The pressure on investment operating models has never been greater. Superannuation funds are managing trillions of dollars, bringing more assets in-house, and expanding their operations across the globe. How can firms build an operating model that is fit for the future?
In our latest AlphaChat podcast, AlphaCert CEO Phil Pietersen and Independent Consultant Stephen Huppert unpack these mega-trends:
Stephen Huppert: Welcome to our next edition of AlphaChat. I'm joined by someone you know very well, AlphaCert's CEO, Phil Pietersen. Hi Phil, and thanks for joining us.
Phil Pietersen: Hi, Stephen. Yeah, thanks very much for the opportunity. Great to see you.
Stephen Huppert: Look, Phil, we've both been in the industry a long time, and when you look back over the last decade,
what do you see as the biggest shifts in how superannuation funds and asset managers are operating?
Phil Pietersen: Yeah, we've seen some major changes over the last decade or so. Driven through a variety of different macro trends. It started off some many years ago, probably around 10 years ago with internalisation. In other words, large super funds wishing to move from a fully outsourced model of using fund managers to running money internally. And then along the way we've also seen two other mega trends around cloud and digital transformation. A trend we're seeing right now is
internationalisation. So those are probably the key three areas that we've seen and the areas that we've probably been most involved in from an AlphaCert perspective.
Stephen Huppert: Great. And we'll talk about all those three. I guess the biggest issue that confronts everybody when we look at Australia's superannuation pool, is it's now over 4 trillion and growing and I guess scale's probably one of the most significant factors that's changed operating models.
Phil Pietersen: Yeah, absolutely. We've certainly seen that. So some of the large super funds that we've been working with over the last decade or more, we've seen the funds under management grow enormously, sometimes fourfold
Stephen Huppert: Wow.
Phil Pietersen: over the last decade. And that presents a huge challenge on many different fronts. The target operating model,
the reporting, the compliance, but also the oversight from the regulator, and not least, you know, finding high return assets,
on a regular basis. So a lot of huge challenges for these organisations, and I've been very impressed with the performance that they've had and their overall performance over that time.
Stephen Huppert: Yeah, they've ingested an awful lot of money. So in terms of internalisation, I guess one of the key drivers is cost savings, but it creates some other operational issues. So what have you seen with funds that have internalised and the types of realities, risks they've had to deal with?
Phil Pietersen: Yeah, well obviously, probably the key thing is the operational infrastructure. And what I mean by that is a transformation of the
target operating model.
Stephen Huppert: Mm-hmm.
Phil Pietersen: When you naturally, when you're running a fully outsourced model with managers, which was very common in the early stages of the super journey. The reporting requirements and the amount of data that was required and your whole governance framework was quite different when you start running money internally. So we've seen some significant changes to operating models and to the amount of infrastructure and operations and the front office to support running money internally. And that varies from actual systems to run portfolios, trading systems, order management systems. But broader that through to risk, performance and attribution right across the board. And what we've seen fundamentally is that the data needs have exponentially grown over that period of time. In a fully outsourced model using fund managers, amount of data that you're required and that nature of that data is very different from running an IBOR and an ABOR on a daily basis.
Stephen Huppert: And you know, as part of that insourcing, I guess it's important to make sure you've got both the systems, the data, and the culture. And how do you see those three interplaying and which seems to be the hardest part when funds move in-house?
Phil Pietersen: So the data, the systems, and the culture, I think there's sort of a maturity curve.
Stephen Huppert: Mm-hmm.
Phil Pietersen: It starts off without having the data at hand, it's very hard to do anything, but that data is driven by systems. In an outsource model, obviously that data's all being sourced externally and being brought in from fund managers and from the custodian, but once the systems are put in place to enable portfolio management and the associated reporting around, risk performance attribution and compliance, the data needs then grow exponentially. In terms of the culture, yeah, it's a totally different mindset across the organisation. But probably more is the governance model that needs to be put in place to govern the data and govern the overall operation of the business.
