In partnership with
Dear readers,
This week we are taking you inside one of the most important shifts happening in tech right now: the moment when AI stops being just a tool and starts becoming a worker.
At the center of this edition is a deeply reported story about Daytona, the $24M-backed Croatian startup that is building what might be the first real “computers for AI agents”. It’s also the story of Ivan Burazin - from early cloud IDE pioneer and founder of one of the biggest developer conferences in the SEE region, to Infobip executive, to now quietly helping define what the next generation of software infrastructure will look like when millions of autonomous agents need somewhere to actually exist, run, fail, retry, and scale.
Additionally, we also track the pulse of the regional ecosystem and this time we feature a founder take from Thea Gherdan, CEO of Proon-Tech, on what will truly matter in 2026: the balance between speed and depth, the rise of AI-first startups, and how companies can build a sustainable impact in today’s shifting ecosystem.
Finally, we add a layer of intrigue, touching on whispered power shifts in Bulgaria and the unresolved legend of one of N. Macedonia’s most mysterious tech anti-heroes.
Enjoy,
Bojan Stojkovski
Editor-in-Chief, IT Logs
The $24M-backed Croatian startup that’s giving AI agents their own computers
The team behind Daytona
I first met Ivan Burazin in Zadar in 2021, at a point when, by every conventional measure, he had already “made it”.
Shift, the developer conference he had built from scratch - with borrowed stages, wild afterparties, and a strong belief that Southeast Europe deserved a real tech community - had become part of Croatia’s first unicorn, Infobip. Ivan now held the title of Chief Developer Experience Officer and had a seat at one of the region’s most successful technology companies.
But he was not behaving like someone who had reached the end of the road, and rather, was talking about computers. He was talking about how developers still spent half their time fighting their environments. About how the cloud, for all its sophistication, had never really solved the basic problem of where work actually happens. And, more quietly, he was talking about software that no longer waited for instructions.
Back then, AI agents were still mostly an experiment. But Ivan was already thinking about what would happen when they stopped being tools and started becoming workers - which actually was the beginning of the startup that would go on to co-found a few years later, Daytona.
From cloud IDEs to developer culture
Ivan’s career has always followed a strange logic: he builds the thing that seems unnecessary right up until the moment it becomes inevitable.
In 2009, he launched PHPanywhere, allowing developers to write and run code in a browser long before anyone else took the idea seriously. It evolved into Codeanywhere, which grew into one of the world’s most widely used cloud development platforms. A decade later, GitHub would call the same idea Codespaces and sell it as the future of software development.
But Ivan didn’t stop at software. In 2012, he founded Shift, which became the most important developer conference in the region. The conference also served as proof that a region better known for outsourcing could produce original technology, culture, and ambition.
Infobip’s acquisition of Shift in 2021 turned Ivan into a corporate executive. However, it also gave him something far more valuable: a view inside a truly global engineering machine. There, Ivan watched thousands of developers struggle with the same invisible tax: environments.
Code lived on laptops, secrets leaked, and DevOps teams spent their lives keeping development machines alive. The cloud had solved deployment but never development, as it was built for servers, not for people - and certainly not for software that wanted to behave like people.
That was when AI agents began to appear. At first, they wrote snippets; then they wrote files; then they started running, debugging, retrying, and branching their own work. These systems did not behave like tools, but rather like junior engineers with infinite stamina.
Ivan understood immediately what that meant - agents would need real computers. They don’t need stateless functions that forget everything, or containers that vanish. They require machines with disk, memory, GPUs, and operating systems that could change mid-execution.
The company that quietly became something else
When Daytona launched in late 2023, it looked almost conventional as it offered self-hosted development environments, an enterprise-grade alternative to GitHub Codespaces. The idea was familiar and their execution turned out to be excellent. A $2 million pre-seed round followed, backed by founders of Postman, Honeycomb, Supabase, and Croatian investors like Damir Sabol, Alan Sumina, Tomislav Car, and Luka Abrus. At the time, I reported that Daytona was Ivan’s $2 million bet on simplifying developer environments. He laughed and corrected me: it was his life’s bet.
In mid-2024, Daytona raised another $5 million to accelerate its runtime and add GPU support, driven by interest from GitHub users and the GPU community. Then the company pivoted and last year, Daytona stopped building tools for humans and started building computers for AI.
