In Business with AI | Leading AI transformation

Melbourne Business School Dean Professor Jenny George sits down with Cynthia Scott, the Group CEO and Managing Director of Zip Co, to unpack how she is transforming Zip into an AI-first organisation.

In Business with AI

In Business with AI is a podcast presented by Melbourne Business School that explores how artificial intelligence is reshaping business, leadership and the future of work.

Hosted by Dean Jenny George, each episode features conversations with AI experts and CEOs who are implementing AI in unique ways in their organisations.

We unpack what’s working, what’s not working, and how leaders can navigate one of the biggest technological transformations of our time. 

Episode 3: Leading AI transformation

In this episode of In Business with AI, Melbourne Business School Dean Professor Jenny George sits down with Cynthia Scott, the Group CEO and Managing Director of Zip Co, to unpack how she is transforming Zip into an AI-first organisation.

Cynthia reveals Zip's three-pillar AI strategy, how employees work alongside an agentic workforce, and how AI is reshaping everything from customer experience and engineering to employee wellbeing.

She also offers advice for leaders looking to put people at the heart of AI transformation. 

Transcript

Jenny George

Hello, and welcome to In Business with AI. I'm Jenny George, I'm the Dean of the Melbourne Business School, and in this podcast, we give leaders takeaways and insights that you can implement in your business today. I'm so excited to be joined by Cynthia Scott. Cynthia's the Group CEO and Managing Director of Zip Co, leading the company's financial performance and strategy across Australia, New Zealand, and the United States. And she spent her career navigating disruption investment banking, in telecommunications, media, and FinTech, and now using AI to disrupt the financial services industry. We're going to unpack what that looks like in practice. So, thank you, Cynthia. 

Cynthia Scott 

Thank you, great to be here. 

Jenny George 

I'm so excited to hear about how you're implementing AI at Zip Co, but I actually think before we get there, it'd be interesting to hear your background. Sure. You've come through many different industries and done lots of different things. How did you end up where you are? And did you have a dream of working in finance all your life? 

Cynthia Scott 

Thanks, Jenny. It's a good question. I had a dream of what I was going to do with my life. I have to confess it probably wasn't to work in finance. Sure. And I have certainly had what people describe or I'm having at people describe as a non-linear career. So actually my dream in life, some people may be aware, was to be a ballerina. And so I left school at 15, trained as a classical ballerina, and that didn't work out. So I then went to university. I got the HSE, went to university, and then had a sort of a finance path. But yes, I spent 20 years in investment banking, and I thought that's what my career was just going to be. But then I moved into corporate. And then so for the last 10 years or so, I've had a number of different corporate positions across various industries, with the privilege of leading teams across all different sorts of countries and cultures and contexts. And then five years ago, I joined Zip. 

Jenny George 

And one of the things I noticed was that you were involved with M&A for Telstra. Yes. And so I think more than 25 different deals. Tell us a little bit, because it's quite a transportable skill. How do you look at something and say, that's a good bet, or we shouldn't do this one? 

Cynthia Scott 

Yeah, it's interesting. So when I was an investment banker, actually my technical skill set was capital markets and capital management. So I came up through the fixed income debt capital markets side of the balance sheet and then moved into management. When I then made the decision to leave investment banking and move into corporate, I knew that it would be a big career step for me. And so I wanted to find an organization where, I used to describe the fact that my technical skill set was managing capital, or my passion was managing human capital. So I needed to find an organization where they had a lot of capital or needed a lot of financial capital, or they had a lot of human capital, or there was change. And Telstra just fit the bill perfectly. And yes, I did end up, one of my first roles, I had responsibility for M&A, but actually M&A was not my technical skill set. So I sort of moved into that world. And I think since the Telstra role, almost every role I've had has had some element of strategy and also investments and also things, as you mentioned, like venture investments and so forth. So I think one thing that if I drew a sort of red thread, if you like, through everything that I've done and think about what's sort of the common pieces of what I've done, it has been this willingness and ability to look at opportunities and to find the proverbial needle in the haystack or find something that is very appealing about a particular business or strategic direction or challenge and find a path through. So in the case of Telstra, yes, I had responsibility for M&A. And yes, in my time there, we undertook a lot of mergers and acquisitions, but I also was the chair of Telstra Ventures, which was the venture capital fund was called Telstra Ventures back then. And so I had the pleasure of working with the two managing directors there and looking at lots of early stage investments. And I also had responsibility for Telstra Health, which was the digital health business, where again, we made lots of acquisitions. So the question that I've often asked when I think about M&A is why, particularly if it's an established business, why is it that we would be a better owner for this business than the current owners? And what is it that we can bring to this opportunity that will actually really bring benefit to the stakeholders, not only the owners of the business, but also the customers and whoever else the stakeholders in and around the business. So I think it's important to be able to identify the opportunity, but be very clear about what you need to do and what conditions you need to actually deliver that value. 

