Dear readers,

This Wednesday, we’re looking at what it takes to build an AI ecosystem that goes beyond adopting the latest tools, from Bosnia and Herzegovina’s efforts to connect AI research with real-world applications to a fresh wave of funding for Romanian-founded AI startups. 

We also look at a Bulgaria-founded space manufacturing company raising $250 million to scale pharmaceutical production in microgravity, while regional founders increasingly turn to EU grants to fill some of the gaps left by traditional VCs. 

Plus, Zephyr Angels’ Igor Mishevski shares his take on where the smartest money is going in 2026, why real customer demand still matters more than AI hype, and what investors want to see before committing bigger checks. 

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

In 2013, Elmir Babović built a drone in a home workshop in Bosnia and Herzegovina. He had no laboratory, no institutional backing and no research budget. He cut aluminum profiles by hand, tracked down components that were difficult to find and wrote his own flight-controller firmware. He named the hexacopter Neretva, after the river that runs through Mostar.

More than a decade later, Babović is trying to solve a problem that goes well beyond building a machine. As managing director of Blum Institut, Bosnia and Herzegovina’s first institute dedicated to AI research and development, he wants to create the kind of environment that young researchers need to turn their ideas into technology that can compete internationally.

That ambition is also part of the thinking behind Kiss the Future, the AI summit returning to Sarajevo for its third edition on October 15 and 16. With a dedicated research track bringing together people working on AI, robotics and the commercialization of scientific research, the event is an attempt to connect parts of the ecosystem that too often operate independently.

For Babović, the challenge facing Bosnia and Herzegovina is not a shortage of intelligent people or promising ideas. It is what happens after somebody has one.

It is an argument shaped by his own experience, and one that helps explain why Blum Institut and Kiss the Future are trying to address a problem that cannot be solved simply by encouraging more people to use AI tools.

The problem with building alone

Babović's interest in robotics and intelligent systems predates the current AI boom. In 2011, he traveled to Singapore to present a research paper on metamorphic hyperspace in robotics at an international scientific conference. A friend paid for his plane ticket and conference fee. Two years later, he built Neretva, the homemade hexacopter that became an early example of his engineering ambitions.

These were not projects supported by a well-funded university laboratory or a mature research ecosystem. They were the work of someone determined to experiment despite the limited resources available to him.

The Neretva project required Babović to improvise at almost every step. He sourced the materials himself and developed the flight-control software around an 18F4550 microcontroller, manually timing pulse-width-modulation signals because the polished, ready-made flight-control solutions he needed were not available for his particular application.

The drone was reportedly the first made in Bosnia and Herzegovina. But the more important part of the story is what it illustrates about the environment in which ambitious technical projects emerge.

A talented engineer can sometimes overcome a lack of infrastructure to produce a prototype. That does not mean the same engineer can easily find the resources, collaborators, funding and institutional support needed to develop it further.

Babović sees this difference as one of the central weaknesses of Bosnia and Herzegovina's innovation ecosystem.

He compares the situation to football. A country can produce excellent individual players without necessarily having a team capable of winning a World Cup. Talent matters, but so do the institutions, coaching, infrastructure and coordination that allow individuals to perform together.

For technology, that means more than access to laptops and software development tools. It means research facilities, computing resources, reliable datasets, long-term financing and companies willing to invest in work that might take years to produce commercial results.

“We have strong individuals. What we are still missing are strong institutions and connections around them,” Babović says.

It is a problem that extends beyond Bosnia and Herzegovina. Across the Western Balkans, engineers and researchers work for leading international companies and universities, demonstrating that the region has no shortage of technical ability. The difficulty is creating conditions that allow more of that expertise to produce globally competitive products and research without requiring people to leave.

For smaller countries, Babović believes, regional cooperation is not merely useful. In some areas, it may be necessary.

Building the institution he never had

Blum Institut, established with the backing of Sarajevo-based software company Bloomteq, is an attempt to create some of the missing connections.

Babović leads the institute as managing director and heads research there and at Bloomteq. He also remains a professor at the Faculty of Information Technologies in Mostar, maintaining a connection between academic research and the practical demands of technology development.

The institute operates across two complementary tracks. One focuses on research that can be translated into commercial applications, while the other emphasizes scientific work and published research. The idea is to avoid the familiar separation between academia, where researchers develop knowledge, and industry, where companies need solutions that work under real-world conditions.

That bridge could make a difference because the journey from a promising experiment to a marketable product is rarely straightforward. A prototype might work in a controlled environment but fail when confronted with industrial conditions, inconsistent data or the practical requirements of a customer.

An institute can provide researchers with access to equipment, experienced colleagues and real problems to solve, while giving companies a way to explore technologies that they might not have the capacity to develop independently.

