Transcript: How AI is Transforming Merchant Experiences 

Johnny Ferguson  00:00:00

If we can give them intelligence that says, well, maybe you can open a branch over here. There's opportunities in this area for a coffee shop, or you might be able to get a better price because we see supply costs for coffee or alcohol or something like that.

Will Laws  00:00:13

Right now, we're seeing about 39% of people have discovered what product they're going to buy with AI today.

Dave Ellett  00:00:20

Wow.

Will Laws  00:00:20

39%. And if you look at what's happening with Gen Z, that's increased to 50%. And what's really important for merchants is that AI drives a 4.4 times increase from discovery to purchase over traditional search.

Johnny Ferguson  00:00:35

I think for a business that is looking to experiment, the single most important thing they need to do is educate their executive on what is possible.

Will Laws  00:00:45

The key thing I think about with this is: trust doesn't get disrupted by AI. It gets amplified.

Dave Ellett  00:01:00

Welcome to the AP+ Point of View podcast. We're coming to you live today from Beyond Tomorrow, our annual payments summit. You may hear some background noise through this episode as we've got a live audience in the room. With so many amazing guests here today, we want to take the opportunity to sit down and speak with a few of them. I'm going to be talking about AI and payments, and I'm delighted to have Will Laws from AWS and Johnny Ferguson from Tyro. Welcome to the podcast today. AI is obviously a huge thing that's been talked about in the press — you can't go a day without it being mentioned. And obviously agentic commerce and payments is a big topic, but today we want to focus more on AI and payments. I know both Tyro and AWS are heavy users of AI. Will, I might start with you. How is AI actually showing up in payments today globally, and how does it apply to Australia? And Jonny, I'd love to hear your views on Australia as well.

Will Laws  00:02:08

There are two main areas where we really see AI taking off in payments today: merchant onboarding and merchant intelligence. These are two challenges we see across the industry. We want to see how our customers at AWS can compress that onboarding time from days or weeks into hours, or even minutes. But that means addressing a couple of core challenges — KYC, verifying identity, verifying a business — which usually involves a lot of documentation. So we think a lot about documentation ingestion. We also look at the ways AI can work through intelligence: value-added services that provide huge insights directly to the merchant. We want to get our customers from a position of not just the statistics, but the next answer that comes out of that data.

Dave Ellett  00:03:13

And Jonny, what are you seeing in Australia?

Johnny Ferguson  00:03:17

I think the low barrier to entry for using AI is really for unregulated players or actors. Fraudsters embrace new technologies quickly — they don't have to comply with any regulations, they work on weekends. So the first place we're seeing the most action is around transaction monitoring and anti-fraud. Tyro is a banking, lending and payments company, and we see data from payments all the way through to lending and banking. What merchants really need is to get insights into their business. We're focused on providing merchant-understandable insights based on all of the data we hold on their behalf. That's where some of the really interesting work in the industry is being done.

Dave Ellett  00:04:18

How are you thinking about balancing innovation with the need for safeguards and governance? Will, I'll come to you first.

Will Laws  00:04:26

I don't think this actually gets talked about nearly enough. You have to balance innovation and safeguards. But in financial services and payments specifically, we have a huge opportunity to do this better than a lot of industries, because of our regulations. This is enabling us to say exactly where our data goes and how we can use it — and that's really important for AI. We think of this as an advantage rather than a challenge. Given that we build a great data strategy and already know where all of our information is, that's what really enables AI. That's where the actual intelligence comes from.

Dave Ellett  00:05:04

And Jonny, how are you thinking about balancing innovation and governance?

Johnny Ferguson  00:05:08

We actually just went through this journey with AWS. We like to harness the creativity of our people — the majority of our staff use AI daily, whether for development or marketing. The key is getting access to data safely. We invested quite a lot in building sandboxes so that our people can go and creatively look at data and start to come up with questions. People often answer this with "start with the business problem," but sometimes you don't know that concretely. So asking the data itself to look at problems — why is there a peak in traffic here, why are payments decreasing at a particular time of year — that's where you start being creative. You might just try something on a whim and end up with a product like Slack, which may have started as a gaming commerce tool. That's where you see innovation: being able to riff on an idea in a walled garden, taking those risks.

Dave Ellett  00:06:45

A lot of companies have been adopting AI. It's been around in payments for a little while, but I'm intrigued about some of the use cases you've seen in your organisations. We talked about data being the underpinning, but how is AI supporting payments specifically? A couple of use cases, Jonny — you start.

