AI Is Becoming an Operating System

Wonderful has entered a new phase of expansion after raising $550 million in its Series C funding round, which valued the company at $5 billion, as it continues developing its AI operating system for large enterprises. Since its Series B round in March 2026, the company has expanded to more than 35 markets and grown its global workforce to 650 employees, with an increasing focus on Central and Southeast Europe.

Most of the new capital will continue to go into people, local teams, and specialists who work directly with customers, Wonderful regional director Vedran Bajer tells ICTbusiness Media - ICTbusiness.biz. In the Adriatic region, the company started with just two people a year ago and has since grown the team to nearly 20, with around ten additional positions planned by the end of the year. Wonderful is placing particular emphasis on forward-deployed engineers who work inside the customer's business environment and connect AI solutions with existing systems and processes.

Bajer argues that companies in Croatia and neighbouring markets no longer lag behind major Western markets in terms of technological readiness, although there are still differences in decision-making speed and willingness to move AI projects from pilots into production. Clearly defined business processes, rather than the technology itself, are in his experience the most important prerequisite for rapid implementation. At the same time, Wonderful no longer wants to position itself merely as a provider of individual AI agents, but as a platform that connects agents, data, applications, and end-to-end business processes.

This approach allows integrations and security mechanisms built for the first use case to be reused when additional agents are deployed. For large enterprises, model independence, the ability to operate in the cloud or on-premises, and compliance with security and regulatory requirements are also becoming increasingly important. Bajer expects employees over the next few years to spend less time executing individual operational steps and more time defining objectives and supervising outcomes, while AI agents coordinate a growing share of day-to-day business operations.

Wonderful raised $550 million in its Series C round at a valuation of $5 billion. What does an investment of that scale concretely change for the company, particularly in Croatia and the region you manage? How much of the new capital will go towards business expansion, product development, and strengthening local teams across the Adriatic region, Hungary, and the Baltic states? Should we expect a significant increase in hiring in Zagreb and across the region, and which professional profiles will you be looking for most? Wonderful's valuation has risen very quickly to $5 billion. Which concrete business indicators – revenue growth, customer numbers or the scale of production deployments – justify that valuation and investor expectations today?

Most of the capital is going where we have always invested the most – into hiring. When we launched the Adriatic operation a year ago, we started with two people, one in Zagreb and one in Belgrade. Today, that team is close to 20 people, and by the end of the year we plan to add around ten more positions. What we are looking for most are forward-deployed engineers: people who physically sit with the customer, integrate our technology with the customer's systems, and test solutions in Croatian, Serbian, or Hungarian rather than doing that work from a remote hub.

We currently have Forward Deployed Engineer positions open across all three regions I manage – the Adriatic, the Baltics, and Hungary – and each comes with a slightly different challenge. Fundamentally, we are looking for more people who can step into a customer's operations, take an ambiguous problem and turn it into AI agents that actually work, from the first conversation all the way through to production.

The common denominator across all three roles is that we are not looking for people who arrive with a ready-made solution. We are looking for people who are prepared to learn, sit with the customer, understand the process from the inside, and take responsibility for the entire journey to production. That is the profile that genuinely gives us speed.

As for the valuation, what convinced investors was not the story, but production. Since our Series B round in March this year, we have expanded to more than 35 markets and grown to 650 employees worldwide. We see the same pattern among customers everywhere, regardless of the region: seven out of ten move into additional business processes within the first three months after deploying their first solution.

That is what investors are really looking at in funding rounds of this size – not the idea itself, but the speed at which an initial project turns into a long-term relationship and additional deployments.

You argue that companies in the Adriatic region, Hungary, and the Baltic states are no longer lagging behind major Western markets in the adoption of artificial intelligence. What does Wonderful's business in Croatia and the wider region look like today? How important is the Croatian market within the region you manage, how many active projects or customers do you have, and in which industries are you currently seeing the strongest demand? Are there significant differences between Croatia, the rest of the Adriatic region, Hungary, and the Baltics when it comes to companies' willingness to move AI from pilot projects into real production? When you say customers can get a production AI solution up and running within weeks rather than months or quarters, what needs to happen inside the company to make that speed possible, and what most commonly slows implementation down?

Croatia is personally my starting point. I opened our first office here, and I live here, so it is difficult to separate how much of its importance is business-related and how much is personal. Today, we see the strongest demand in telecommunications, banking, insurance and logistics – sectors with high volumes of work and where mistakes translate directly into costs. We will soon be able to talk publicly about some of those projects, while for others we are still waiting to agree with customers on how they can be communicated.