Jenny George 

And that's a great strategic lens, isn't it? Because in order to be the better owner, there has to be some rationale. It's not just money. It's not just this is a potential thing we could do, but this is a natural home for this business. 

Cynthia Scott 

Yes, yeah, no, that's right. And actually, if I think through the different roles that I've had in the different industries that I've been in and the different operating environments. At times, there was a very stable operating environment. At times, there was quite significant change and disruption. The core principles that I often come back to are what are our strategic assets and what are our strategic differentiators?  

And I actually think it's really important for every business and every team to be able to articulate what are your strategic assets and what are your strategic differentiators?  

And it's a good exercise actually to do in your own team, to actually get people to communicate, well, what are your strategic differentiators? And I'll often sort of provoke people and say, well, your competitors will probably say they've got the same strategic differentiators. So is it really actually? Is it different? Yeah, and so I do think that that's a good lens or a good exercise to do for teams and for companies. 

Jenny George 

And let's then talk about Zip Co. Yes. This is not the boiling water company that we think of when we hear the words in. Tell us a little bit about Zip Co and what you do.  

Cynthia Scott 

Yes, there's certainly brand confusion. So in Australia, often when I say people, I tell people that I work at Zip, they say, oh, we've got them at home. And I'm like, no, no, no, you've got the Zip Industries hot water taps. We're Zipco, the financial services organization. There's a couple of companies called Zip. So brand confusion is certainly real for us. So Zipco is a digital financial services business and it was founded in 2013 by two fabulous Australians, Larry Diamond and Peter Gray, who wanted, effectively they saw a customer problem and they wanted to provide a better solution for customers. And the opportunity that they saw was finance at point of sale. So back in those days, actually getting finance when you wanted to buy something was clunky, it was paper-based. And they saw an opportunity to effectively disrupt credit cards and to digitize what was a very laborious process. And from that, Zip was born in 2013. They listed in 2015 and then grew very quickly to around 2021, we were operating in 14 countries. And we've been on quite a journey since then to simplify the company, really drive profitability. And now, as you mentioned, we operate in Australia and in the US. And while Australia is our home market and we're very proud to have nearly 10% of the adult population as Zip customers, 80% of the business is now actually, the earnings are now actually generated from the US and our US business has grown at 40%, or more than 40% for the last two years. So it is very much an Australian digital financial services company with our roots here in Australia, but a lot of our growth is now coming out of the US as well. 

Jenny George 

And some of that I know is strategic choices that you've led. So you've been there in the business, you've delivered profitability, a strength and balance sheet. And that narrowing the focus to two core markets must have been a little bit nerve-wracking because in a sense, we're used to the idea of growth being about doing more. And in this case, growth was maybe about being more focused. 

Cynthia Scott 

It was, and I would say to you, leading a business is always contextual. It really depends what's the operating environment in which you're leading the business. And when I joined Zip initially as the CEO for Australia and New Zealand and then became the Group CEO three years ago. And at the time that I became Group CEO, the capital markets were very challenged for financial services, consumer finance, business, FinTech as we are.  

And it very much was the case that the market and our capital providers wanted us to demonstrate sustainable profitability. So I was very confident that we had a very strong business. We had really good strategic assets, but we did need to narrow our focus and really ruthlessly prioritise driving or delivering and then driving sustainable profitable growth. And that's what we did in FY23 and 24. And then we've gone on to deliver 12 consecutive quarters of profitability. So it was very important.  

Jenny George  

Congratulations. 

Cynthia Scott 

Thank you.  

Jenny George  

That's amazing. 

Cynthia Scott  

Well, the team have done it, but it was very important at the time to narrow the focus of the team and really drive that very deliberate energy around profitability and growth and make sure that we delivered that so that we then earned the right to grow in other ways as we're now doing. 

 Jenny George 

One of the interesting things I think about that narrow focus is you now have quite in-depth knowledge of two important markets. Yes. As you look at the US market and the Australian market, where do you see the main differences? 

Cynthia Scott 

Yeah, well, actually our businesses are quite different in each market. So you're right. We've got, as I said, nearly 10% of the adult population in Australia are zip customers. And our zip customer in Australia may be a little bit different to what people expect it to be.  

We have an average age of our zip customer in Australia is 40 years old, and we skew roughly equal, 50-50 male and female. A lot of the spend that we see from our customer base in Australia is both discretionary and non-discretionary, but we also provide finance for things that people probably don't automatically associate with zip.  

So whether that is solar, it might be you're putting a pool in, you're renovating your kitchen.  Interesting. And so actually us, zip being there for our customers in Australia for everyday purchases, like literally our customers use us more than 25 times a year on average for everyday purchases, but then also for life's bigger purchases, we are there for our customers.  

