Now He Builds AI That Classifies Phishing Attacks Before They Reach Your Inbox.

A conversation with: Sam Gelbart, Head of Innovation at SYNAQ

Most people who end up in cybersecurity planned to be there.

Sam didn’t set out to build the intelligence layer behind one of South Africa’s largest email security platforms.

When SYNAQ’s founders first approached him, he was an IT Manager in medical aid administration — present in the room not as a principal, but as the technical voice someone suggested to be there. There was no grand plan. No calculated pivot into cybersecurity.

What followed was two decades of staying curious, staying committed, and going deeper. Today, Sam leads the engineering behind the machine learning models that determine, in real time, whether an email is a legitimate business communication or a threat — across more than a billion messages a year.

Here’s how that happened.

“I knew these guys were going to build something interesting”

 

When SYNAQ was founded in 2003, Sam had a front-row seat — not as an employee, but as someone who could see the founders clearly.

“I was genuinely impressed with David for his technical capability, and Yossi for his business acumen and creative deal-making. I knew these guys were going to build some very interesting and cool products, and I loved the way they were approaching it.”

That early read on the founders’ potential was what made the eventual invitation to join — and run product development — an easy yes.

That invitation put him inside the architecture of a platform that was, at the time, South Africa’s first cloud-based email security system. What followed was two decades of building from the inside.

 

Here is what he learned that you can’t learn anywhere else

On constraints:

“Not having all the resources in the world forces you to come up with novel and innovative ways to solve technology and scaling problems. Constraints matter — they make you think differently.”

On data:

“What I love about data is that when it’s used well, it tells the absolute truth. Having a sound data strategy is critical in any business, and translating data into actionable insight is a core part of what I do.”

On AI — and the discipline required to navigate it:

“We live in a world where AI-driven innovation is shipping almost hourly. To navigate that deluge, you have to step back and evaluate what’s stable, secure, and proven — because a lot of what’s flavour of the month today gets superseded tomorrow.”

That grounded scepticism shapes how SYNAQ approaches innovation: embrace what genuinely accelerates problem-solving, stay rigorous about what actually belongs in a security platform.

The Moment Everything Changed

The turning point, Sam says, was specific and immediate.

“Pretty much the moment Yossi and David asked me to join and run product development. Building out the next evolution of the SecureMail platform from day one meant I was going to be delving into software development at a level I could only have dreamed of.”

 

The Decision That Defined Everything

Of all the steps in Sam’s journey at SYNAQ, one stands apart.

“Taking on responsibility for learning traditional machine learning skills. That opened my eyes to what ML and AI can truly achieve.”

The proof of concept was a predictive model built to classify email URLs as either phishing or benign. It sounds straightforward. It wasn’t.

And the insight it produced went beyond the model itself:

“Most security problems are fundamentally data-driven and can be modelled to make our customers safer and our products better.”

That principle now runs through everything SYNAQ builds. It’s what sits behind the 3,500 individual security checks that run on every email through the platform — in seconds, every time.

Building for the Customer, Not Just the System

What’s stayed consistent across two decades, Sam says, is a deliberate focus on the person on the other side of the product.

“While the work is deeply technical, I’ve learned to focus not just on the systems we build but on how our customers actually use them. That drive to make our products user-friendly, effective, and efficient is a part of my job I really enjoy.”

It’s a product philosophy that’s easy to state and genuinely difficult to maintain when you’re operating at the intersection of AI, machine learning, and enterprise security infrastructure.

 

What This Looks Like in Practice

The platform Sam has spent two decades building now:

• Processes over 1 billion emails per year

• Runs 3,500 security checks per email

• Carries the only 100% punitive phishing SLA in South Africa — meaning SYNAQ bears the financial consequence if a qualifying phishing email gets through

That kind of commitment doesn’t come from confidence in marketing language. It comes from confidence in the models.

One More Thing Sam Said

When asked how working at SYNAQ has shifted his thinking about technology broadly, his answer wasn’t about infrastructure or threat intelligence.

It was about creative energy.

“Tools like Claude have massively shifted our team’s mindset when it comes to tackling difficult problems. Anything feels possible, and that creates an environment filled with creative energy.”

That’s not a small thing to say inside a company whose entire job is to stay a step ahead of people trying to break in.

Interested in what a security platform built on two decades of South African-specific machine learning looks like for your organisation?

synaq.com | sales@synaq.com

SYNAQ has protected South African email infrastructure since 2004 — built here, run here, accountable here.