How Fintech Leaders Are Translating Innovation Into Real-World Impact
HOST: Jason Henrichs
GUEST: Harriet Rees, Chief Information Officer at Starling Bank
What Were the Big Takeaways from Money20/20?
HOST (Jason Henrichs):
Harriet, Money20/20 always gives us a pulse check on where fintech is heading. From your perspective, what stood out most this year?
GUEST (Harriet Rees):
It’s fascinating to see how fintechs are maturing. The conversation has shifted from disruption to integration. We’re seeing established players and startups collaborate more intentionally, focusing less on hype and more on impact.
The biggest theme this year was about turning innovation into something useful. Everyone’s talking about AI, but the question isn’t if it’s transformative. It’s how we ensure it creates value for people.
- Focus on sustainable business models, not just valuation.
- Partnerships between traditional banks and fintechs are now mission-critical.
- The tone has shifted from “move fast” to “build smart.”
QUOTE: “AI isn’t the innovation itself, it’s the enabler that helps us deliver better experiences faster.”
How Starling Bank Is Scaling Human-Centered Banking
HOST:
Starling Bank has always positioned itself as both tech-forward and customer-first. How do you maintain that balance as you scale?
GUEST:
Starling’s philosophy has never been “tech for tech’s sake.” Every new capability we develop starts with a question: What problem does this solve for our customers?
We build everything in-house, which gives us agility, but also responsibility. That control allows us to experiment responsibly and move fast without breaking trust.
- Empathy-driven design: User experience is part of our product DNA.
- Data-informed decision-making: We don’t just collect data, we interpret it in context.
- End-to-end ownership: Our teams don’t hand off projects; they own them.
QUOTE: “Building in-house isn’t about control, it’s about accountability for every click, tap, and transaction.”
The Role of AI and Data in Next-Gen Banking
HOST:
Let’s talk about AI. How is Starling using it responsibly, especially in such a regulated space?
GUEST:
AI is at the core of what we’re doing, but with clear boundaries. It’s not about replacing people; it’s about enhancing decisions.
We use AI to analyze patterns that would be impossible for humans to spot, like real-time fraud detection or credit modeling—but the human remains in control. Transparency is key.
- AI applications must meet ethical, explainable, and regulatory standards.
- Automation helps us streamline operations, not eliminate human oversight.
- Data is treated as a trust asset, not a commodity.
QUOTE: “AI should make banking simpler and safer, not just smarter.”
From Data Overload to Strategic Clarity
HOST:
Many banks struggle with too much data but too little insight. How does Starling navigate that challenge?
GUEST:
That’s a real problem across the industry. The goal isn’t more data, it’s better data usage. We’ve built infrastructure that allows us to act on insights quickly.
Our teams are structured around outcomes, not functions. That helps us prioritize clarity over noise.
- Simplify reporting to focus on decisions that move the needle.
- Integrate systems for a single view of customer journeys.
- Align data strategy with business strategy, not just compliance.
QUOTE: “The future of banking belongs to organizations that can turn information into intuition.”
The Fintech Ecosystem: Collaboration as a Competitive Edge
HOST:
Do you think collaboration is still the future, or is fintech heading back toward competition?
GUEST:
Collaboration has evolved, it’s now about strategic complementarity. Banks and fintechs each bring strengths.
- Fintechs offer speed, creativity, and customer obsession.
- Banks bring scale, stability, and regulatory depth.
At Starling, we’re proof that blending both worlds creates something durable. The key is aligning purpose, not just profit.
QUOTE: “The future isn’t fintech vs. banks, it’s fintech with banks.”
Closing Thoughts: Designing for What’s Next
HOST:
What’s next for Starling and the broader fintech world?
GUEST:
The next wave of transformation will be invisible. When technology works perfectly, it disappears into the background.
We’ll see more focus on embedded finance, contextual intelligence, and personalized ecosystems. It’s less about apps—and more about experiences that anticipate needs.
QUOTE: “The best technology in banking will feel effortless, and that’s the hardest thing to build.”