Our US business is slightly different in so far as the nature of the products that we've got in the US are different to the products that we have here in Australia in terms of the size and the duration and the structure, but also our customer base is quite different. While we've been operating in Australia since 2013, we've been operating in the US and the products themselves have been available in the US for only about six or seven years. 

So it's much earlier in the penetration of consumer finance at point of sale. And our customer base in the US is very much a low to middle income American. So about a third of the US population, more than 100 million Americans live paycheck to paycheck and really need financial services products to help them smooth the mismatch between their income and their expenses. And that's where zip comes in. And that's really what people, our customers in the US are using our products for. It is largely non-discretionary. So I think grocery fuel, everyday expenses, where they just need some help in a month, getting through that sort of lumpy cashflow that they have to manage. 

Jenny George 

So one of the things that's really fascinating is looking at disruption of the financial services industry and AI is a big part of that. How are you viewing the disruption that's coming for financial services in particular? 

Cynthia Scott 

Yeah, it's a really good question Jenny. And actually, if I think about the various roles I've had in different industries over the years, one thing that I've always said to people is, it's really good to have a healthy dose of paranoia and thinking about disruption, even in a business like Zip that is relatively young, I'm always thinking about, well, who's disrupting us?  

Who's coming, who's around the corner, looking at our business, looking at our customer base, at our merchant relationships, the nature of the products we've got and what is the potential disruption that might derail the growth that we've got at the moment, for example. And I think you can look at that from a product lens, a customer lens, or just the way that you do business. AI is an enormous opportunity. And we made a strategic decision two and a half years ago now to really lean into AI. And I just wanna contextualise for everyone, we are a digital business. And so we are a business full of technologists, engineers, who really want to lean into technology, but also who have a natural bent to want to disrupt and to innovate. So it is somewhat easier in an organisation like zip, you're pushing on an open door where you've got hundreds and hundreds of our zipsters who want actually to embrace technology and who want to use these tools. So we made a strategic decision a couple of years ago to lean into AI, and to really harness the power of AI as a tool to reimagine how we do business. And the way that we went about it, again, two and a half years ago, it was very deliberately to set up two things. 

One was we said we've got three pillars of our AI strategy, AI powered people, AI powered processes, and AI powered products. And I'll talk a little bit about why we started with people first. But the second thing that we said is that across the company, we said to everyone, we want you to embrace this as an opportunity. And so when you think about AI, think about our business and use the opportunity of AI to automate the mundane, and then we want to humanise the complex. And so the reason why we started with AI powered people was we wanted to give all of our, we call them zipsters, wanted to give all of our zipsters the tools to begin to experiment with AI, but we also wanted to give them the training. And so whether that was our engineers doing a bespoke university course in the US, or every zipster going through LinkedIn learning, we wanted to demystify AI and help them understand that this is actually a really big opportunity for them as individuals and for us as a business. So once we got our teams to embrace AI across the board and not think of it as a threat or something that potentially could be disruptive to them, once we got them to embrace the tooling, we actually then said to them very deliberately, you know our business better than we do, I do. So you go and find the processes where you know that this is an incredible opportunity to take repetitive parts of our business and use AI, use an agentic workforce to deliver that part of the business. And then we need you and all the fabulous things about humans to actually really focus on the important things for, and for example, engaging with our customers, more complex queries from our customers, hardship, credit decisioning, things where we need our humans and we need our team to actually really lean into that kind of work, to give them capacity to do that, we use AI and an agentic workforce to take a lot of the more repetitive, mundane parts of our business out. And I will say the real acceleration we saw in AI adoption, so now 100% of our team use it, you know, sort of 85, 86% of our code is now AI enabled. That real acceleration in adoption across the company happened when we gave the tools to the team and we said to the team, you go and find the processes because you know this business much, much better than we do. And the Zipsters absolutely lent in. 

Jenny George 

And of course, one of the things in point of sale is what's point of sale going to look like when you've got armies of AI agents who are the point of sale? And that must be such a fascinating question to ask because in some ways that opens enormous opportunities. It does. Or something like Zip Co. 

Cynthia Scott 

It does, Jenny. And so, you know, what we were just talking about was AI and the impact on a business like ours operationally, internally. You understand that Internally, the more strategic, more like the as interesting question is actually the strategic one. And this is something that we started to grapple with about 18 months ago is we need to completely reimagine our company in a world of agentic commerce. And by that I say, look, 87% of commerce still happens in a physical store. I agree with that, but the payment obviously still happens through payment rails. But if you think about, if I sort of break it down, if I think about a business like ours, which is a digital first business, digital native business, our business, like most consumer facing businesses, was largely created using human centered design. You know, that's where you sort of, you think about, well, how is a human going to move through our ecosystem? How do humans search for products? How do you recommend products? How do they make decisions? And that's how Zip Co was created. And not only created, but also how the business has been refined over years and years and years. 