Johnny Ferguson  00:07:08

With Tyro being in payments, banking and lending, when you see the full 360 degrees of a merchant — the incoming and outgoing payments — you have a corpus of data much greater than any one of those alone. We can provide decisioning to the merchant. Rather than giving them a dashboard with a graph going up or down, it's more along the lines of: the graph is going up because of this, and here's what you can do. That's what we want to empower SMEs with. They don't have dedicated finance teams, bookkeepers, accountants or lawyers. If we can provide them with insights across those areas, we're providing protection and avenues for growth. And it's not hard to grow as an SMB — whether it's opening a new coffee shop, adding a product line, or in health, maybe opening a branch in a new area. When you provide actionable insights based on AI recommendations built on very high quality data, that's the value of AI. It's not necessarily a one-trick pony — it's everything together.

Dave Ellett  00:08:55

Will, what are some of the use cases you're seeing?

Will Laws  00:08:58

There are so many, but I think the under-talked-about ones are the ones that aren't as flashy. The boring ones are where we see so much business value right now. As you said, how can I get those actionable insights directly back to the merchant, or even directly to the customer? We also think a lot about document ingestion — it's really boring to sit there reading through documents, validating information. Having a process that enables a team to look at that and say, "these documents match, we understand where these things are" — while keeping a human in the loop for decisioning — that's genuinely impactful. But it's only possible when you connect your data, and it's still really important to have a data strategy and strong data foundations.

Dave Ellett  00:09:53

You have to have accurate data, because if the data isn't accurate, AI won't be able to add value to it in any meaningful way.

Will Laws  00:10:00

100%.

Johnny Ferguson  00:10:01

Just riffing on that — Tyro just acquired a company called Friday, and one of the amazing things they do is categorisation. For a merchant, reconciliation and categorisation of payments can take hours and hours a week — sometimes thousands of transactions. Friday gives us the capability to use AI to do categorisation for the merchant, which makes bank statements easier and tax easier too. If you use a product like that and it saves you six hours a week, instead of sitting at your computer late at night when you're tired, our merchants can actually be front of house welcoming people into their café or building intuition for their business. The boring jobs to be done are actually the sexy opportunities.

Dave Ellett  00:11:03

AI is changing the speed of payments — everything's happening at such a fast pace. Do you think it levels the playing field for everyone, or does it create new advantages for some?

Will Laws  00:11:18

I get this question a lot, and I think it's a bit of both. I see 200-person fintechs doing what used to need thousands of engineers to do. The pace of development is really fast. But there's still an advantage to established businesses: the playing field isn't just the model you have — it's the data you have. Established businesses have that in spades. For new businesses, the playing field means they can enter much quicker. So I expect we'll see a large number of successful fintechs coming up through the pipeline quickly, but also major plays by the established industry. The key thing I think about is that trust doesn't get disrupted by AI — it gets amplified. If you have that trust and relationship with your merchants today, this is an opportunity to expand your offering. If you're entering the field today, you still have to think about how you'll establish that trust, but you'll be able to deliver results faster.

Dave Ellett  00:12:38

That was actually a really good answer — you gave both sides, so I'll let you have that one. Jonny, I'm interested in your views.

Johnny Ferguson  00:12:46

I've got a slightly different opinion. It will lower the bar to entry across all industries — the tools are democratised, you can get them for $20 a month or even free. But I see that customers who have already made investments in high-quality data and a well-governed environment will be accelerated. The gap between organisations that haven't embraced this and those that are cloud-native or AI-native will widen. Those organisations will accelerate faster, and you'll see profitability shift significantly.

Dave Ellett  00:13:48

I know we said we wouldn't talk about agentic, but I will. In payments it's obviously a big theme — we've seen many players talk about it globally. Will, I'll come to you first with your global view. What are you seeing around agentic?

Will Laws  00:14:06

Agentic is expanding so quickly. Right now we're seeing about 39% of people have discovered what product they're going to buy with AI today. 39%. And if you look at what's happening with Gen Z, that's increased to 50%. It's not a technology of the future — it's something happening right now. And what's really important for merchants is that AI drives a 4.4 times increase from discovery to purchase over traditional search. Like this jacket I'm wearing: I wanted a new jacket, went to Claude, typed in my criteria and where I wanted to buy, and it gave me a list. I went to the store starting from the bottom of the list as an experiment, and the one I purchased ended up being the top recommendation. But when the agent is finding the product, helping you compare prices, and then completing the payment — that's the next step. And the question I put back to the payments industry is: what happens when the agent is choosing the payment method the consumer uses? With least-cost routing currently handled by a payment provider, what happens when it's potentially handled by a consumer-focused agent?