Raw Transcript:
616 Money 2020 Take-aways & News from Starling Bank and Google
Welcome to Breaking Banks, the number one global fintech radio show and podcast. I’m Brett King. And I’m Jason Henricks.
Every week since 2013, we explore the personalities, startups, innovators, and industry players driving disruption in financial services. From incumbents to unicorns, and from cutting edge technology to the people using it to help create a more innovative, inclusive, and healthy financial future. I’m J.P. Nichols, and this is Breaking Banks.
Hello, fintech friends. I am fresh back from Las Vegas in May 2020. Actually, probably anything but fresh back from Las Vegas.
And all of my bank and fintech friends probably feel the same. But I am always happy that I leave Tuesday morning on that first flight out because 48 hours is more than enough. Couple things.
Later in this episode, I’m going to give you some quick thoughts and impressions of what I was hearing, seeing, and thinking about in Las Vegas. And then have an exclusive clip with Harriet Reese, the group CIO of Starling Bank out of the UK on an announcement that they made with Google from the stage there in a behind-the-scenes look at what they are doing using AI around fraud and their approach to doing it. And I really respect it because it is not a fully baked, well-ironed-out, down-to-the-five-nines of how they understand it will work.
But they recognize fraud has become such a big issue for so many of their customers. They are taking a test-and-learn approach to how they are rolling it out and continuing to evolve. So if you are looking for some deep insights on what happened in Vegas that is not staying in Vegas, if you are not already a subscriber to Alex Johnson’s Fintech Takes, he had a great post called 3-2-1 Money Time from Las Vegas that actually has some really great insights.
And coming out very shortly, one I love, Kia Haslett, formerly a bank director, also now is at Workweek with Alex. This was her first Money 2020. And I think that is really interesting to see because we have had this argument with a number of banks, as well as banking associations and some other large conferences, that they should all be attending things like Money 2020, Fintech Meetup, Finnovate.
And it always gets poo-pooed as like, oh, that is for the other banks or the Bass banks, the innovative banks. It is now mainstream that banks need to be thinking on a much bigger platform than just their local community. And it is not enough just to go to your local banker association.
It is not enough to go to the ABA community bankers. We are not saying do not go to those things anymore. It is just not enough.
You compete on a very different scale if you work on banking. It is equally important if you work in fintech and even if you are early stage, and I am not suggesting you lobby surf because Scarlett Sieber would come after me for that. But it is important that you pay attention to the news coming out because more than ever, we are seeing Me Too startups because the rise not only of SaaS tools that allow you to create startups more cheaply and products more cheaply than ever before, the idea of using some of these AI tools to code to be able to create a product on a shoestring means you need to pay more attention than ever because there is more competition than ever at the earliest stages in what is actually getting traction.
So a couple of observations. One, probably no surprise, stablecoins, agentic AI, and especially agentic commerce dominated the conversations. Now, I will, in full disclaimer, have been pooh-poohing stablecoin where the use case has largely been pointed to as cross-border, cross-border, cross-border that everyone wants to talk about.
In cross-border, when you actually factor in the full KYC, KYB on the front end and the back end, the efficiency of stablecoin in most applications on the transmission, which is the part immediately after the KYC and immediately before the release of the funds on a KYC, KYB, the improvement there, there is a lot of speculation it may actually be more expensive to use a stablecoin depending on gas fees and other things in between. I actually am going to change my tune a little bit that I think the first use cases of stablecoin and agentic commerce that we’re seeing that the time window between things that are actually only incremental improvements to the things that will actually be large improvements is going to be much shorter than ever before. This is also true on the agentic commerce side.
I saw and heard and recorded a bunch of pitches that most of which I was like, the value is very incremental. I’m not sure that there is enough there there to drive adoption. But even in the span of the weeks leading up to money 2020, in the time since then, my thinking is evolving because you’re already starting to see some new improvements, some incremental improvements.
I say, hey, is it a fundamental shift? Not yet. But it is beginning to see a wedge, an improvement that may actually make things different. Now, the part that I still want to go after is more than ever, there were a lot of announcements and proclamations from the stage at money 2020, most of which I kind of doubt the validity of.
