you're ready to give us another update.
00:43:01.151 --> 00:43:06.111
It was fantastic having you on. and, yeah, looking forward to seeing some of
00:43:06.111 --> 00:43:11.631
these insights that come to fruition, if they do, and supporting any initiatives
00:43:11.631 --> 00:43:13.011
that we're back to the camera before.
00:43:13.391 --> 00:43:16.771
Thanks, Mark. Thanks, Matt. It's been great to be here. Yeah, awesome.
00:43:23.031 --> 00:43:29.071
Thanks for tuning in to PowerTime, your source for practical insights across
00:43:29.071 --> 00:43:33.411
Microsoft AI, low-code and digital transformation.
00:43:33.771 --> 00:43:40.071
Just a reminder, this podcast does not represent the official views of Microsoft.
00:43:40.811 --> 00:43:44.131
The opinions shared are those of our guests and hosts.
00:43:44.531 --> 00:43:49.911
If you found value in today's episode, share it on LinkedIn,
00:43:50.591 --> 00:43:56.031
drop us a rating on Apple or Spotify and subscribe on YouTube for more.
00:43:56.311 --> 00:43:58.191
You can also check out our full
00:43:58.191 --> 00:44:04.331
archive and behind-the-scenes updates at powertime.au and on Substack.
00:44:05.291 --> 00:44:10.011
And finally, if you or someone you know is struggling with mental health,
00:44:10.191 --> 00:44:11.711
please don't hesitate to reach out.
00:44:12.011 --> 00:44:18.571
In Australia, you can contact Lifeline on 13 11 14 or Kids Helpline on 1800 14.
00:44:18.480 --> 00:44:38.163
Music.