Fireside Chat with Matt Comyn and Sri Ambati at H2O World Sydney 2022
Read the Full Transcript
Sri Ambati:
Today, it's an incredible pleasure and privilege to be joined by the CEO of Australia's largest bank, Commonwealth Bank of Australia, Matt Comyn. Matt has more than 20 years of experience in banking across businesses, in institutional retail and wealth management in the bank ever since he joined in 1999. As CEO, Matt is focused on delivering global best digital experiences, and we are very proud to have partnered with CBA to help some of that journey. Super excited to have Matt on stage and talk about how they are bringing Australia and Commonwealth Bank and our community and customers to the forefront of AI with great pleasure. I invite Matt to the stage
Matt Comyn:
That was Sri. How are you? Thank you. Very good to see you.
Sri Ambati:
Pleasure.
Matt Comyn:
Thank you. Morning folks.
Sri Ambati:
This must be your venue.
Matt Comyn:
Yeah, I guess so.
Sri Ambati:
CBA has been investing a lot in digital and technology over the years, and it's super exciting to see all the investments in AI. What's the bank's strategic maybe vision with AI? We'd love to know the general direction of the importance of AI for Commonwealth Bank.
Visions Of AI For Commonwealth Bank
Matt Comyn:
Yeah, look, I mean, it's clearly a huge priority. I think industries fundamentally are being reshaped and digital and data played an incredibly important role. I guess from us both strategically, as you probably know, Sri, we talk certainly internally about building tomorrow's bank today for our customers. There's enormous power and benefits that can be delivered to customers. I mean, we also should say fundamentally, it starts with high quality data that's well protected, and there's great controls around that. Our customer's privacy is absolutely at the forefront of that. But as we have really built up and you'll hear, some of the team that we're very proud of we've built up some really good capability, which has enabled us to really rethink a number of the experiences that we deliver for our customers.
I'm happy to go into some of my more favored use cases. But I think it's partly also bringing both to the more than 10 million customers that we're very fortunate to serve. A lot of really personalized and relevant actions and conversations with our customers as well as building world class capability internally. I think challenging the way people think about some of the traditional problems and some of the things that we've worked on together with H2O has shown us just the enormous benefit from that. I think it probably goes back many years when we started building something. We've spoken about quite a bit externally, which our customer engagement engine, which has continued to grow now where I think we've got about 450 machine learning models.
We're using that to basically in real time consider and decide what's the next best action or conversation across more than 16 million customers with obviously a couple of hundred billion data points. There's a lot of compute power that goes into that. But also a lot of the work that we've done and the team have led have been trying to also democratize some of those skills and capability. That's one of the really important roles that you and the team at as well as the, some serious IP and, and horsepower also bringing products and that. People who don't necessarily have the same level of data science can actually access. We've got more than a thousand people now using your H2O tool set.
Sri Ambati:
We are incredibly fortunate for the partnership, but also the culture of data and quantitative thinking that pervades almost the senior most echelons of this bank and this institution. What makes AI first or AI some of the signature aspects of being here, first organization or data driven organization that others in my community and customers can aspire to?
Signature Aspects Of AI At Commonwealth Bank
Matt Comyn:
Yeah. Well, I mean, obviously it has to be really grounded in what are the benefits that we can deliver for customers. And like I said, some of the use cases that we've talked about certainly internally and externally around things like the customer engagement engine around the benefits finder, which fundamentally we put together. Obviously the improved analytical techniques helped to build matches for customers about benefits that they're currently missing out on. That was a piece of work that started many years ago. I think we've now had more than 2.1 million of our customers have taken out a benefit that they probably previously wouldn't have known what was available to them. We've used it in lots of topical areas. Everything from emergency situations in bushfires, in floods to both give advanced notification to customers to give them realtime information.
More recently we've thought about what probably are more popular at the moment is the fuel finder, which obviously does a lot of analysis in terms of spend patterns and also what's available basically serves up as a next best conversation in the app. Where's the cheapest place to go and get fuel and there's lots of those sorts of things. There's a whole range of personalizations that we're serving up that we want to build into our shopping experiences. We can see ways to add a lot of value for our more than a million business customers in terms of trying to match our customer, our retail customer base with the best , and most personalized deals, almost appropriate for them. Yeah. The list of use cases goes on.