Jenny George 

And of course, best practice. 

Cynthia Scott 

And we know. 

Jenny George 

That's exactly what we're all supposed to do. 

Cynthia Scott 

Absolutely, and so we know, if for example, you see the zip widget on a product page at a retailer, that you are more likely to convert your basket because you know that you can get finance.  

We know how to move a human through a merchant's ecosystem and our ecosystem, and we know how we can maximise basket size and conversion, et cetera, et cetera, for humans. Where commerce becomes really interesting is where it's actually not a human in your ecosystem, but an agent. Because an agent is not going to respond to the same sort of stimulus that a human will. And so human-centered design is maybe a little less relevant when you're thinking about a world of agentic commerce.  

So about 18 months ago, we made a decision to really lean into agentic commerce and actually have a seat at the table of the most important decisions that are happening at the moment around agentic commerce. And by that I mean, we are a provider of credit, but we're also a facilitator of payments infrastructure. And there are some very large participants in our ecosystem who are very interested in what happens with agentic commerce. They're much bigger than we are and they've got much deeper pockets, much deeper pockets than we do. But they are all very, very focused on what will the world, the future of agentic commerce look like. And so our objective at Zipp is to be at that table.  

And that's why you've seen us enter into partnerships and work closely with Google, with Stripe, with Visa, with Ixopay, because these are the organisations that are defining all of the aspects of a value chain in agentic commerce. And I don't think that you or I or anyone yet can actually really opine on whether or not it's going to be Google or it's going to be Visa, it's going to be Ixopay. Look, all the protocols are being built now. And so our objective is to make sure that we have a seat at the table so that we are aware of what's going on and we can make a contribution to the landscape of agentic commerce as it's unfolding. 

Jenny George 

And it's so interesting, isn't it? Questions about whether essentially the agent will use the payment mechanisms of the human or will they be given their own identities? Will the agent in fact be its own kind of separate payment structure, if you like, that gets built for agents specifically? 

Cynthia Scott 

So there is certainly a school of thought that says that the whole internet protocols and everything, there'll be a human one and an agent one. But it's true today that agents are in our ecosystem and they are in everyone's ecosystems. What's interesting that we need to think about from our perspective is we are a provider of credit and ultimately, as a zip customer, Jenny, we will credit decision you. And so it's you that we're lending to.  

If you then actually have an agent acting on your behalf, you can imagine there's a myriad of considerations from our perspective if it's actually your agent in our ecosystem seeking to affect a payment. So, you know, and these are the sort of protocols that I was referring to. Identity verification, the authenticity of the instruction that you gave to your agent becomes really important. And so at various steps along the way, there are now protocols being developed around agent authentication, determination of who takes with your agent that you gave an instruction to, does not do what you said. Is that your responsibility or our responsibility? 

So it is actually a really multi-layered complex ecosystem that's unfolding, but it's also really exciting. And then the other thing I'll just flag is that because of the nature of our role in this, we do see millions and millions of transactions move across our platforms. And we've got 4.6 million customers, you know, in the US, and we've got 1.9 million customers here in Australia. So we've been surveying them every quarter on consumer behaviour around agentic commerce. And what we're interested to understand is that to what level of customers or humans comfortable using AI tooling, and to what degree are they comfortable with all of the various stages of an agentic commerce journey? You won't be surprised to hear. The vast majority of customers are very comfortable and are already using AI tooling, and in some cases agents, for search and recommendation. And roughly three quarters of our customers say that what they found out through their AI search helped them or made them change their recommendation of what they were going to purchase, which is super interesting. But still close to 80% of customers say that they're not yet comfortable handing the actual payment over to an agent. Now my hypothesis is that's going to change quickly, and we need to be ready for that. But right now the consumer behaviour is such they're not quite at the stage of wanting to do that. So again, not surprisingly, the most important things as this evolves is identity, making sure we actually really verify who is undertaking the transaction, but also trust. And so trust becomes a very, very key component in our relationship with our customers and the world of agenda commerce in totality. 

Jenny George 

Do you see a difference across geographies? One of the really interesting things that we found, there's a Melbourne Business School Research study, 47 countries found that Australia is one of the lower countries for trust in AI, not in usage, quite high in usage, but quite low in trust. But other countries had some different approaches and certainly emerging economies, China, India, much higher trust, much more willingness to put more things into the hands of AI. Are you seeing that yourself? Are there differences between the US and Australia? 