And that’s a little disappointing because some of them are brands that I love talking about in the interest of being shown as a thought leader around stable coin or a thought leader around agentic commerce that are announcing things that, well, I frankly doubt the validity of and how massaged the data and the messaging was to be able to make those claims that the desire to be able to be staking out a place in the future is far outrunning actually the credibility that they have. Now, one that this is not true of, and this is why I love my conversation with Harriet, which is coming up right now, the group CIO from Starling Bank is, she’s very clear about what they have baked and what they don’t and the reasons behind that. So here, listen to Harriet and I from the floor of the Venetian at money 2020.
As always, if you liked this episode, drop us a like on your favorite podcast platform, whether that be Spotify, Apple, or who knows what. And thanks for listening. I’ll be honest, Harriet, I am sick of hearing about agentic AI.
Between stable coin and agentic, it has been too much. And we’re only in the first morning of money 2020. But what I find particularly interesting is what Starling is up to in your application of agentic AI.
And I think, you know, like all things technology related, it is typically the bad people that make the greatest adoption of it. But Starling is trying to get ahead of the curve. What did you guys announce from the stage today in the partnership related to it? Because I think that is one of the more interesting things that I’ve seen in Las Vegas so far.
Yeah, we’re super excited to be announcing our second customer facing application of generative AI. We’re working really closely with Google Cloud using their Gemini models. And today we launched scam intelligence.
The purpose of this tool is to help customers to identify and spot the signs of a scam in the app. We invite customers to upload images such as screenshots or pictures of marketplace listings and messages between sellers. They upload that right into the Starling app.
And then Gemini under the hood analyzes that image for indicators of a scam. So then right within the banking app, we tell customers things they can do to satisfy themselves that this is or isn’t a scam. And then they can carry on with their purchase if they’re comfortable.
So one of the things that I find personally difficult is changing user behavior. How are you addressing the fact that I think far too often users don’t take enough responsibility for their own actions and they will call it shop now and pay later, not in a financing perspective, but the consequences of those actions when they become victim to fraud don’t come to bear until much later. How are you addressing that? Yeah, and I think this is where at the moment Starling is really pioneering in using our prime real estate, which is our banking app, to put the invitation to customers to engage with these tools and features.
We saw this earlier this year with another feature that we launched, spending intelligence. And we’re doing it again with scam intelligence is utilizing the fact that customers are already in our banking app to make payments. And right there in that flow, offering them the opportunity to get this insight, learn how to spot the indicators and scam themselves, and frankly, make it an engaging experience.
I think for too long, banks have relied on customers doing that due diligence outside of their platform. And there’s a litany of reasons why that’s difficult or just too time consuming for customers, not least the fact that they don’t know who they can trust out there. Whereas when they’re in our bank, a banking app, we know they can trust us.
And therefore, we need to use that prime real estate to the good of the customer. Customer support in the world of financial services is hard. Between policy compliance, regulations, audit, security, and complicated customer queries, the stakes are high.
That’s where FIN, the best performing AI agent for financial services support, comes in. FIN handles high stakes performance and ensures policy compliance with no code controls. And if you’re wondering whether it really works, just ask the thousands of customers using FIN to provide end-to-end customer support every day.
To learn more about how FIN can handle the most complex financial queries and transform support teams, head to fin.ai breakingbanks. That’s fin.ai breakingbanks. One of the things I find interesting about Starling’s approach is, frankly, you don’t have it all figured out.
You’re not waiting to have it all figured out. You’re taking this test and learn approach. What are your early indicators? Can you unpack the insight that started taking you here, and where do you think you go next? Obviously, you can’t give the roadmap to the fraudsters about where you’re headed, but talk to us about how you got here.
Yeah, so Starling’s been innovating in the data science space now since I arrived, which is over seven years ago. You didn’t even actually, technically, have data before you arrived. Exactly, but I was ready to do the science on it.
I was there. But in that time, we’ve built up great confidence and understood the capability of these models. And let’s be frank, these models can’t do all things.