A couple of the other ones that we've worked on, maybe I can talk about them a little bit more detail, but we've used AI around what we call it document AI, which actually we were working on a particular remediation project, but we've actually now worked out, or we think we've worked out, we've probably got the most reliable and accurate document scanning algorithm that we can see publicly. Obviously that has huge potential for the way we serve customers to make things know your customer processes more streamlined and seamless, and of course more accurate. Clearly the potential and one of the key things that the team does is obviously deliver. We've got a center of excellence. We've got a large team and some very gifted individuals and leaders, but it's also about fundamentally retraining and helping other parts of the organization access that.
The quality and improvement and making sure we've got high quality data, as I said, that is well protected, fundamentally, that just delivers so many benefits for a financial institution. There's a lot of things that we rely on that data, obviously to help serve our customers, but also fundamentally to manage risk, both financial and non-financial. I think we spend probably as an organization, I'm sure as many others do, a lot of time thinking about how do we build that capability. Then of course, working with world class partners that we've been prepared to invest in, work together in. It's a source of pride for the team to be able to work on exciting projects that have real impact certainly at a national scale and to work with globally renowned experts. It is at the moment, a tight labor market, particularly in these sorts of skills. We want to be known as an organization where you can come and work with large data sets. You can make a real impact, deliver great things for customers, work as part of a great team on leading edge tools and with some of our partners.
Sri Ambati:
Awesome. I think it's a testament to the amount of talent CBA has amassed over the last decade, I would imagine. our partnership has been in the making five, six years but we've seen nothing but excellence on the data side. I'm super excited with the partnership you're seeing with your team. In terms of culture of allowing experimentation and increasing the velocity and the learning rate of the organization. A lot of our customers, other customers in the world are also grappling with some of the same challenges of AI transformation. How has CBA brought that culture of agility across the organization.
Commonwealth Bank's Agility In Technological Advancements
Matt Comyn:
Yeah, you touched on an important point, I think, Sri. Certainly in my view over the last five or seven years as we've evolved there's a confluence of capabilities that you need to be able to really scale. You obviously need some really deep technical expertise but then you also need leaders who can translate that simply and practically. As I think about some of the work, when we were building out the customer engagement engine, we solved some of the technical problems. I wouldn't say early on. It certainly took some time to get the design right. Actually, one of the more challenging aspects of it is having the right leaders who can then engage and interface with the rest of the organization.
They can also basically be trusted as the center of the analytics inside the organization. People have a traditional approach generally, which as you'd know is often in a consumer business that sends a lot of messages and communication to customers and hope that some of them are worthwhile and accurate. Actually, when you have the right capability and leadership, then it becomes really important to be able to trust the customer engagement engine. Obviously, of course, to be able to track it, to be able to measure it, and monitor it. Maybe we will come to some other really important aspects of it as well in terms of ethical AI and explainability of individual models and the whole range of different things that we need to keep a close eye on and work with our regulators on.
It's such an important topic. I think getting the fundamental capabilities in place is hugely important. Then, trying to build and scale that up, and then also try to challenge across some of the more traditional use cases. I know one that maybe the team will talk about later looking at a very traditional area of financial institutions, which is managing fraud. You actually run a more traditional model against an AI model. The fraud detection rates are dramatically, I think about 25% better. Clearly there's applicability. There's a whole range of models that we use for credit scoring. You don't always get the same level of uplift, but I do think that scaling out and building the confidence inside the organization, and then as I said earlier as well, democratizing.
You've got a large number of leaders that even if they haven't classically been trained in some of the statistical and analytical techniques, they can actually get to understand and work with tools that make it easier to access. I know there was a case that we were running a couple of my direct reports actually competing and trying to work out for, I can't remember the exact, but it was basically which offer would your customer best respond to and a whole bunch of variables. Then being able to select which one of those variables and then run it in real time. As you would imagine, some of those variables are actually quite counterintuitive. It's those sorts of things that at scale you can try to see and unlock the potential that's there.
Sri Ambati:
Yeah. This is the head of marketing and the leader of operations.