Cynthia Scott 

Yes, we are seeing that and yes, there are differences. It's interesting, there's also, as I said, our Australian customer on average is 40 years old, our US customer is slightly younger, not too much younger. Our US customer skews a bit more female, our Australian customer is roughly 50-50. What we do see is a little bit as you've just described, I've only got those two markets to compare at the moment, but it's actually pretty similar in terms of the use of AI and the use of Agentic for search and recommendation and so forth, but right now, yes, there's probably a greater willingness of US customers to go as far down the payment processing piece than Australian customers. And I think there's that element of do it for me, and we see that a bit more in the US than we do here in Australia. 

Jenny George 

Yeah, fascinating. I'll take you back a little bit to, you were talking about your internal AI strategy and I loved this focus on people, processes and their products. Tell me a little bit about how you go around encouraging experimentation. You gave people the tools, you did some training, you told them to experiment. How do you make sure that that experimentation isn't just having fun, but it's also really aimed on things that have got good eventual return on investment? 

Cynthia Scott 

It's a really good question. And let me take you all the way back to when we started this journey, so two and a half years ago, one of the great things about being a technology company and a digital first company is we know we want to experiment, but also our partners know that we want to experiment. And so what was happening across Zipp was we had a lot of our partners coming to us saying, let's do a pilot. And we just found all these pilots popping up across the company in different teams in different areas, whether it was customer experience, marketing, the engineering team. And it was somewhat uncontrolled in so far as there was not alignment to a strategy. And so what we initially did was we said, let's just focus our time and energy and our work with partners on marketing, CX, and our engineering. And that's where we started. And so yes, we encouraged experimentation, we encouraged innovation, but we tried to focus people's attention on not only certain areas, but also on our corporate strategy. And what were the biggest rocks that we wanted to actually complete in that particular year? 

If I scoot forward to where we are now, because AI and energetic workforce is just very much part of our DNA now, it's not a tool for us now, it's actually how we do business. We've institutionalized that experimentation. And so we have hackathons and we have one big hackathon a year in each region, but then we now have smaller, more frequent hackathons. And in fact, in Australia this year, in our Australian hackathon, we had more than 400 zipsters participate in the hackathon, which was phenomenal. And we had a hacker track and a learner track.  

So we encouraged people who maybe didn't see themselves as hackers to go down the learner path. And we got fabulous feedback from the team on demystifying how you use these tools to innovate and to disrupt. So even when we create formal events like that, we still give a theme. And so we will encourage people to focus the innovation and harness their energy around a particular customer solution or a particular product opportunity that we're building. So it's not unbridled and people can just do whatever they want. It is also very much focused on our big strategic goals and our big customer needs or merchant needs or product needs, and making sure that the team are rallying around that. 

Jenny George 

And one of the things I was interested in with your choices early on was those are things that probably don't immediately head you into problems with regulation. Because in financial services, there are some things where you're highly regulated. And experimentation in those areas is a bit trickier. 

Cynthia Scott 

Well, there's also just some no-go zones. So as I said earlier, there's parts of the business where either from a regulatory perspective, so within credit decisioning, or from a customer experience perspective, for example, within our hardship processes, that they are human processes. But then there are still opportunities within certain parts of the business to actually reimagine our processes, even within a regulatory constraint.  

So a good example would be here in Australia, there's very formal processes around collections. So when someone misses a payment, when they're late, it is very prescriptive about how many times you need to contact them, the form of contact, the evidence that you've provided contact, and you literally work your way through a very prescriptive process from a regulatory perspective, and you need to be able to evidence and demonstrate that.  

One of the agents that we've built, who is live and active in our business in Australia, is what we call the collections dialer. You can imagine as you get through to later stage collections, we need to call people to query why they've missed their payment, for example, and the consumer behaviour is typically they don't answer the phone, but we still need to leave a message, we need to send emails, there's a regulatory process that we need to go through. 

We've now got an agentic workforce handling a lot of those calls, leaving the messages, sending the emails, and of course if our customer picks up the phone, it gets routed directly to a human who then speaks to them directly. But in that scenario, our agents can do 4X the volume that a human can do. And so that's great, because it just means that then our customers are getting a better experience, because our human workforce can get through the volume faster. 

Jenny George 

And you've been on both sides of the boardroom on this, at the non-exec level as well as the executive. Thinking from a board's perspective, what questions do you think directors should be asking their CEOs about the way that they're thinking about AI or implementing AI? What do you recommend is on a board agenda? 

Cynthia Scott 

Well, yes, so I obviously sit on the board of Zipp as the managing director, and then I have some other non-executive roles where AI obviously comes up as a very hot topic. I think I'm very fortunate to have such a supportive board at Zip who have been on this journey with us and who've been asking all the right questions and ensuring that we've got robust governance along the way as we've moved very fast. I have seen a difference in the speed between the Australian business and the US business, just generally, not specifically at Zipp, but just generally, and often the risk appetite of the Australian businesses that I've spoken to is a little less than some of the US businesses that I've spoken to. And so being very clear about the governance structures, processes that need to be put in place to ensure that the boards have line of sight of, particularly how decisions are being made in the company, is critical.  