There are differences between markets, but they are not so much about technological readiness as they are about the speed of decision-making, and that varies more from company to company than from country to country. What really determines whether something remains a pilot or reaches production within a few weeks is how clearly the process has been mapped before we even start building the agent.

The biggest factor slowing things down is not the technology. It is companies that have not clearly defined for themselves how their own processes actually work. When we work through that part with the customer in advance, the first agent can be in production within a month, and every subsequent one, because it inherits part of what has already been built, can be deployed in days rather than months.

Wonderful no longer positions itself simply as a company developing AI agents, but as an AI operating system for large enterprises. What does an AI OS provide that companies cannot obtain by using individual AI agents, copilots, or generative AI tools? Can you give a concrete example of how the Wonderful AI OS connects agents, business processes, data, existing applications, and AI-native applications, and which parts of a business process can then be handled by artificial intelligence? Which results do customers most commonly measure after implementation – cost reduction, productivity gains, faster customer service, or revenue growth – and how quickly can they expect measurable return on investment? Wonderful also stresses that the system is model-agnostic and can operate both in the cloud and on-premises. How important are data security, regulatory requirements, integration with existing IT systems, and avoiding dependence on a single AI model in purchasing decisions by large enterprises today?

The best example is a large telecom operator in the wider region for which we initially started building a solution solely for customer support. Already during the first call, it became clear that the conversation touched user authentication, package and pricing checks, troubleshooting a satellite receiver, profile management, and even post-call surveys – ten different areas within a single company.

If we had built those as ten separate projects, each would have required its own budget and approval process. Instead, a single agent on one platform covered the entire workflow, and every integration we built for that use case remains reusable for the next one.

Customers typically track three things: how much the time required to resolve a request has been reduced, how many employees have been freed up to deal with more complex cases that genuinely require judgement, and how quickly the first agent expands into additional processes.

That third metric is actually the best indicator of return on investment, because when the second and third agents are built on the foundations of the first, the cost of each additional implementation falls while the value increases.

Security and independence from any single model are no longer optional. They are a prerequisite even to enter a serious discussion with a bank or insurer. That is why our system is model-agnostic and can run in any cloud environment or on-premises, depending on what the regulator or the company's internal policy requires.

We build governance around the strictest standards applicable in the markets where we operate – GDPR, ISO standards, SOC 2 Type II and DORA requirements – and so far we have not failed a single security or compliance review conducted by a financial institution.

In only a few years, generative AI has moved from chatbots and copilots to agents capable of executing entire business processes. How do you see the next phase of enterprise AI developing, and how much will the way large companies operate change over the next two to three years? Will we move from today's model, in which an employee uses an AI tool, towards one in which autonomous AI agents coordinate with one another and execute a large share of operational work while people primarily set objectives and supervise outcomes? Where do you draw the line when it comes to AI autonomy, and which decisions must remain under direct human control for reasons of accountability, security, or regulation? Could the current generation of AI technology allow Croatia and the wider region to leapfrog part of the technological gap with larger markets, and what do companies need to do today to avoid once again becoming late adopters two or three years from now?

I think we are moving towards a model in which people perform fewer individual steps and increasingly set the objective and supervise the outcome, while agents coordinate a greater share of operational work among themselves.

That is already happening in parts of organisations with the highest volumes and the clearest rules, such as customer support or request processing, and it is expanding towards more complex processes such as loan approvals or insurance claims settlement.

For me, the boundary is very concrete. It is not about what an agent is allowed to read, but what it is allowed to write or change. An agent that only reads data can make a good demo, but the moment it has authority to approve a request, initiate a payment, or change the status of a contract, there must be an identity, clearly defined permissions, an audit trail for every action, and a point at which a human must approve.

As for the opportunity for the region, I genuinely believe this is the moment when geography stops being a differentiator in terms of access to capabilities. Companies in Zagreb or Ljubljana today have access to the same technology as their competitors in Singapore or New York.

What turns that advantage into real results is the local language, local regulation, and people who sit inside the company and build the solution together with it rather than delivering a finished product from a distance.

If the region is to avoid ending up as a late adopter again, companies today need to treat AI as a board-level decision rather than the purchase of a tool for a single department. They first need to clearly define their own processes, because an agent simply accelerates whatever you give it – including your own ambiguity if you fail to resolve it before you begin.

For us, it always comes back to the same principle: acta, non verba.