Matt Comyn:
That's right. That's exactly the one I was thinking. I'm sorry.
Sri Ambati:
I think that's one of those things we need to repeat across the rest of the world. I think getting this C level of the topmost executives hands-on playing with data, playing with models, or understanding the vulnerable days offer model and its applicability goes a long way in transforming the organization. I think that's something that we've seen really uniquely here in CBA. So super excited then.
Matt Comyn:
Yeah. I think clearly intuition and experience still play a really important role, but there's a lot of things that objectively can be measured and modeled. The data and the facts speak for themselves. If you look at, I think some of the companies globally that have been really successful over the last decade or more, they tend to have a pretty good culture in my view around being very data driven and having that institutionalized inside the organization. I think that's certainly a good add to a number of the very good capabilities and some of the things that, as you know, as you said, the largest bank in Australia, some of the things that we more traditionally are very much grounded in that very strong customer service ethos and technology more broadly. We've definitely tried to scale up our capabilities in this area
Sri Ambati:
During COVID, we worked very closely with the Commons to make sure services were on point and helped a lot of the people, communities here. Almost every time I've seen you talk about customers, there was a community that came right almost before or right after, in two words, we feel very privileged for the community we built in Australia without ever coming to our open source. How do we take the success from Commonwealth Bank and the AI efforts here to the rest of the nation?
Taking Commonwealth Bank's Success To The Rest Of The Nation
Matt Comyn:
Yeah. I think clearly this is not something that many institutions, both from the corporate sector as well as governments, are undoubtedly around the world. Maybe, I'll speak from recent experience. In September, I was down in Canberra and actually speaking at a conference there with a senior member of government administration. I'd say particularly COVID served two major benefits. One it created very or much closer linkages between corporations, not just the banks, but a number of companies and government. It really demonstrated the power and quality of high velocity, basically real time or near real time data, given the considerable uncertainty. I think one of the questions that people were asking us who holds more data and how's it better used?
Of course, we were having a bit of a debate about whether that's the aggregate of corporate or the government. Actually, if you look at some of the ways that both inside the government and just the service delivery, and undoubtedly there's still enormous potential, but, I know that they're very focused on that. I've seen that up close at a federal level where we are fortunate to do quite a bit of work with both the federal government. We're also the primary or transaction banker for a number of the states as well. Yeah. I think there's the same mindset and commitment to trying to deliver better services and better outcomes. Of course, even one of the things that I mentioned at the beginning of this conversation, sorry, around benefits finder, that that's something that we've worked on with a number of different government departments.
I think some people, at least early on, were questioning whether that would be welcome because fundamentally we're delivering hundreds of millions of dollars into our customer's bank accounts. Obviously, that has to be paid for, but actually the reverse of that is nothing worse than having all of these initiatives that are in place and support that's available for customers who need it. They don't avail themselves of it just simply because they're not aware of it. We've found the level of support that we've had has been very, very good. I think there's enormous potential in a post-COVID world for us all to learn the lessons about what worked most effectively during that time. Clearly there's always risks that come with at least some of that work and certainly, I guess recent events here. There's, as there should be, a huge focus on customer privacy and data security, but there's certainly ways that that can be done very safely and deliver high quality service and measurable improvements in outcomes.
Sri Ambati:
It's a certain level of upskilling that needs to happen as well as part of that. I think the education and the university's collaborations, the universities and your spot on with the data ecosystem. I was in the audience at the IPaa when you were talking. It was interesting that the public sector has more data, and the private sector is interesting. Together we have more data to serve our customers. That's right. Building that ecosystem partnership would be super keen.
Matt Comyn:
Yeah. Clearly it's an area where there's really strong demand for skills and expertise, and I think a fundamental asset of leaders more broadly. That's also one of the reasons, I guess, why we're sitting here as well. We want to be, as I said earlier, doing really high quality work that makes a big difference at a national scale and be able to attract the best people in market to come and work as part of that team.
Sri Ambati:
I remember ever since our first conversations with the bank have been your team wanted to be the world's best users of AI world's best users of H2O. Today we have the world's largest gathering, or Australia's largest gathering of grand masters of data science. One is to five gold medals to be a grandmaster, most of them in the audience here. Hopefully, we'll want to come back a year from now. I know we met we, we are fortunate to have your partnership in strategic investments last year, a year almost to date. We are here trying to build the grassroots. I think hopefully a year from now we'll have a few grand masters from ComBank or, or rest of the continent. We are hopeful to upskill the teams.