And that can be a challenge because that can have a significant impact of culture because it can change the individual's day-to-day ways of working. And, sometimes humans are reluctant to change things that they've been doing very successfully for many years or many decades.
Stephen Huppert: That's for sure. And that term governance is one that I'm hearing more and more as the industry matures. And one of the things that we've had to change and somehow let go of a bit was the plethora of legacy spreadsheets all over the funds and the operations.
So, how does that shift from using all these spreadsheets to digital and cloud? It's much more than just a nice to have now, isn't it?
Phil Pietersen: Absolutely. I think, you know, there's always, been a place for spreadsheets. I think there always will be. Where the challenge arises and we've got a very popular download on our website, is the risks around spreadsheet management. I know Stephen, you were involved in creating that back in the day. The issue I see with spreadsheets is where they become mission critical because, the governance and management of those spreadsheets is much more difficult than if the business process has been systemised. So that's a key issue that we see. Where we see those large, complex spreadsheets that are effectively working as defacto databases that have a lot of historical data in them, or require a lot of data to do the calculations, that presents an enormous risk.
Stephen Huppert: Yeah.
Phil Pietersen: The reviews we've done over the years assessing spreadsheets oftentimes we find errors in those spreadsheets, which may have sat undetected for a period of time. The impact and the criticality is always challenging to say what they
Stephen Huppert: Yep.
Phil Pietersen: were downstream. But oftentimes there's errors that are sitting there. So that spreadsheet migration, particularly a) where they're used as a database, and b) they're mission critical is a key change to the operating system.
So those are the core investment operations spreadsheets. The models, the different testing models and calculation models that are used in the front office, you know, that's completely different. Modeling and testing different scenarios. That's completely different because it tends not to be absolutely mission critical for the operation of the system, particularly if you're looking through a governance lens from a CPS 230 or some of the regulatory compliance over in the UK and
in the US. But we have seen in one particular example, early in our life of this company, a client had 400 operational spreadsheets that they used every month. And within 18 months, systemising that through the use of modern technology, we were able to turn off 325 of those spreadsheets.
So within 18 months, that's a massive step forward. And from a risk and internal audit and external audit perspective, that is a huge step forward. So it was a very cool project to be involved in.
Stephen Huppert: Yeah.
And as you rightly point out, it's not just the time saved, the cost, but there's that risk. And when something goes wrong, and inevitably it will with a spreadsheet, the cost of remediation can be massive and the reputation.
Phil Pietersen: That's right. And then tracking these things down can be sometimes quite difficult. But you know, from a human perspective, it puts a lot of pressure on what we call single person dependency, where there will always be one person that is an expert in that spreadsheet, and the pressure's on where at month end or year end, where things are going wrong.
It puts pressure on them if they want to get away and have a holiday that weekend.
Stephen Huppert: Yes absolutely does. So with that migration to a more modern infrastructure, especially with cloud, digital, what are some of the things that funds have really got right and, what still seems to be tripping them up?
Phil Pietersen: Yeah, I think the appetite for data consumption just seems to grow exponentially. So we found that very interesting that the more data that's in place and the better quality of the data, the more hunger there is for data. A lot of is driven by regulation as well. Deeper, wider, and more in depth analysis across a lot of the data fields there. That's been a mega trend and a lot of the organisations, even after working on their data platforms for many years, are still looking to advance that. We do see from the regulator perspective more and more look through and underpinning data that's required. And in complex portfolios and complex organisations, that's very difficult to manage on a spreadsheet.
You need a sophisticated data model to drive that. So I think we've seen that come through. Obviously in the world of LLM'S and of AI, there's a hunger to be able to consume that data. But a trend that's just emerging is around data quality.
A lot of organisations will have data lakes, data warehouses, databases, where not every single field is necessarily to the same standard of quality because it may not be used on an operational, day-to-day basis.