And so the questions that the board challenged me on for all good reasons is, show us how decisions are being made with AI. Because it's obviously not acceptable for directors not to understand how decisions are being made in a company. And so as AI becomes a particularly energetic workforce, becomes just more and more of how we're doing business, we are spending a lot of time with our board who are very much leaning into not only the AI opportunity, but ensuring that we've got robust governance and policies and processes around that work that we're doing. 

Jenny George 

And in financial services in particular, there's already some really important use cases and detecting fraud is one of them. Predictive analytics has been used for quite some time, great results showing that we can not just detect more fraud, but also have fewer false positives, hopefully get better at this. How have you gone about trying to improve the analytics using AI? 

Cynthia Scott 

It's so, again, it's very much now just how we do business. And so whether it's fraud, whether it's cyber, whether it's customer experience, marketing, literally there's not a part of our business where we're not now adopting and implementing AI. And the analytics is a huge part of that. It's interesting when I look at something like engineering. So within engineering, again, AI is very much part of how we do business, but typically when you're writing code, you would then have someone assess the quality of the code, debug it, et cetera, et cetera. And we've got agents who actually now do that work for us. And of course they are typically faster, the agents that are writing the code, and then the assessors who are assessing the quality of the code are typically faster and have less bugs and errors than our typical workforce. And so it's just meaning that we're getting through work faster and that the quality of the code is going up. So really across the business, it's absolutely very much part of how we do business. 

Jenny George 

And engineering coding in particular is clearly one of the earliest places where AI is completely transforming the way that things look. I'm kind of interested in the fact that your agents are both writing the code and you're so QA-ing it. Do you use different models to do that? Because sometimes if you use the same model, that has the same blights on. 

Cynthia Scott 

Yeah, no, we do. Yeah, yeah, we do. No, we absolutely do. And we also do champion-challenger. So when I say an AI, an agent might be QA-ing something. They might be QA-ing the humans work and the human might be QA-ing the agents work and so forth. And then sometimes we'll let the agent QA agents work. But yes, we do. And Jenny, you mentioned like engineering is an obvious place, but here's maybe a less obvious place where we got very early traction in our AI adoption through this. 

So when I said to you, we encouraged the team to go out and find the processes where AI could support them every day. And pleasingly, one of the first agents that was built that ultimately went viral across the company was an agent to support writing performance reviews. Because everybody leans into writing performance reviews and you know that time of view where you get all those emails saying, can you please review this person and here's your 10 people that you've got to write their performance reviews. 

Jenny George 

That's not your favourite task, perhaps. 

Cynthia Scott 

Well, it can just be really time-consuming and obviously it's important and you need to do it well. So actually our people and culture team built the agent, not the engineering team, the PNC team built the agent that ingests an individual's self-review, their peer reviews and then the manager's feedback and creates the first draft of the performance review. Now that agent is probably one of our greatest agent adoption experiences that we've had because it's very tangible that everyone can actually see just how beneficial that can be to completing that important task but in a much more efficient way. 

Jenny George 

And is that the kind of thing I guess that in a hackathon allows everyone across the business to be able to see maybe there's something I can do? 

Cynthia Scott 

Oh yeah, very much so. And in the hackathon we just did recently in Australia, the 25 ideas that came out of the hackathon. So this was a two-day hackathon and the teams had to build production-ready products, features, whatever it might be. And that's exactly what the team did. And the breadth of ideas that came out was incredible, whether it was ways to change our customer experience, new product features, ways to reimagine parts of our business today, it was just inspirational, candidly, not only what they did, but also how quickly AI enabled them to build a product or build the code to create a feature, to be production-ready. It was just every year we're doing it, it just gets faster and faster, which is fabulous. 

Jenny George 

And I noticed one of your strategic areas was CX, customer experience. Yes, yeah. I think you've talked about using propensity model to deliver nudges to customers. Yes, yes. How does that work and where does AI fit into some of that work? 

Cynthia Scott 

Well, actually, where we first focused AI in customer experience was not so much about nudging customers, and we certainly, we can use that because we've now put Zia, we've got two agents, Zia in the US and Ziggy in Australia. They are AI-enabled bots, if you like, but they're now both behind the login, so they're personalized, they can see what customers are doing and they can actually use propensity models to actually prompt behaviours.  

But where probably the bigger impact initially was when we started to adopt AI within CX was for the agent, like our agents, our customer experience agents themselves, and really helping them navigate the conversation with the customer, and if you like, using a propensity model to prompt our zipsters, our agents, to actually provide a better level of service to our customers. So ingesting all of our help material, all of our product material, all of our behavioural modelling, and actually helping the agent provide our customer with a better experience. 