Dave Ellett  00:15:48

I think that's going to be a significant disruption for the industry. Jonny, your views on agentic?

Johnny Ferguson  00:15:59

We accept in-store and online payments, so agentic commerce is something we're building our systems to be ready for now. There are protocols like MCP that you need to be able to implement to be ready for identifying agents going forward. It's really around the infrastructure and identity, and then building the endpoints at a payments provider like us to be able to work interactively with a high degree of trust. When agents can act in a highly assured and predictable manner, that becomes really interesting from an innovation point of view.

Dave Ellett  00:16:46

You raise a great point about trust, and trust and identity coming together in that use case. Do you trust the bot to do what you expect? My wife and I have been using a lot of AI to plan a holiday in Europe and it's amazing how good it is. But would I go that next step and use it for a cross-border payment? That's the question for me at the moment. I don't think we're quite there yet, but there are a lot of players trying to solve that gap. As we look ahead to what's possible in payments, what excites you? What's happening globally? Will, you start.

Will Laws  00:17:27

The way that developers work today is changing really rapidly. We call that process the AI software development lifecycle, or AI SDLC. It's a cultural change happening in organisations: when I can output a huge amount of code with my team and get a deliverable executed quickly, the business needs to evolve how it thinks about deploying that into production. I have some personal projects I like to work on that I wouldn't have been able to start five years ago because I didn't have the time. Now I can sit down, watch half an hour of TV, and build a little app. When you apply the foundations and guardrails that businesses have around what's possible, real work is actually getting done.

Dave Ellett  00:18:17

Sounds like another SMB coming online here, Jonny. What excites you?

Johnny Ferguson  00:18:21

We've got an online solution for Will — he can get a special rate.

00:18:27

It all comes back to the SMEs for us. Tying banking and payments together — directional data gives us a unique proposition in the market to synthesise that data and surface it. I really love the idea of a merchant just talking to an interface we've provided and being able to ask questions like: which was the best-performing coffee shop I had, which supplier is giving me the best rate, what was my biggest month last year — and having it come up immediately on screen with a visual. That sort of assistance to the merchant, where it saves them real time and helps them grow, becomes really rewarding for us.

Dave Ellett  00:19:24

One final question for both of you. What's one specific action you would like to see to drive progress in AI in payments?

Will Laws  00:19:33

Start with your data foundations. If you can't make personalised suggestions or focus directly on what the merchant needs, you're not going to be capable of delivering value. Start with one problem, pick it, solve it, build the value out — but then sit back and collect the statistics. What is the customer satisfaction on this? How am I going to iterate going forward so I can expand to the next idea? Just take it one step at a time. As your data gets richer from that process, your model and way of working will get better, and that will provide your merchant more value. That is a flywheel — moving that data back into the pipeline, creating the next iteration of that algorithm.

Dave Ellett  00:20:24

And Jonny.

Johnny Ferguson  00:20:25

I've got two elements to this answer. For a business looking to experiment, the single most important thing is to educate your executive on what is possible. You would have seen a lot of memes around AI being sprayed on everything. When you overpromise and then underdeliver, you lose trust with your exec — and for me, that's the single biggest thing. I think as an industry — AP+, AWS, Tyro, and everyone else here today — the interoperability and basis on standards to develop trust is critical. We're not going to trust agents to do payments if we can't have confidence in identity and standards. That interoperability, and the collaboration between different entity groups — whether it's the CDR standards, AP+, cloud providers, or businesses like us — needs to happen in a very open and embracing way. And it's strongly related to CDR. Open banking and payments have to be completely integrated.

Dave Ellett  00:22:04

It has to be an industry focus to get this right. I did want to leave on the quote that Will gave us: if you get AI right, trust gets amplified by AI. That was thought-provoking for me. So thank you for that, Will. And thank you both for joining the podcast today.

Johnny Ferguson  00:22:21

Thanks for having us.

Will Laws  00:22:22

Thanks for having us.