“Each of these agents is growing faster and faster and faster. New SaaS startups are no longer deterministic web or mobile applications. They’re actually agents - and they are the new product,” this is how Ivan presented his vision during a presentation at another big dev conference, WeAreDevelopers 2025 in Berlin.
The key to Daytona’s pivot was something the company calls a sandbox. They are programmatic, composable computers - CPU, memory, disk, GPU, networking, and operating system that can be assembled on demand, then started, paused, snapshotted, forked, or destroyed in milliseconds. An AI agent can explore multiple execution paths in parallel, keep the best ones, and throw the rest away.
Ivan often points to the rise of people running AI agents on Mac Minis as proof that the need is real. “On the level of one agent or enthusiast, a Mac Mini makes sense. That’s like someone having their own server in the basement. But when you’re building something like Lovable, ChatGPT, or any production-grade agent system, physical machines stop being a solution.” he explains.
“You can’t scale to thousands or millions of agents with Mac Minis. That’s where we come in. Virtual computers, or sandboxes, give agents everything they need - OS, disk, memory, GPU, network - but as a software primitive that can be started, paused, forked, and shut down on demand. The Mac Mini proves the need. Daytona is the answer to scaling that need.”
When the market caught up
Once Daytona released its agent-native runtime, the market moved quickly. Customers like LangChain, Turing, Writer, and SambaNova began running real workloads on Daytona’s sandboxes. Early-stage startups and Fortune 100 companies alike arrived for the same reason: their agents finally had somewhere to exist.
Ivan Burazin during WeAreDevelopers 2025
In just a few months, Daytona reached $1 million in ARR. Adoption accelerated and that momentum led to its latest milestone: a $24 million Series A led by FirstMark Capital, with participation from Pace Capital, Upfront Ventures, Darkmode, and E2VC, plus strategic investments from Datadog and Figma Ventures.
Matt Turck of FirstMark, who joined Daytona’s board, summed it up simply. “We believe there is a shift from human-centric cloud infrastructure to agent-native infrastructure. Daytona turns the idea of ‘computers for agents’ from theory into reality.”
The fresh funds are not for growth hacks, but for hardware. Daytona will invest heavily in its own agent-optimized cloud - real servers built specifically to run sandboxes at scale. The company plans to hire at least 25 people in 2026, across Split, Zagreb, and its new office in San Francisco, with a focus on infrastructure, platform engineering, product, and developer experience.
When I asked him whether compute becomes the new labor in a world where agents, not people, do most of the work, his answer was immediate. “Our type of compute is not the labour. It’s the tool that enables the labour when agents become the primary workers.” Ivan tells IT Logs.
The timing is not accidental - as large models move from clever demos to indispensable tools inside real workflows, the entire market around them is beginning to expand. With models becoming more and more useful, the belief inside Daytona is that this market will grow by an order of magnitude in 2026 - not just in capital invested, but in how many developers, companies, and industries start relying on AI systems that can actually operate computers on their behalf.
Learn how to make every AI investment count.
Successful AI transformation starts with deeply understanding your organization’s most critical use cases. We recommend this practical guide from You.com that walks through a proven framework to identify, prioritize, and document high-value AI opportunities.
In this AI Use Case Discovery Guide, you’ll learn how to:
Map internal workflows and customer journeys to pinpoint where AI can drive measurable ROI
Ask the right questions when it comes to AI use cases
Align cross-functional teams and stakeholders for a unified, scalable approach
Across the region…
Digital Serbia Initiative-supported Business Angels Network and Omorika Ventures have jointly invested €300K in Shosha Games, a Serbian indie game studio developing the cooperative title Water Me & You, giving the team fresh capital to accelerate production and bring its emotionally driven, multiplayer experience closer to market.
A new Romanian VC fund has entered the scene. Vest Ventures is launching with a €16.6M budget, co-financed through the Regional Programme Vest 2021–2027, with a clear mandate to back ambitious pre-seed startups. The fund will combine direct equity investments with hands-on acceleration, tailored mentorship, and access to a wider network of investors, partners, and international markets.
A latest cohort of 21 Greek startups has been selected to join the Greek AI Accelerator, a national program aimed at boosting the country’s innovation ecosystem through artificial intelligence. The announcement was made at an event organized by Endeavor Greece as part of a strategic partnership between the Greek government, OpenAI, and Endeavor Greece. The selected companies will receive access to AI tools, expertise, networks, and funding to help them integrate responsible AI into the products and services they are building.
