Why AI-native banking is a step-change, not a rebranding exercise

In many ways, AI is the digital equivalent of coriander. Or FC Bayern Munich. People are typically either wildly enthusiastic about its possibilities, or dead-set against it.

There's rarely a middle ground.

Banking is no different.

In 2026, banks are allocating up to 5% of their overall budgets to AI.

Meanwhile, our research found that consumers are, at best, ambivalent. They use AI. They can see and appreciate its benefits. But they're worried about it, too. And one of their biggest worries is that its proliferation will mean losing access to human support.

Ecem - catchy CTA as discussed

Given these findings, it might seem counter-intuitive for Solaris to have decided to go all in and become Europe's first AI-native bank.

But AI-native doesn't mean automating everything as much as (in)humanly possible. It means using the most powerful technological tools we've ever had at our disposal in the ways that make most sense.

That is, in ways that will make life easier for banking staff and the banking experience better for customers.

AI-native vs AI-first: A subtle but important difference

So what does AI-native actually mean? And how is it different (or better) than being AI-first — an increasingly common claim across a growing number of sectors?

The distinction is structural.

AI-first operates at the surface level. Chatbots that handle customer queries, for instance. Or AI assistants bolted-on to an existing user journey.

AI-native, on the other hand, means embedding AI into the infrastructure and, so, making it integral to how the bank's processes are designed, governed, and run.

Think of it this way. Say you stepped into a dark room.

In an AI-first scenario, AI turns on the light for you, even when you'd rather turn on the light yourself.

In an AI-native scenario, you turn on the light, while AI handles what goes on in the background. Its job is to ensure that, when you flick the switch, the light turns on quickly, reliably, and safely.

Back to banking, AI-native means AI isn't a product feature — bells and whistles that, all too often, customers didn't ask for and might not even appreciate — but integral to the foundations.

Crucially, AI-native capabilities are built around customer needs and compliance requirements, and not the other way round.

Deconstructing AI-native banking

So far, so good. But what does AI-native banking look like in practice?

To answer this question, we need to take a brief detour into the evolution of banking.

In the beginning, banks were a simple solution to a relatively simple problem: keeping valuables safe. This meant that, for most of their history, bank accounts were, by and large, a luxury for the wealthy. It was only after WWII that they went mainstream.

Today, bank accounts are essential, and those without are a small minority — about 4% of Europeans, according to the latest available numbers. This proliferation, and initiatives such as the EU's basic bank account, have led to standardization. A bank account is a bank account is a bank account, regardless of which brand logo is on the statement.

With the product no longer the key differentiator, the execution makes all the difference. Whoever can deliver faster, cheaper, more securely, and more usefully for the customer gains the advantage.

To date, banks have sought to improve execution through efficiency gains, a broader feature-set, and their internal culture. Being AI-native goes beyond this, turning the focus onto process organization — how the work itself is done. The processes, workflows, checks, and controls that sit behind customer interactions.

The AI ladder

Process organization demands a rigorous approach. You can't shoehorn existing processes into an AI model and call it a success. The starting point shouldn't be to look for areas where you can add AI, but to ask:

Is a specific process capable of being automated?

And, if so, would automating it with AI be a genuine improvement on the current state? These aren't strict yes/no questions. Arriving at a satisfactory answer requires a structured approach.

For me, a three-step framework has always worked: documentation, standardization, and, only then (and only if warranted), automation.

1. Documentation

How many steps does the process involve? What are they? Where are the decision points? It's only once these are clearly laid out that you can identify areas where you can improve and evaluate if it's possible to automate.

As business theorist and scholar W. Edward Derning once put it: "If you can't describe what you're doing as a process, you don't know what you're doing."

2. Standardization

One of AI's biggest strengths is its ability to make nuanced decisions. But that doesn't mean it should be allowed to act on a case-by-case basis in every single instance. Standardization matters because it ensures predictability and reproducibility, which are essential in a tightly regulated sector like banking.

Standardizing requires the establishment of a baseline and benchmarks. What does "good" look like? How does the process unfold in different scenarios?

Just as important, is the process entirely digital, or do key steps take place offline, for example using email or spreadsheets?

3. Automation

A cornerstone of Solaris' approach is that automation doesn't necessarily mean: "Let's add AI." The latter is (and should be) one tool, not the entire toolbox.

A good rule of thumb is that run-of-the-mill, repeatable tasks should be automated, while complex, emotionally charged cases should be human-led, with AI in a supporting role.

Consider card-blocking. A customer who has misplaced their card doesn't need to call up their bank and wait in a queue while somebody blocks it. Nor should they be required to talk to a chatbot. A simple on/off switch within their banking app is more than enough.

A fraud case, on the other hand, is very different. The customer will be distressed and require reassurance, so being able to speak to a human quickly makes all the difference.

At the same time, AI can make the experience less distressing by handling the admin — collecting information, pre-filling forms, and offering guidance to both customers and staff.

Our research supports this approach. The 4,000 German respondents we interviewed for our white paper AI-driven financial services: Why trust and human expertise are essential for success overwhelmingly said they want to feel empowered and in control.

AI-native banking must be customer-first

The high levels of trust they enjoy and robust regulation put EU banks in a uniquely advantageous position when it comes to AI.

But, in order to make the most of the opportunity, they cannot implement AI for AI's sake. A successful AI strategy requires intentionality. Automation decisions based on what value is added, not efficiency alone. Humans taking charge when warranted, even in situations that could be automated. Customers in control instead of feeling like AI is an imposition (and, all too often, a step down in service quality).

This is what Solaris is striving for as an AI-native bank.

You wouldn't use a hammer when what's really required is a screwdriver. Similarly, being AI-native means recognizing that there's a time and place for AI and that, ultimately, it's the outcome and the impact that will have on the customer which should determine what that time and place are.