In the last decade or so, almost every financial interaction we make has moved from big physical branches to a small, tiny screen.

Be it your own phone or laptop, you can really do anything and everything. You can open a bank account, sign up for a lender, send money to a friend, or pay for a delivery on that screen of yours.

Now, your entire financial journey happens through a digital front door.

But the same door that customers usually walk through also happens to be the very door criminals use, too, often at the same moment.

That is the backdrop behind SEON’s philosophy, and it is also where the company’s President of GTM, Matt DeLauro, begins when he talks about the problem with how most institutions are set up today.

The issue, he says, is structural. It has nothing to do with budgets or software. According to Matt, it comes down to the long-standing split between fraud teams and anti-money laundering teams.

A split that made perfect sense when banks lived in the physical world, but now creates something much worse than inefficiency. It creates blind spots.

“Fraud is actually a predicate crime. There’s no money laundering without fraud,” he explains. “[Hence], if you have two different teams looking at the same problem from different angles, they miss a lot of the context.”

What Matt is trying to say is that criminals do not think in silos. But institutions, well, they still do.

Why Splitting Fraud and AML No Longer Works

In the old model, people walked into a branch with physical documents. Fraud and money laundering were separated because the risks occurred in different parts of the customer journey.

Digital environments, however, are a bit different and do not necessarily work that way.

Everyone enters through the same website or app, and every criminal uses the same devices, IP addresses, and behavioural tricks to impersonate or manipulate the system.

AML teams traditionally monitor the flow of funds, while fraud teams monitor intent, behaviour, and the legitimacy of someone trying to create or access an account.

Those two worlds should inform each other, but in reality, they usually sit in different reporting lines, use different tools, and view different datasets.

Matt puts it bluntly.

“AML teams rarely have access to the signals fraud teams see. Location, IP, device, email, phone. Even when they do, they have to [kind of] beg, borrow, and steal for that data.”

That disconnect is exactly where criminals operate, and rather efficiently, I must say.

Synthetic identities slip through gaps between teams. Transaction monitoring flags alerts that fraud analysts have context for, but AML officers, they never seem to see them.

The outcome is predictable. More work, less certainty, and long investigation queues that stretch for months.

For SEON, the fix begins with unifying the intelligence layer. And increasingly, that means turning to AI.

But not the kind of AI that creates a new black box.

The Industry Has Enough Black Boxes

AI has become one of the most overused words in financial services. Every pitch deck, conference booth, and vendor website puts it front and centre.

Yet most institutions still struggle to understand how their AI systems make decisions.

That is now a regulatory issue, especially in markets with strict reporting obligations.

Regulators expect financial institutions to justify why they made or did not make a report. That means knowing how the model evaluated a case, what factors influenced the outcome, and whether bias or error might have contributed.

Matt breaks down the problem in a way most compliance teams recognise. Models that promise accuracy often require months of training before they deliver value.

And secondly, models that deliver day one value often cannot adapt quickly enough to new fraud patterns. Neither solves the real-world pressure that banks and fintechs face.

Matt stresses that either the AML or the fraud team has the capacity to wait ten months just to set up a solution. Their job is to stop fraud on day one, ASAP.

“[So], that’s why we built both, the rules for immediate protection, and the algorithms for long-term precision,” he said.

The hybrid approach is the key feature of SEON’s platform.

The company offers a white-box rules engine that teams can configure instantly, combined with algorithms that learn subtle patterns across millions of data points. To make this usable, SEON recently introduced a natural language rule builder.

Analysts can write a sentence the way they would explain a risk scenario to a colleague, and the system turns it into a rule.

It gives investigators a clear view into how decisions are made, while also increasing the speed at which new threats can be mitigated.

APAC’s Synthetic Identity Crisis Needs a Different Kind of Intelligence

One of the clearest examples of why traditional tools fall short is synthetic identity fraud, and the problem is particularly severe across APAC.

Here, it is harder to detect, harder to trace, and harder to prevent through legacy checks that rely heavily on government databases. Matt does not sugarcoat it.

“Your government ID numbers are already out there on the dark web. Definitely mine, probably yours,” he said in a jokingly manner.

Fraudsters have learned that the fastest way into a financial system is to pair a valid identification number with completely new digital credentials.

A fresh email address. A temporary phone number. A device that cannot be traced back to an earlier account.

Matt explains that a typical synthetic identity scheme involves taking a legitimate ID and pairing it with a fresh email or disposable phone number.

By doing so, it gives the fraudster full control while the system assumes the identity belongs to the real person.

And to make things a tad bit worrying is that traditional KYC systems validate just the ID itself, not the digital behaviour around it.

What this means is that if the number is real, the document looks legitimate, and the face matches, the system typically allows the onboarding to continue. But the identifiers that criminals create are often too new or too shallow to be real.