The data might be being ingest say from a custodian, but may not be looked at, and if there are issues with that data, the AI models don't know that and will give you something that looks like a fact, whereas in fact it's based on data that might be spurious. I think data quality and data cleansing is gonna really be a major focus as we start to leverage some of these AI tools that are coming out.
I think that's being broadly accepted now that the, that DQ
is a major mega trend.
Stephen Huppert: And it really shifts data governance, data quality from being just a compliance exercise to something quite strategic then and forward looking.
Phil Pietersen: Yeah, absolutely. I think a lot of the drivers for data governance, say a few years ago was ensuring that external data that was being purchased was being done effectively and efficiently, and there wasn't duplicates in a large, complex organisation, the management and the governance of that data, but now it's shifting more towards a federated data model, where different parts of the organisation, different teams want to have control of their own data. It's a much more sophisticated data governance model. And the cost savings through data efficiency are being dwarfed now by the demands of those teams using machine learning and AI tools, which are really highlighting data quality issues that may be nascent in the underlying data repositories.
Stephen Huppert: And then the other trend that you flagged earlier with the funds getting bigger and bigger is the shifting of portfolios offshore. And for a lot of funds, internationalisation is a big issue with huge offices globally. Some of the funds, half the assets are being managed offshore.
So, from an operations and day-to-day data management perspective, what are some of the challenges that's bringing?
Phil Pietersen: Yeah, that's definitely a trend we're seeing in, the Australasian super industry, the internationalisation mega trend. The precursors for that is really having an infrastructure that is portable and globally accessible.
Stephen Huppert: Hmm.
Phil Pietersen: The trends around digital migration, cloud migration that we've seen occur over the last 10 years have really provided that foundation. When we started working in the space more than 15 years ago, there was still a lot of on-premise, a lot of databases that may have sat in the front office under somebody's desk. A lot of that has been eliminated. The move to global cloud infrastructure providers is probably complete for the most of the large organisations.
So that was really a precursor. Some of that also enabled a whole lot of changes that were able to be made around for in-house or customised product systems to be able to run continuous integration and continuous deployment pipelines, have the associated cybersecurity, and to be able to move quicker and faster and have more cybersecurity and more, business continuity and business resilience.
So that was very much a precursor that we've seen to the internationalisation component. Now we are seeing a lot of those super funds that we've dealt with, setting up offices, particularly in Europe and building large teams, large teams in Europe and the infrastructure and the support that we provide through our support services and also through our core AlphaCert platform has leveraged those digital and cloud migration mega trends to enable that support to be done very easily in different time zones and different geographies.
Stephen Huppert: And you mention the different time zones and the upside of that, funds talk a bit about follow the sun model, being able to have more round the clock, but that also puts a lot more requirements for rigor into the whole operating model.
Phil Pietersen: Yeah, it does. And you know, a number of the funds I've spoken to are looking at a Follow the Sun. Another term I've heard used loosely is follow the book as well.
Stephen Huppert: Okay.
Phil Pietersen: So the will be passed from region to region, depending on the completion. And any outstanding issues move to the next region for that team to pick up and continue working on a 24 hour cycle. So there's a lot of advantages to that, but that can put pressure on reporting lines, controls and organisational structure, and even to a certain extent, culture. Because the reason why a book might not be complete at the end of the day in one time zone means there's issues. And the next time zone has to pick up the previous team's issues.
So that can sometimes cause contention over a long period of time. So we've seen that, but it does enable, the follow the sun enables a 24 by seven opportunity to work through issues, make improvements, and really accelerate the rate of change that we are able to inject from a data and platform perspective.
Stephen Huppert: And that also brings with it challenges of multi-currency, cross border regulation and the different cultures in those different regions. It's all adds to the stress on the operating model.
Phil Pietersen: Yeah, definitely. And we've seen the time zone can be a challenge. But the key thing obviously is operating in different jurisdictions. And different regions and what the local, regulatory compliance regimes are. And for a large organisation like a Superfund, you know, that is pretty taxing, operating in these different regimes.
