Dear readers,

This week, we look at one of the biggest questions shaping the AI era: how do we know what to trust online? As synthetic voices, deepfakes, and AI-generated content become more convincing by the day, a new wave of startups is emerging to solve the trust problem before it spirals further - and of the most promising ones comes from the region.

Therefore, we bring you the story of Montenegro-based startup DeepMark, which is taking on one of AI’s most pressing challenges, proving that quality deep tech can emerge far beyond Silicon Valley. In our Investor Take, where we unpack what is really driving startup valuations in 2026, from AI momentum and founder quality to the signals investors are watching most closely.

Our goal with each edition remains the same: we are here to help you better understand not just what is happening in tech, but why it matters, where the opportunities are emerging, and which trends are worth paying attention to before everyone else catches on.

Happy reading!
Bojan Stojkovski
Editor-in-Chief, IT Logs

The first thing Slavko Kovačević says about himself is not what you would expect from the founder of one of the Balkans’ most promising AI security startups.

“Accomplishment number one, being born aside, I’m CEO of DeepMark.” he tells the audience during a panel at the SEE AI Summit in Novi Sad.

It is funny, slightly awkward, and also revealing. The Montenegrin founder does not perform confidence in the Silicon Valley sense - he is thoughtful, direct, and unusually honest about mistakes, weaknesses, and the uncomfortable realities of building a company in one of the world’s most difficult categories: deep tech.

That becomes important once the conversation turns to what DeepMark is actually trying to solve. As generative AI becomes better, cheaper, and easier to access, the internet is entering a new trust crisis. Images can be manipulated at scale, videos can be synthesized, and perhaps most unsettling of all, voices can now be cloned with never before seen accuracy. 

The question is no longer whether synthetic content exists. It is whether ordinary people, businesses, or institutions will be able to reliably distinguish authentic media from fabricated content.

That philosophy sits at the heart of DeepMark, which has ben developing AI-driven watermarking technology for digital assets such as audio, video, and images.

From research to real-world trust

Kovačević likes to explain DeepMark’s product using a metaphor that immediately clicks. “We call this solution a blue checkmark for voice,” he says. “Basically, we are creating an identity for voice. Voice used to be an identity, but now we can’t trust voice anymore because everyone can clone a voice.”

For decades, hearing someone’s voice carried implicit trust. Voice acted as a natural biometric signature, something personal and difficult to replicate. AI has fundamentally broken that assumption. In a world where a scammer, a bad actor, or even an AI agent can sound exactly like someone you know, trust becomes something that can no longer rely on instinct alone.

DeepMark’s answer emerged not from market hype but from years of academic research. Kovačević and his co-founders were early pioneers in deep learning-based audio watermarking, long before synthetic media became a mainstream concern. 

Recognition from global tech giants validated the science, but it also created frustration. The research was valuable, influential even, yet remained trapped in papers and citations instead of solving real-world problems. “You have this gem, this miracle, and everyone is citing it, but no one is actually using it. We invested years into it and I wanted that technology, two or three years of my life, not to go to waste.” he tells IT Logs.

That frustration became the seed for DeepMark. Unlike the romanticized startup stories often repeated in tech circles, Kovačević is refreshingly unsentimental about how he became a founder. “To be honest, I didn’t decide to switch to entrepreneurship, I decided to leave university,” he says. “The cool startup story would be I got investment and left my job, but that’s not what happened. I just didn’t want to stay at university. I wanted a change.” 

DeepMark arrived at the right moment, as it gave him both technical purpose and personal reinvention.

Stop waiting for opportunities and go after them

“My whole life, I was waiting for opportunities to come,” he says with his characteristic self-deprecating humor. “I did get some, probably more than many people my age, and I created some through hard work, but I was not creating them by pursuing them. Other people were offering them.” Slavko explains.