This is where SEON’s approach to digital footprints starts to matter.

Rather than asking whether a phone number simply exists in a static database, the system looks for signs of life across the broader digital world.

It checks whether an email or mobile number has been active on everyday platforms such as Grab, WeChat, or WhatsApp, and whether its activity resembles the natural patterns of a real user rather than something freshly created for a crime.

As Matt puts it, it would be unusual for someone to apply for a digital bank account yet have no presence on apps that are practically essential in the region.

That wider footprint has become one of the most reliable early markers of synthetic identity fraud, especially in APAC’s mobile-first markets.

It also shows why institutions need tighter and smarter security controls in the first place.

Stopping Fraud Without Stopping Growth

But the problem is that tighter controls usually mean more friction for legitimate customers.

Security controls often come at the expense of user experience. The safer an onboarding flow becomes, the more hoops a customer has to jump through. That is the trade-off most companies assume they must accept.

Matt, however, argues the opposite.

“Legacy tools introduce a lot of friction. What we offer is a frictionless surface.”

His point is that most risks can be evaluated without interrupting a user. SEON’s SDK and APIs collect behavioural biometric signals as the user interacts with the app.

The system captures typing patterns, device orientation, IP address consistency, whether the device is jailbroken, and whether it is hiding behind a residential proxy.

All of this happens in the background, with no additional steps for the customer. The risk engine then decides whether to escalate, flag, or green-light the onboarding.

In a region as diverse as APAC, where a user in Jakarta behaves very differently from a user in Sydney, this passive, contextual approach is often far more accurate than rigid verification steps.

It also avoids the pitfall of rejecting genuine customers simply because they behave differently from a predefined “normal.”

Regulators Are Moving Faster Than Systems Can Keep Up

The pressure on compliance teams has increased sharply. One example is Singapore’s MAS rule that gives institutions only five days to file a suspicious activity report from the moment they detect something suspicious.

Anyone who has ever written a SAR knows this is tight.

“Timing is the most difficult part of running a compliance team,” Matt says. “A lot of teams are six or eight months behind on investigations.”

Most of that delay comes from narrative creation. A SAR is not a checkbox but is more of a detailed report that describes the behaviour, the transactions, the risks, and the rationale behind the suspicion.

Investigators often spend hours drafting a narrative, pulling together evidence, and formatting the final submission. SEON now uses large language models to take on most of that heavy lifting.

Instead of starting from a blank page, the system produces a near-complete draft that the investigator only needs to review and refine, cutting the workload down dramatically.

Matt says that the efficiency gains are huge.

“A five-month backlog can be reduced to 30 days,” he said.

For teams that face regulatory deadlines, this kind of workflow automation is the difference between staying compliant and drowning under case volume.

One Command Centre, Not a Spaghetti Bowl

With so many fraud, KYC, and AML vendors in the market, it is reasonable to ask what truly distinguishes SEON. When Matt explains how clients describe their setup, the answer becomes clear.

“Most clients today live in a spaghetti mess of silos and disconnected systems,” he says. “We offer a unified command centre. A single source of truth. And we can be integrated in one to two weeks.”

The appeal is obvious. Instead of juggling five or six systems across different parts of the customer journey, institutions get one place where fraud and AML signals converge.

One dashboard. One policy layer. One investigation flow.

This is the platform approach many banks are now trying to build internally, but rarely manage to stitch together effectively.

SEON began life as the disruptor to slow, legacy systems. The company is now larger, better funded, and operating with enterprise-level clients. So how does a company evolve without losing its original agility?

Matt believes the answer is simple.

“The crown jewels of SEON are that we’re easy to work with and easy to integrate. We think of ourselves as the Stripe of fraud and compliance.”

To protect that identity, SEON’s leadership team spends a surprising amount of time speaking directly to new customers, asking about their onboarding experience and where friction still exists.

“A lot of companies become successful and forget what got them there,” he says.

By anchoring the company culture around developer experience, transparency, and speed, SEON hopes to avoid becoming the legacy system it once sought to replace.

The Fight Needs Both Hands

Matt ends the interview with a simple observation. Banks and fintechs cannot afford to fight financial crime with one hand tied behind their backs. They need to remember that fraud and money laundering are deeply connected.

Teams, tools, and workflows that treat them as separate will always be slower than the criminals they are trying to stop.

SEON’s bet is that unifying these systems is not just more efficient. It is necessary.

And in a region as diverse and fast-moving as APAC, where synthetic identities are growing more sophisticated and regulatory timelines are tightening, that unified approach may become the new baseline rather than the exception.

The digital economy is expanding at a pace no one can fully track. And fraud, well, they are evolving just as quickly. What companies build today will define how they protect customers tomorrow.

Featured image: Edited by Fintech News Singapore based on images by ismode via Freepik and SEON.