Jenny George 

So more of a what's the next question? 

Cynthia Scott 

What's the next topic to address? Or what actually might be happening here. But an interesting, two interesting consequences of AI in a CX business, which obviously we identified and we've managed, one is that if you think about traditionally how you would measure the performance of a CX agent or a CX business, typically you've got metrics like call handling time, call wait time, the volumes, et cetera, et cetera.  

But in a world where all of your easier customer queries are actually being handled within a CX bot or an agent, those simple tasks go to the agent and get handled really fast, go to the CX agent, get handled really fast, and then the more complex customer transactions go to the human. 

And so all of your traditional metrics around things like call handling time go up because your humans are handling the more complex transactions and your AI agent is handling the simpler transactions. So we had to rethink how we manage and how we measure performance within our call centers. And then the other-- So 

Jenny George 

does that go to more of a quality rather than a volume metric? 

Cynthia Scott 

Yes, it definitely does. And then the other consequence which again goes to quality is we saw and we have seen in both businesses an increase in customer NPS since we've adopted AI. And that's because the simple tasks are being handled really quickly by Zia and by Ziggy, our AI agents, and then the customer is getting in front of a human to handle the more complex things that they need faster.  

And so then our NPS has gone up, which is great. But the other sort of consequence that we've had to adapt to if you like is that if you think about a CX agent, a human, their day would typically have ebb and flow. So sometimes they're doing really simple queries, answering emails, making small changes, and then sometimes they're dealing with more complex customer queries that you might have a not happy customer, for example. In the scenario that we've got now, all the simple work is being done by our AI agents, Zia and Ziggy. 

Jenny George 

And now everything's high stress. 

Cynthia Scott 

Now everything's high stress, or just more complex and longer calls. And so actually we've now also adopted AI usage to monitor the wellbeing of our agent humans, because we can record and monitor the tone of the conversation. How long is the conversation? Where, what's the stress level of the customer versus the agent? And our team leaders can monitor that on a dashboard real time. And so we then know if a particular agent's had a couple of difficult calls in a row, the team leader can go into Veean and say, I see, you know, on the dashboard that you've had a few challenging calls, why don't you take a break, for example. So it's just another consequence of thinking through the benefits of AI, but also some of the considerations around managing your team. 

Jenny George 

And I know, AI has been used for fatigue management, for truck drivers, for example. And so this is, I suppose, the call center equivalent. 

Cynthia Scott 

Yeah, it is the call center equivalent. Yeah, exactly, exactly. So there's always, and we often think about what's the customer impact of this? And you know, so we measure customer NPS, but then what's the impact on our zipsters? And then what's the impact on our business financially and operationally as well? 

Jenny George  

There's a giant fear, I suppose, that AI is gonna take away jobs. And in a business where coding is increasingly able to be done by agents, easy calls are taken by agents. It's an interesting question for a leader about how you alleviate anxiety, and how you just manage to bring people along with the AI innovations that you're trying to make sure are going into the business. How do you think about that? 

Cynthia Scott 

Well, I think adopting a strategy early on of starting with people was very deliberate for that reason. And we have most certainly seen a shift in the nature of work, but also how work is being delivered. And by that, I mean, like a good example of engineering. It used to be the case, and I think anyone who's got an engineering team would understand this as well and be seeing this. It used to be the case that when you're hiring for an engineering business, you would hire a mobile engineer, you would hire a front-end engineer, you would hire a backend engineer, and they all had discrete tasks. Typically these days, we, like most companies, just hire full-stack engineers. 

You need a more generalist engineer, and then they use AI to supplement the core skills that may not be their technical capability. It's more important that they're full-stack end-to-end. And so what that means, and what we're certainly seeing, is that while we used to have more of a roles-based structure, you know, my role is to deliver blah, now our structures are becoming more team or delivery-based. The output that we need is X. 

 And so who are the team members? What are the skill sets that we need to bring together to deliver that output? And so actually it's resulting in a shift in how you think about organizational design and delivery, as well as how you think about who do you recruit, and how do you continue to train people, and how do you promote people?  

So it's actually a super interesting development, and we see it as an opportunity, and we've also experienced that it's a way to unlock different career paths for people that they may not have considered. 

Jenny George 

And I suppose two things about that that really interest me. One is if a generalist is more useful in this AI augmented world, that says something about training, but it also may say we want people with a diverse set of experiences. And one of the things that's quite interesting about that is entry-level people may well enter through a specialization, through a particular niche, and then grow into a more generalist role.  

That suggests that entry-level is going to be more difficult because it's harder to be a generalist, maybe, as entry-level, or maybe not. I'm interested in your thoughts. 