Matt Comyn:
I'm extremely hopeful about that too, Sri. I was a bit wary of naming them individually in case people will go out and try and poach them, but there's some extremely good talent capability, which I think it's wonderful they've got, and a couple of them made some pretty strong commitments about the progress we're going to make over the next year or two. I think it's fantastic to have such a highly motivated, engaged team with really ambitious goals and long-term goals with lots of perseverance that deliver really great outcomes.
Sri Ambati:
Yeah, I think the cross collaboration between technology, data and business is going to be super, super important. That's something we're seeing in a really generous mountain in our conversations with your teams. It's super. We're super, super hopeful to make Commbank world's blessed bank using AI. In terms of taking the responsibility, you touched on it briefly. We have a talk later today on AI in data security and how to prevent inadvertent loss of data, which we touched on in September with stronger AI as a defense. From a responsible AI standpoint, how do you or what kind of questions are you seeing from the region and as well as the global interactions here.
AI Used For Data Security
Matt Comyn:
Yeah, I mean, again, enormous amount of interest as you'd expect, both from directors on the board of the Commonwealth Bank, as well as other stakeholders including our regulators. That's, again, been a real priority and focus for the team. I think we're one of the four companies in Australia that's worked alongside the development of the ethical principles of AI with the government. I think we have recently walked through a whole range of different elements of that with regulators that our approach. How those models are supervised. How we manage for a whole range of different things to ensure that there's explainability, no bias. We understand drift. The team, I think, have done a great job around leading implementation and management framework for that.
Of course, I think that's an area that will clearly continue to evolve. And It's something that we're certainly always looking globally to see if there are better practices, but I think we've certainly had it at the forefront given the sorts of services that we provide to customers. I mean, nobody likes the thought of unsupervised models making a whole range of different decisions. We have to be very prudent and thoughtful about exactly how we deploy models. As you'd expect, there's lots of frameworks and policies and processes inside large financial institutions to make sure that that's really well managed.
Sri Ambati:
How do you keep up with all the changes of AI? It is one of the fastest moving spaces, and as far as senior leaders go, you're probably one of the foremost we've seen in terms of being able to keep up with changes happening in AI as well as the talent changes. How do you personally keep up to date on some of these spaces?
How Commonwealth Bank Stays Up-To-Date With Fast Moving AI
Matt Comyn:
Well, I think probably in the knowledge level of just enough to be dangerous. I rely on my team, as you know Sri. I ask lots of questions both internally and externally and you know we're overseas either seeing investors or seeing other institutions or tech companies or people who are making investments or some of the companies that are coming through. Yeah, try to flip across it. There's a lot of talent globally going into this space. It's very important, from my perspective, that we've got, as I said earlier, world class talent and capability that's extremely engaged. You want to just keep pushing on making sure it's a great place to work. I just think it's very hard to make a lot of progress without a great team of people in this particular area.I want to make sure I'm certainly up to date and up to speed. I think about how I and the team can make it easier for those people to get high quality work done.
Sri Ambati:
I'm fascinated every time we have a conversation on the latest you're able to keep up with. Despite your own self estimate, I think it's really one of the most incredible leadership team we have. We've been working with amongst several companies globally. Some of our, I mean, there's probably one or two companies that come nearly close to this level of sophistication and deployments as well as care and making sure it's responsible or tight. It's very pervasive quantitative thinking across your senior team. It's just absolutely a pleasure to work with as well for us.
Matt Comyn:
Well, I mean, I guess in the same spirit, we love objectively measuring ourselves and benchmarking. It's great to work with you. There's a number of other companies as well that we try to get a bit of a sense of: how do we stack up, where are we, where are we doing well? Where are we on par, maybe where have we had a lot of progress? Where are we lagging? It's like I said, it has to be right the way through the stack as well in terms of starting with all of the data and then to the end deployment production. We've definitely increased our progress in a number of the areas that we're tracking pretty closely.