Croatian tech company SWEN Hospitality has completed the acquisition of Diventa, one of the longest-running and most widely used integrated hotel information systems in the region. The merger creates a new hospitality IT group that is expected to cover around 30 percent of the regional market after full integration.
Gui Wainwright, CEO of Occam
Occam Industries, a defense tech startup focused on autonomous drone operations, has been cleared for integration testing by Ukraine’s Brave1 defense tech cluster and has raised a €3 million pre-seed round to scale its work. The funding was led by Presto Tech Horizons, backed by CSG, with participation from Antler, Freedom Fund, and TYR.vc, supporting further product development and deployment.
Rumor has it…
The word on the street is that a famous Bulgarian brand quietly changed hands, bought by a wealthy insider as a kind of legacy move for his daughter. Since then, a new CEO has stepped in, someone deeply embedded in the startup and mentorship scene. What makes it even more interesting is the whisper that there might have been some mentoring of the new management long before the deal ever went through - though that part, as always, lives firmly in rumour territory.
In this edition, we’re taking a throwback to one of N. Macedonia’s most mysterious tech figures. Often described as the startup ecosystem’s anti-hero, he was behind one of the biggest exits the country has ever seen - yet the myths around him refuse to fade. Among the most persistent is a rumor that he faced intense pressure and underhanded tactics from rivals years ago, around the time he decided to eventually leave the country.
The Founder take…
Thea Gherdan, co-founder and CEO of Proon-Tech
IT Logs: In 2026, what will help startups win faster - moving quickly or building something truly deep?
Thea Gherdan: This is a schemingly good question, because the answer is not either–or. I believe in 2026, velocity still wins, but only when it is paired with depth, context, and purpose. Moving fast to solve a real, felt problem, the “pebble in the shoe”, will always be welcome. Speed matters when it shortens friction or facilitates access, reduces pain, or unlocks clarity for people and systems that are already under pressure or burning need.
What will no longer work is speed without substance. We’ve entered a phase where the environment, whether economic, ecological, geopolitical, changes seemingly unpredictably and is far less forgiving. Teams that move fast while being deeply aware of their context will win: aware of their market, their users, their supply chains, their regulatory realities, and the second-order effects of what they build.
Depth does not mean slow. It means intentional. The startups I see winning in 2026 are those that can iterate quickly without doing shallow work, building fast, but not carelessly; innovating rapidly, but with a long-term systems view. That’s how speed becomes sustainable rather than extractive/ generative or essence-less.
What will slow founders down most next year: hiring, fundraising, or regulation?
Oh, this is an even better question! All three matter, but none of them exist in isolation. What will slow founders down most in 2026 is a lack of contextual intelligence and situational awareness. I actually expect capital to unlock more selectively, but more courageously.
It means more caution in the way money is released, yet at the same time, I see a higher propensity for unlocking funding for experimentation, for fucking around and finding out, because the situation in the global context requires us to get solutions or perspectives that were, until now, not looked at or approached this way. The global situation demands new perspectives, not recycled playbooks.
Do you think AI-first startups will clearly beat those just adding AI on top in 2026?
In the long run, yes, but nuance matters. Having worked hands-on with proprietary, purpose-built AI systems, not just applying existing models, but building algorithms from the ground up to address problems that have been overlooked for decades. In my view, AI that is truly applied, embedded deeply into how a product works, learns, and evolves will create disproportionate impact. Especially in fields like agriculture, climate, materials, sensing, defense/ dual-use MRV, aerospace, and biotech, AI-first architectures will unlock capabilities that simply weren’t possible before. Especially, due to the unique perspective of the humans behind it.
Ultimately, tech in service of humanity will define the winners. The more we lower the friction of adopting meaningful technology - making AI understandable, trustworthy, and genuinely useful in people’s lives, the faster we’ll solve problems many didn’t even realize could be solved.
Upcoming events in the region…
Money Motion 2026 - March 11-12, Zagreb
BWIGA 2026 - March 25, Belgrade
Adriatics Tech Summit 2026 - March 30 - April 1, Sarajevo
Albania Dreaming 2026 - April 2026, Tirana
SaaStanak 2026 - May 25-27, Sibenik