And the UK is subject pretty much to European regulation, and addition if they're doing work in that region. There's a lot of additional pressure, which means additional data. And that might be different from the APRA
that might be captured today. There might be additional or different sort of data that's required tomorrow in different jurisdictions.
So it makes life interesting, put it that way.
Stephen Huppert: It certainly does. You've touched already and it's very hard to have a conversation in the industry these days without it shifting to AI. So I guess the obvious question that everyone wants to know, is it hype or a really genuine game changer for investment management?
Phil Pietersen: Look, if you look into the future and imagine a future in 10 years time and looking at the agentic AI models that are self-learning, self-correcting, and are advancing on an exponential basis, it is quite incredible to imagine what is possible in that relatively short timeframe. I think there are lot of mega trends that we're seeing around AI and having traveled quite a lot in Europe, UK, and the US recently is some organisations are coming off a very low base of knowledge, maturity and access to the tools, whereas others are at a much higher level of utilisation, understanding, and leveraging of those tools.
So the first mission for us all is to try and close the gap, from the less knowledgeable to the most knowledgeable, from the have-nots and the haves closing that gap. Only when that gap is closed efficiently, will we really be able to accelerate the use of these tools there. I think the jury is still out in certain, in areas in terms of the utility today through use of those tools directly for efficiency gains. But the, nature and the propensity of these models to produce hallucinations is obviously a concern for the operations teams that are used to operating to five decimal
places or more of accuracy. So the use of AI tools, I think depends on where in the value chain you're looking at, further down the value chain where you're doing advanced analytics and you're looking for trend analysis and correlations that are not
obvious when you look at the data, those tools are extremely good for that. And if there's a human in the loop step to
Stephen Huppert: Yep.
Phil Pietersen: verify the information before it gets used. It's a natural part of the research and the chase of alpha. I can understand that and I can fully appreciate the value of that further up the value chain when it's looking at anomalies and actually processing the data, accurate to many different decimal places. That I would have a real concern with today in terms of relying absolutely on those systems. In 10 years time, who knows where we might end up, but I think it's only gonna exponentially advance. And humans are not always necessarily great at understanding exponentially. Talking to people about compound interest and getting them to understand that, we don't natively think exponentially. We tend to think yearly. So that's, I think the challenge is who knows quite what it's gonna be like in 10 years time, and it is pretty exciting, but it's also pretty scary about having some of these models get off the rails and need to be reigned back in.
So governance more than ever before is going to be key.
Stephen Huppert: I mean, that's that word again, governance. And I think for all sorts of dimensions, governance, whether it's looking at data, looking at your AI, is critical. Because it is, I like your description of the exponential growth. Because if we just look back over three years, and it really is only, three years since these larger language models have been everywhere, what chat GPT could do three years ago compared to what it can do now is just phenomenal.
And I think we do forget very quickly how quickly it is moving. So there are some real risks for funds who rush into AI without the right data foundations isn't there?
Phil Pietersen: Absolutely. And as I've said before, is that data quality is probably more important than ever. And it'll be exponentially more important as time goes by. The use of the technology is advancing so fast and we're actually running a series of demos at the moment,
Stephen Huppert: Hmm.
Phil Pietersen: leveraging the portfolio data that we have in the AlphaCert platform.
And we are teaming up in this case with Bloomberg to look at some of their regulatory data and using tools sitting on top of these platforms to do advanced correlations that are quite difficult to do if you were building spreadsheets, or building SQL statements and using traditional tools, whereas using technology like the MCP server, laying that on top of those data sources and plugging an LLM model into that to ask the queries, we're getting phenomenal correlations very quickly and very easy because the MCP technology doesn't need to be told what the database structure is and the offset database structure is certainly more complex now than it ever has been. It can self discover and self regulate the correlation. So combining portfolio data with regulatory data is a fantastic example of us showing the power of AI and it's to build proof of concepts. I'm talking about POCs at this stage, not production systems is unbelievably quick and unbelievably easy, and the results that are coming out are truly, mind blowing. So if we imagine what we're gonna be like in the next, five to 10 years, it truly is impressive.