Sri Ambati:
I think when we have an AI maturity model that's usually trying to track and see how the models are being deployed, reused, the feedback loops across different parts of the organization and how it can cross pollination and cross teams work. I think we're beginning to see some really good measures quantitatively even across CBA. Yeah. As a force for good. And one of the exciting aspects of our strategic partnership a year ago was wanting to use it to fight bush fires or wildfires. How best can we bring AI to help the community and the community of your customers as well as the community at large here?
How Can AI Be Used To Help The Community
Matt Comyn:
That's a really good place to start. As you know, Australia sadly is prone to a number of natural disasters and emergencies. Clearly that's been one fundamentally, I think also just being able to make sure that we're helping our customers. There's a lot of benefit to doing that at scale. I think the other area that you and I have discussed as well, which I know something you are passionate about as well will be in and around climate and the transition, and we're certainly using that in some areas around across our retail customer base in terms of measuring emissions and being able to see show people's carbon footprint. I think there's a lot of improvements that can be made generally in that space.
We're doing a lot of work with CSIRO at the moment, which we hope to be able to publish in the new year. The pathways for a number of priority sectors within Australia, which we're obviously bringing together a lot of the data that we have with the work that CSIRO are doing. I think that's clearly, it's extremely complex and challenging for the globe and clearly for the country as well. I think that undoubtedly will be an area that will continue to refine. They'd certainly be at the broadest level, I think where I would say we're really prioritizing. But like I'd say the other one I talked a little earlier about a particular improvement on a fraud model.
Another area we're doing a lot of work on at the moment and a number of us are passionate about is in scams. I'm assuming it's the same in the US. We've seen a dramatic increase in scams during, or certainly accelerated during COVID and coming out the other side of COVID. I think there's enormous potential to try and identify. There's a number of things I won't pre-announce, but we'll probably be talking about them in the coming weeks and months. Both partnerships and just ways that we can help serve up better predictors for customers to really try to create more friction and prevent customers from losing money in scams. I think it's an extremely important topic and one that we want to make sure that we're leading in that space
Sri Ambati:
Many of your customers specifically have been asking about scams and using graph data and ways to really address their problems. Certainly, I think the talks after this are going to be touching some of the transactional abuse use cases as well. Yeah. So it is quite a meaningful difference for the customers and community with AI.
Matt Comyn:
I think that can make a huge difference at scale as there's hundreds of millions of dollars, if not billions across the economy that's being lost. It's a very worthwhile goal to go after, to try and eradicate or try and certainly make a very significant dent in that.
Sri Ambati:
One of the interesting things you're seeing is the co-creation with the customer. and specifically CommBank has such incredible data assets, but also documents, and obviously we touched on the KYC, being able to co-create and then see if that's useful for other agencies locally or globally to start bringing more asset light teams to the core business. We need to see some of that in the US where one of our customers manages to take as a service to the governments and other their own customers. Do you see the bank thinking about a partnership where data and AI could become not just a force multiplier insight, but also start becoming a very strong source of brand and community building?
AI Brand And Community Building
Matt Comyn:
Yeah, I mean, we'd certainly like to think so. I think the one that we were talking about, particularly the document AI. It just starts with a fundamental problem, unfortunately, like a legacy remediation, a complex error around financial advice where you're otherwise individually assessing and reviewing huge volumes of data. It's a good example insofar as them, let's put the team alongside that. Let's see what other things are available today? Is there a different way that we could build a service that's actually much more reliable, much more efficient and much more accurate? Then, of course, that turned out to work extremely effectively. Then the question is, okay, well, if that's so good at recognizing particular documents, what's the broader scope and applicability?
Clearly we and governments and other corporations will be digitizing all of that , those traditional paper records. I think there's ways that again, link to things like fraud and scams and a whole range of other important, both obligations or ways for us to protect the community, those sorts of services. I think there'd definitely be no hesitation. If we think that we've got something, and I dare say the team would say we do that's unique, that would add value to other partners, then that's certainly something that we'd be prepared to share.
Sri Ambati:
I think the heart of that thinking is organizations should start thinking of data and AI as a fundamental asset and start seeing how to and I mean, what we would say at least early lessons in that journey that we can that some of our audience can aspire to learn from being, making using data and AI as an asset.