One of Blum Institut's projects illustrates this practical approach. Its computer vision system for inspecting solar panel installations is designed to identify faults from photographs, reducing the need for technicians to inspect each panel manually.

It is not the sort of AI application that dominates headlines about artificial general intelligence or humanoid robots. It is, however, a concrete example of how research can address a problem in an existing industry.

The institute also wants to give younger researchers a place to develop their own ideas with proper guidance. Babović calls it “a sanctuary for young people,” particularly master's and doctoral students who need mentorship, research infrastructure and the opportunity to work on projects that extend beyond academic exercises.

The phrase carries an obvious connection to his own experience. In 2011, he needed a friend to help finance an international conference trip. In 2013, he was building a drone in a workshop without institutional support. The goal now is to ensure that the next generation has somewhere to go after building its first prototype.

“The difference is what happens after the garage. A mature innovation ecosystem gives that person somewhere to go next.” he points out. 

Why Kiss the Future needs a research track

This is where Kiss the Future becomes more than another gathering devoted to artificial intelligence. 

The Sarajevo summit's third edition is built around the idea that AI development requires more than companies adopting existing tools or executives discussing how the technology might improve productivity. Researchers, engineers, investors and businesses need opportunities to work together, particularly when the objective is to develop new technology rather than simply implement something that already exists.

The summit's Research Track is intended to address that gap. While other parts of the event focus on business applications and enterprise deployment, the research program concentrates on fundamental and applied AI work that has not necessarily reached the market.

Two hands-on workshops offer a glimpse of the approach. Florin-Catalin Chici, head of startup acceleration at Silicon Allee, the deep-tech venture lab embedded within Fraunhofer HHI in Berlin, will lead a session titled “From Research to Market.”

The workshop addresses a familiar problem for researchers: how to move from a scientific result to a company capable of developing and selling a product.

Silicon Allee's model connects founders with Fraunhofer HHI's patents, laboratories and researchers, alongside funding for up to 12 months. Rather than charging licensing fees, the venture lab uses a virtual equity arrangement that pays out if a startup eventually exits.

The structure offers a practical example of how scientific research can be connected to entrepreneurship. It also reflects the type of institutional support Babović believes is missing in Bosnia and Herzegovina: a mechanism through which a promising idea can access resources and survive the transition from research to commercialization.

The second workshop, “Physical AI – Robotics,” takes a different approach to the same challenge.

Led by Nikodem Bartnik, an engineer on the LeRobot team at Hugging Face, the session focuses on the relationship between AI models and physical machines. The open-source LeRobot toolkit allows researchers to teleoperate robots, record demonstrations, train models to imitate those actions and deploy the resulting models on hardware. Furthermore, Bartnik has also developed tutorials for low-cost robotic platforms and built a YouTube audience around accessible robotics and DIY electronics.

Why AI research needs long-term investment

For young researchers in the Western Balkans, that combination can offer something more valuable than another general introduction to generative AI - an exposure to the technical work involved in building intelligent systems and the institutional structures needed to turn that work into something sustainable.

Babović wants researchers to understand not only how to use AI, but also how intelligent systems are designed, trained, tested and developed. 

There is, of course, a limit to what Blum Institut or a two-day summit can achieve. Bosnia and Herzegovina still faces challenges involving research funding, computing infrastructure, access to datasets, technology transfer and long-term institutional support. These are structural problems that cannot be resolved by a single organization or a series of conferences.

The Bosnian entrepreneur is explicit about that limitation. “A research institute can help connect some of the missing pieces, but it cannot replace an ecosystem,” he says.

He argues that universities need stronger research capacity, companies need incentives to invest in research and development, governments must treat science and innovation as strategic infrastructure, and investors need to become more comfortable with technologies that may take years to mature.

That last point is particularly important for deep tech. Unlike a software product that can be launched quickly using existing infrastructure, advanced AI and robotics projects may require extensive testing, specialized equipment and sustained research before they become commercially viable.

Short-term funding and isolated projects are unlikely to create the conditions needed for that work to flourish. Competitive AI ecosystems, Babović argues, require five- or ten-year commitments to research groups, doctoral programs, laboratories and companies prepared to take technological risks.

Thus, Blum Institut's ambition is to demonstrate what is possible while helping establish some of those commitments. In that manner, Kiss the Future provides a meeting point where researchers, industry representatives and people working on commercialization can exchange ideas and potentially find collaborators.

Whether the summit can contribute to lasting change will depend on what happens after its participants leave Sarajevo. Babović's own trajectory offers a useful reminder of why that next step is important: building a drone without a laboratory was possible, but developing an ecosystem in which hundreds of other researchers can take their work further is a considerably harder task.

IT Logs: Where do you think the smartest money is 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?