Stephen Huppert: And thinking about that technology and what you're capable of doing. Again, I think we need to remind us all to have the human in the loop still. So it's a tool to help us, not a set and forget and AI goes and does it, and we sit back.
Phil Pietersen: Yeah, absolutely. And it certainly changes the nature of our roles.
Stephen Huppert: Hmm.
Phil Pietersen: The efficiency gains are incredible. Our team of building the systems using, AI tools every day, and the tools that we're using today probably didn't even exist two or perhaps three years ago. Everybody is using AI to build the code, write the code. But more importantly, check the code, validate the code,
do code reviews to different standards and different international guidelines that we put together, as well as looking for anomalies, potential errors, and building out use cases, test cases, and acceptance criteria. Fully automating that process end to end, it's absolutely incredible the gains that we've seen.
So from a pure programming perspective, we've seen phenomenal gains occur, but the programming part of the end-to-end life cycle from an idea, to putting into the hands of the user, the programming part is a relatively small part. So there's still opportunities elsewhere in that value chain to gain those efficiencies.
But keeping an eye on that and keeping the human in the loop is absolutely critical, and I hope that continues for a long time, before the agents take over the world.
Stephen Huppert: Yes. Yes, definitely. So putting you on the spot a little bit, if you're advising a superannuation fund, COO or CTO today, where would you advise them to focus their energy first? We've talked about internalisation, internationalisation, cloud transformation, AI. Any thoughts about where they should focus their energy first?
Phil Pietersen: Yeah, I think making sure those foundations are in place and the data foundations are in place is the starting point. Part of those foundations is to make sure that you have some flexibility and some agility
Stephen Huppert: Mm-hmm.
Phil Pietersen: in the data management, so that as waves of data are required, new waves from regulation or prosecuting new opportunities out there in the investment world.
Having that flexible model that can be relied upon is absolutely critical. So the actual technical foundations, the data itself and the data governance over the top of that. But, you know, in the world of internationalisation, a lot of organisations are looking to shift a lot of their presence up into to markets where they can be closer to the assets that they are working with, either from a public markets perspective or from a private markets perspective. So having an infrastructure and having a support model and an operating model that is time zone aware and can flex into other time zones is absolutely critical. And that does require some quite cautious planning, some careful enterprise architecture thinking, and a really sophisticated operating model and an operating platform such as what AlphaCert has been able to demonstrate.
Stephen Huppert: Yeah, that's really sound advice that, before you go rushing off to any of those transformations, make sure you got the foundation really solid and in place. That's really good advice. Finally if I gave you a magic wand and I asked you to fix one problem for the industry looking forward, is there one single problem you'd like to try and solve with that magic wand, Phil?
Phil Pietersen: I think it is around data, Stephen, I really do believe it's around data. Data really is the, platform for supporting everything that's happened into the past into today, and then looking at what we can do into the future. If you have a data platform and a data infrastructure that is adaptable, agile, can move fast, and provides confidence, it becomes a competitive advantage to be able to move quickly, to make decisions um, in a high confidence environment.
So that's really our mission, is to try and look at improving data quality, and the management, and the flexibility, and move it from some sort of dark art to an environment where it's well understood, well governed, is effective, and is delivering fantastic results for the teams right through, from the back, middle, and the front office.
So if I had my magic wand, it would be bang, it's all about the data.
Stephen Huppert: Yeah. And leading to confidence. It's not just about the means to an end, it's wanting to be confident. Phil, thanks very much for joining me. It's always a pleasure to have these conversations. If you'd like to read more about these topics, there's lots of really good insights on the AlphaCert website.
So thanks for joining us and we'll catch you next time.