In conversation with Sam Everington

Read time: 2:30 Minutes

Sam Everington, CEO at Engine by Starling, attended a Club Stage Deep Dive on Agentic AI. He then stayed on to answer further questions.

In your panel “Deep Dive Agentic AI”, you mentioned that AI is making banks more human. What do you mean by that?

Sam Everington: Yeah, it’s from a few different angles. Ultimately, banks are looking at it from a cost and efficiency perspective and driving that efficiency leaves them more time and capacity to better serve customers.

We’re focusing on how to build tools that pick up on things that humans are missing. We have lots of humans carrying out lots of processes inside a bank and, as we all do, they make mistakes. Can the agent monitor what the human is doing? Can you invert that completely and have the human monitor what the agent is doing?

You can’t economically afford to have two people looking at every action in banks today, so you accept a certain level of error. Agentic AI means you can effectively get each action done twice and compare the outcomes.

So you’d say AI and humans are completing each other in that sense?

Everington: Working in support of each other, yes. Complementing each other. They pick up on things that the other party has missed. The AI will miss things, the human will miss things. The same way you get two independent humans to look at something, you find errors.

Talking about Starling Bank, do you have examples of how you’re using agentic AI at the moment?

Everington: We’ve got quite a few.

We have scam intelligence for example, which is a tool where you can upload screenshots or images related to a payment, and we give guidance on the risk factors. If you’re purchasing something online, we can tell you whether a deal looks too good to be true. The agent can flag that the price is unbelievable and probably a scam. If you have a WhatsApp conversation where someone is trying to persuade you to send money in a stressful scenario, it can highlight the risk factors and gives you advise on what to do.

We have an agentic assistant, the Starling assistant, which went live a couple of weeks ago. It’s a conversational interface in the banking app that has context on you, your transactions, your savings, your goals, and your card, and it can take action.

You can explain what outcome you’re trying to achieve, and it will guide you through the steps. That can be creating a new savings space, setting up a regular contribution, or if your card has been stolen. You might think you need to contact the bank, but it can freeze the card for you in the meantime.

Do you think the UK, given London’s role as a financial and fintech hub, is seeing greater adoption of agentic AI compared to Europe, or is it more balanced?

Everington: There are good examples across Europe, but the UK has been a big finance hub and a big technology hub for a long time now. It’s also been a very competitive banking market for a number of years, which isn’t true across all European countries.

Is competition driving more aggressive adoption of Agentic AI in the UK?

Everington: Yes, competition is driving the investment, which makes the capabilities available more broadly. As awareness builds, adoption is increasing. I don’t think people in any country are predisposed to act differently. It’s more that in a competitive environment, they are exposed to more and further developed products.