Using Data And AI As An Asset
Matt Comyn:
Yeah. Well, I mean, Sri, I think you and others will have far more pearls of wisdom than I will. But as we've been staring into this problem for many years, it's large incumbent organizations have an enormous amount of data that is in lots of different places. It's taken a lot of work to get high quality and consistent ways to put the data with a very clear mechanism to be able to serve that data up, to store it, to keep it safe, to keep a lineage and a record to have controls around that. Then to be able to create capabilities inside the organization. The team will talk a little bit about the workbench.
They've basically created some consistency around the way basically to access data. Because what we found at least is that several years ago, there's lots and lots of teams that are working on data at any one point in time. Everyone's pulling data from one place to another and putting in individual data stores, and there's incredible inefficiency and risks around that. That's been an enormous focus. Then of course, just having a common and consistent framework tool set that they can then use and deploy and then quite a large team, but with a real center of excellence that's looking across the landscape of the organization. I think you touched on it earlier, really challenging ourselves around what are some ambitious goals that we can really try and persevere and and try to make some real progress.
As I said, the team has done a great job, I mean, bringing your company to us, and it was really internally led by our data team, and that was fundamentally one of the reasons why we were prepared to make the investment. I'm sure it was great anyway, but I wasn't necessarily going to be able to evaluate that. People inside the organization who are really passionate about it and it's as much about supporting and having confidence in them to keep evolving and pushing the thinking. I'm sure in 12 months, 18 months we will have made a lot of progress, but also, it's just fundamentally such an important direction across all of the dimensions that we've discussed today to get right. I just think there's enormous unlock and benefits for us to be able to provide to our customers.
Sri Ambati:
No, we feel privileged to be a partner in that journey. To the extent that we have a fully knowledgeable customer on our board, we have and thanks to your support, and the board as well, and we get direct feedback to learn from your journeys and our customer experiences. So super, super thrilled at what we've managed to even build over the last three, four years. Actually, many of those years were in COVID, so we couldn't actually come see, but as you promised I remember well during the race that as soon as it's, we skies reopen, I'll be there, I'll be here, but you beat me to the punch and came to Silicon Valley before I could come. I'm glad to have built a very strong partnership with the bank as well as the whole team of makers in the community globally. One final probably thirty second question, probably more learning from what is a leader of today? You're a very inspiring leader for most of the generation. A leader of a large institution with incredible talent and vision and ambition. What should a leader of today aspire to and learn from your walk and journey?
What Is A Leader Today And Lessons Learned
Matt Comyn:
Well, maybe I'll just say, what are we really focused on? We've obviously been touched on a number of them. I think one of the things we really learned through COVID and again, I don't, this is a startling insight, but how both the level of engagement but also pride in being part of something that's bigger. I think for us, it's one of the reasons why we updated the purpose to be about building a broader future. We've thought about and orient a lot of our strategic thinking on a bigger purpose. I mean, clearly we're a financial institution. We serve millions of customers. I mean, there's a lot of things that we do in the day-to-day, which are incredibly important in terms of our operational performance and service of those customers.
Also, I think there's a lot to be unlocked, a lot of discretionary effort, a lot of engagement, as I said, pride. I think also long-term competitive advantage to being seen to be, and actually delivering something that's a lot more important than the near term. This is like directionally an area which unlocks so many different benefits and potentials and I think enables us to really challenge some of the things that we'd otherwise be thinking about. I think that's critical for leaders of today to be able to attract the right talent, to be able to retain them and to be able to really think long term and to push that thinking.
Sri Ambati:
What a pleasure, what an amazing piece of advice for future leaders as well. We in the audience and what a pleasure it has been to have you at H2O World kick us off and, and inspire our community with the incredible journey that the bank is taking and your leadership. Super thrilled to have spent the time with you and look forward to many more sessions. Hopefully, a year from now, we'll be able to look at having changed 5% or 10% of the bank to be AI makers. Hopefully, that's a goal we can start to strive for. But super, super keen too and very excited and thrilled. It's a privilege. Thank you.
Matt Comyn:
Thanks, Sri. Always a pleasure.