He says that same mindset followed him into startup life and nearly slowed DeepMark’s momentum at a critical stage. “We made the same mistake with DeepMark. We were kind of waiting for some time. Then I decided we shouldn’t wait anymore and we completely changed our mentality.” he explains.

For founders in South Eastern Europe, networks often matter more than capital in the earliest stages. Trust compounds fast in smaller ecosystems.

Diaspora connections have been equally important. Kovačević believes professionals from the Balkans who built careers abroad often feel an obligation to help those who stayed. 

“They generally want to help people here because they had the same problems we have now,” he says. “They left to find happiness and results, and they want us not to do the same thing.” 

That motivation shapes DeepMark’s broader purpose beyond AI security. “One motivation I don’t tell investors is that we want to create a safe haven for smart, intelligent, hardworking people to stay and do deep tech.”

Building deep tech from the Balkans

For Kovačević, talent is not the Balkans’ weakness - retention is. “The talent is in the Balkans,” he says. “One thing Communist Yugoslavia did well was education, especially STEM. It’s not getting better, obviously. It’s getting worse. But it still holds.” 

“All our employees studied electrical engineering and were the best students in their generations,” he says. “We have offices in both Belgrade and Montenegro and we got top talent in the region.”

Still, talent alone is not enough. Commercial execution creates a different set of problems, and Kovačević is brutally honest about mistakes. One of the biggest was hiring sales too early. “I hired a guy to do sales, and don’t get me wrong, he’s a great salesman, a thousand times better than me at sales as a craft,” he says. 

Another hard lesson involved trust. “My life philosophy is that everyone is good, and I’m not going to change that,” he says. “But you should also be careful.” Startup life repeatedly challenged that worldview. “A lot of people tried taking advantage of us, both in this region and abroad.” Slavko sighs.

Competing with giants

DeepMark is building in a category crowded by some of the world’s largest players. Kovačević does not pretend otherwise. “Our biggest competitor is actually Google,” he says, referring to Google and its synthetic media verification efforts through SynthID.

That reality shapes DeepMark’s market strategy. “I’m not targeting the giants,” he says. “I’m targeting everyone else.” The logic is pragmatic, as hyperscalers prioritize large enterprise accounts and global infrastructure deals. 

Smaller players often remain underserved, especially when solutions require privacy-sensitive deployment. “For this technology to work, it has to be a trade secret. It has to be on-prem. You can’t open-source everything because this is trust security and if it leaks, it can be misused.”

“Engineering is hard, yes, but if someone already did something, you can usually achieve it too. With research, that doesn’t apply. It’s a gamble. You start with a goal, but you cannot be certain you’ll achieve it.” he exlains.

That uncertainty is why capital matters so much in deep tech. Last year, DeepMark secured €1.2M in a round led by from Czech VC Lighthouse Ventures, angel investors, and grants from Montenegro’s innovation ecosystem. Funding buys time, arguably the most precious resource for research-heavy startups.

That framing got attention, but once the technical teams joined, the conversation changed. “They didn’t care about ElevenLabs being first. They cared only about one thing: how good is the algorithm, and is it going to work or not.”

That notion may define DeepMark better than anything else - vision opens doors, but technical credibility keeps them open. In a world increasingly flooded with synthetic content, the next layer of digital infrastructure may not be about generating more content, but about verifying what deserves trust. 

And Kovačević summarizes that future in a single sentence that feels almost philosophical in its simplicity - “We don’t want to guess.”

On July 16th at 1PM ET, beehiiv is going live with a look at the future of publishing, audience growth, and digital business.

What started as a newsletter platform has evolved into something much bigger: a place where creators and brands can grow, monetize, and own their audiences without stitching together half the internet to make it work.

The next chapter starts live at the Summer Release Event. 

Join us to see what’s coming next.

RSVP now.

IT Logs: Where do you think the smartest money are going this year?

IT Logs: Bigger bets on fewer startups, or smaller checks spread wider?

IT Logs: What is moving valuations more in 2026: real traction, AI buzz, or solid unit economics?