Cynthia Scott 

Oh, look, I know there's been a lot of debate about what does all this mean for entry-level roles. The way that, certainly, my own perspective on this is actually this AI democratizes roles, and I often say it democratizes being a founder. It can, when I talk again to our engineering team, it used to be the case that we would only hire people who had engineering degrees. And now, you don't have to have an engineering degree.  

You have to have a curiosity. You have to be technologically literate. And we can train you, and it opens up the aperture of who may end up in an engineering role rather than having a very narrow definition of what it means to be an engineer. So I actually think if I was, and I've got kids of this sort of age, but if I was thinking about what might my career be like if I was just finishing university and starting out now, actually, I think the students who are coming through unis today are going to have multiple careers, and it may or may not reflect what their underlying course of study was, because I think there's much greater opportunities for them to move laterally across industries, across roles, because of the benefits of AI. And so I would encourage everyone to think laterally about what the opportunities are rather than just limiting. 

Jenny George 

And interestingly, I think both for companies, but also, of course, for us in the tertiary sector, one of the key things is helping people understand what good looks like, because I think that if you're going to augment with AI skills, AI will give you a result, but the human QA on that is, is the result I'm getting good or not good? And understanding that is critical. 

Cynthia Scott 

And there's parts of every business that you just cannot replace the incredible things that humans bring to it. And it's not only, it's interesting, it's not only the delivery of the business day to day, but it's also the DNA of what makes a business really special. And we've thought a lot about this in the context of standing up, we've got a human workforce, and we've now got an agentic workforce.  

So we've thought through what's the performance management process for our humans who are managing agents and humans, for example, what is the performance management process for an agentic worker? And this is something that is very much, we're leaning into it from our perspective at Zip, and you may have seen we just announced a collaboration with Google Cloud to create an AI factory, initially in our US business, to accelerate the work that we're doing around an agentic workforce. And there's all these other considerations around, well, not only performance management, but how do we think about maintaining and developing the culture of the organization, the values of the organization? And this is critical for us as a business, and it's as critical as the operationalization of an agentic workforce. So it's just not as simple to say, AI is better or the agent is better, it's much more nuanced than that. 

Jenny George 

How do you build company values into an agent? 

Cynthia Scott 

Yeah, so-- An AI agent. Well, so it's a really, really interesting conversation. How do you build company values? How do you identify performance? How do you manage performance of an agent? I mean, our values are very much sacrosanct, and they're very much how we do business every day. Someone challenged me once and said, "Would you replace a whole team with an agentic workforce?"  

And I would be really challenged to do that because you're going to lose a lot of the things that make an organization really special from a values perspective. So when people talk about agentic workforces and the agentic opportunity, often they're focusing on the governance side and the policies and the processes, which are critically important, but maintaining and growing the values of a business is also really important. 

Jenny George 

So actually my final question was going to be, what would you never delegate to AI? 

Cynthia Scott 

That's a great question. Well, I mean, operationally, there's some aspects of our business that I just could never see us delegating, whether that is hardship. So the management of the most sensitive aspects that have the greatest impact on our customers when they really need us the most, hardship, I just can't see that being delegated because it's an intensely human relationship and engagement that we need to do. And I think it'll be a long time before we get to the point of being comfortable delegating to AI some of the most important decisions that we make as a company, and another one would be credit decisioning, for example, things that are just really, really fundamental. And then I'll come back to your comment about values, you know, going down the AI path to the point where it would disrupt or reduce the values of a business or the importance of the impact of values in a business is something that, it's certainly not something that I would wanna have happen on my watch because it's just such a core part of who we are as an organization. 

Jenny George 

Wow, thank you so much, Cynthia. This has been just a fabulous conversation. Thanks for being with In Business With AI. We've got a little black book here, and this is one of the things we ask all our guests to help us with. So we'd love for you to give us one piece of advice that you think that people should take into account when they're thinking about being in business with AI. We'll put it on our website, so there'll just be a little kind of picture of this so people can go and have a look at it, but we're building up a record of our guests' advice to our viewers and our listeners. So thank you so much. We are so happy to have had you with us, and what you're doing at Zipp is both innovative, cutting-edge, but also deeply human, and I think that that's a wonderful thing for an Australian leader to be doing in an Australian-founded business. Thank you. It's got a global footprint, so it's fantastic. 

Cynthia Scott 

Thank you, Jenny, it's been a great conversation. That's a lovely thing for you to say. I really appreciate it. Thank you. 

Jenny George 

Thank you very much. Well, thank you for joining us, and please join us here on the podcast in the future. You can sign up to us, and of course, please get involved with the Melbourne Business School, our Institute for Digital Innovation and AI has many events, and we would love to engage with you. See you next time. 

"My advice would be - the winner in the evolution of AI will be "the fastest learner" embrace the opportunity and encourage your teams to lean in to the greatest strategic shift in our generation!" Cynthia Scott