So we need to make sure before we unleash them on customers, we’ve really tried and tested what their capability really is and how confident we can be in them being able to do the thing we need them to do. And we did that by utilizing these models in our back office, in our internal processes. We’ve now got literally tens of production-scaled use cases across our operations center, where we’ve seen amazing efficiencies and amazing improvements to the quality of service that we’ve been able to deliver to customers.
8,000 hours served for our customer service agents, huge reductions in our detection of fraud in the back end. And it was using that experimentation approach and the ability that we could see how confident we were in the application of these models that we then had the confidence to move forward to putting it into the hands of customers. So I’d encourage everyone to use that experimentation in the back office, where you can put human in the loop, you can have enhanced oversight and enhanced monitoring, to really get confident that the models are performing at what you need it to do, and then move to bringing that into the hands of the customer.
So we’re talking about two powerhouses of both technology, customer trust, and advanced analytics working together on this. How did the relationship come to be with, you know, Gemini and the broader Google team? And frankly, was it a challenge to have two powerhouses kind of vying for how do we go do this? Yeah, so Starling and Google have had a relationship really since our inception. Google Cloud has been a great partner to us through their sort of workspace offering.
And so again, naturally, that was an easy place for us to start the experimentation with the capabilities of the Gemini model. All of our staff have access to Gemini, notebook LM, a suite of Google tools to use that as part of their day-to-day job. And then under the hood, we thought, okay, now is the time to experiment with the capability of the Gemini model in its rawest form.
What can this model really do? And that started with basic capabilities like the ability to summarize and synthesize text, and moved to what you’re seeing out there in the application today, the multimodal capabilities of Gemini, using it to analyze images and screenshots provided by customers. But from the Google side, we’ve really found the relationship to be very open and transparent. When we first embarked on this journey, it was really important to me as a bank CIO that I had full control and understanding of where that data was going, how our customer data was going to be used.
And we had a really open and frank conversation with Google about the confidence and the covenants that I would need within that contract to give me the confidence to experiment at scale. And Google were the first provider to be able to provide me with the confidence that I needed to move forward. So since that point, Google have been very encouraging, always going the extra mile to help us to scale these models.
And that’s the reason that we’ve seen such success today. So one of the big challenges in the U.S. is often regulatory pushback in that confidence and transparency you talk about don’t always work together. What is the FCA’s reaction, Ben, as you’re trying to approach this knowing full well it is not completely big, right? You are launching this knowing that it will be test and learn in your approach, but you couldn’t afford to wait.
What is the regulatory response, Ben? Yeah, so we’ve been engaging with the regulator since about 2019 on AI applications in industry. And I think that engaging with the regulator and being really open with them about our ambition here has been a really important part to them having confidence in the way that we’re approaching it. I chair a number of sort of AI initiatives across the industry and working with the Bank of England in particular.
I run a group called the AI Task Force. And through that group, we talk really openly about what the challenges are, what threats there are that are still unresolved on the AI front. But it’s the fact that I’m cognizant of those and we’re having that industry discussion, which gives the regulators confidence that in-house we are therefore thinking about how to mitigate those risks and how to keep an eye on challenges that are still unresolved.
So actually, from my perspective, regulators in the UK have been very encouraging to firms to experiment with AI. There’s a number of regulatory sandboxes that firms can use if they need to, if they don’t have that infrastructure in-house. At Starling, we use our own internal demo environment and as I mentioned, internal testing.
But definitely, there is the acceleration and excitement to move AI forward from the UK regulatory regime. Well, you are a very popular woman today, Harriet. Thank you for taking time to share what Starling is up to and looking forward to having this conversation again as you have more results.
Great. Thanks so much for having me. That’s it for another week of the world’s number one fintech podcast and radio show, Breaking Banks.
This episode was produced by our US-based production team, including producer Elizabeth Severance, audio engineer Kevin Hersham. If you like this episode, don’t forget to tweet it out or post it on your favorite social media or leave us a five-star review on iTunes, Google Podcasts, Facebook or wherever it is that you listen to our show. Those actions help other people find our podcast and in return, that helps us build an audience that can be supported by sponsorship so we can continue to provide you with our award-winning content every week.
Thanks again for joining us. We’ll see you on Breaking Banks next week.
