While AI adoption has quickly moved from buzzword to actual business infrastructure, the gap between companies seeing tangible results and those sticking to theory often comes down to execution. In Romania, only about 5% of companies are actively using AI and automation, but the conversation is no longer about “if” but “how” and “where to start” in a practical and effective way.
In this context, we spoke to Ralph Boccalini, Microsoft 365 Copilot implementation specialist and AI solutions architect at EMGC. He works closely with organizations in Romania and Europe to transform AI from a standalone tool to a real operational benefit. The discussion cuts through the noise and highlights where most businesses struggle, what “AI-ready” actually means, and how small businesses can get started while managing their budgets while achieving measurable impact.
In other words, AI doesn’t solve chaos; it augments what already exists. As Ralph Boccarini emphasizes, the companies that first build a clear, structured digital foundation will be the ones that turn AI into real results, not just hype.
You work with companies in Romania and Europe every day on Microsoft 365 and AI implementations. What is the first thing you notice when you join an organization? Are you ready for AI?
You’ll know if it’s ready in less than 10 minutes. Rather than asking, “Are you ready for the digital transformation of AI?” ask where your company’s knowledge resides and how it is managed. If reporting is still done manually in Excel, departments are siled, and “processes” are email + WhatsApp + paper, then AI won’t solve anything and will only add to the confusion.
At EMGC, we start with the same baseline every time, including data location, permissions, collaboration habits, and security posture. That’s what makes Copilot AI useful and secure. That’s why I’m so particular about the word “preparation.” Most companies don’t need a moonshot. We need a consistent digital environment where information is structured and accessible based on rights.
And yes, there is still a huge gap in Romania. Only 5% of companies use automation/AI, compared to over 33% in major EU countries. This shows that “preparatory work” is still the main task here.
What will AI actually replace in the company you work for, including tasks, entire processes, and roles? Where is the line between “AI assists” and “AI replaces”?
The first thing AI replaces is time. Because time is where impact can be measured fastest. But it’s just an entry point, not the end goal. In most organizations, success begins with repetitive knowledge work tasks such as drafting and summarizing emails, generating documents and reports, analyzing Excel data, and summarizing meetings into action lists. If done correctly, hours turn into minutes consistently.
The real difference begins when we stop treating AI like a “chat window” and embed it where the work actually happens. That’s why I continue to separate Copilot Chat from Microsoft 365 Copilot. Copilot Chat is a secure, enterprise-protected chat experience that is typically web-based by default. Help with research, brainstorming, and drafting. Microsoft 365 Copilot (licensed add-on) is different. Embedded in Word, Excel, PowerPoint, Outlook, and Teams, it’s designed around your organization’s content and permissions, so it works in your actual work context, not just what you paste into a chat.
Where I draw the line is simple and very practical. Where humans own the outcome, such as review, validation, and decision-making, it is “AI-assisted.” “AI replacement” occurs when the workflow is structured to run end-to-end with controls (routing, classification, templated reporting, document processing) and the risk profile is acceptable.
At EMGC, we designed Copilot to be well-managed, based on the right data, and with the right permissions. This allows you to scale from personal productivity to actual process automation without creating any risk. For the avoidance of doubt, Microsoft clarifies that prompt/response and organizational data accessed through Microsoft Graph is not used to train the underlying model, which is a non-negotiable baseline for enterprise deployment.
How do your employees react when they start working with an AI agent? Is there resistance, fear, or enthusiasm? What is the most common feedback you hear from people in the field?
In most companies, the first important “hire” is the business owner/CEO. If leaders don’t personally see the benefits (faster decision-making, less operational pain, less administrative lost time), adoption remains theoretical.
The pattern across Romanian SMEs is that the real obstacle is rarely technology. It’s about pace, habits, and the idea that digitalization is “complex, expensive, and risky.”
When leaders use it consistently and set a clear operating model, employees typically respond in a very predictable way, moving from “curiosity” to “realism.” Not because they like AI, but because it makes their working hours more manageable. There is an important difference here. Deployment is change management, not license activation. You need to define the top use cases for each role, teach people how to clean data access and validate output, and agree on what will be automated and what will remain human property. When done right, especially in an environment where employees juggle 10 to 40 digital tools every day, people become less aware of “another tool” and reduce interruptions, duplication, and management loops.
The most common feedback I’ve heard “from the field” isn’t “I write faster.” That means that you are less likely to get stuck. Less time chasing information, less time rebuilding things from scratch, fewer repetitive steps, and smoother handoffs between people. Measurable aspects are also consistent with real-life experience. Depending on hiring maturity, companies typically save between 45 and 150 minutes of office time per employee per day. This is exactly the type of pressure relief that improves the work environment and reduces the feeling of constant work stagnation and burnout.
Specifically, what kind of budget do you need to fully implement AI? Is there a minimum threshold below which it doesn’t make sense to start?
The true “minimum standard” is not the numbers, but whether the fundamentals exist. If your data is fragmented and access is uncontrolled, you may not be able to afford to consider digitizing your business. The first step is a quick diagnosis of how the company actually operates and where information is being lost. Then configure your environment appropriately.
However, for many small and medium-sized businesses, the entry point is not that great. In fact, a basic digitization package (license + workspace configuration) starts from around 3,000 euros., Depending on the type of business, amount of data, and users, costs are primarily determined by licensing and setup, and not by custom development.
This “think big, start small” approach is exactly what we brought to small and medium-sized businesses through the conference EnterTech 2026, which we held in Bucharest in March, and which we will continue to offer every year with the continuation of this event and its next edition, which is already in preparation.
From an EMGC perspective, we are very pragmatic. Build a minimal secure setup that quickly realizes measurable time savings, then expand. By doing so, you can keep your ROI and trust high, especially for cost-sensitive Romanian SMEs.
What are the tangible, measurable benefits for companies that have already adopted AI, such as productivity, cost, and speed of decision-making?
Productivity is the first thing you can measure. In office work, 20% to 40% of daily tasks can be automated. This translates into a recovery of 2-6 days per month per employee at higher implementation levels, or 45-150 minutes per day depending on usage maturity.
Then you can see the costs and effects of your decisions. When work is no longer manually fragmented, there is less operational effort and error, and readers are no longer making decisions about old Excel files. Romania’s problem is not a “lack of technology” but the speed of adoption. Companies that move faster make faster decisions, which leads to competitive advantage.
That’s why my role as a Microsoft 365 Copilot Architect focuses less on “prompts” and more on architecture, including permissions, governance, measurable adoption, and controlling Copilot into daily workflows.
If a Romanian CEO asked you today, “I only have six months and a limited budget, where should I start with AI?”, what would you say?
Start with a short audit. This is where time is lost, errors are repeated, and information breaks down between departments. Then centralize your knowledge and workflows within Microsoft 365. They typically use SharePoint as a structured knowledge backbone, Teams as a collaboration layer, and clear permissions/governance. This allows Copilot to work with trusted, company-owned information rather than scattered pieces. Without it, AI will never be consistent.
And one more thing, and this is important, is exactly why we built EnterTech 2026 at EMGC. Practical digitalization for small and medium-sized businesses, not theory, but practical content and a lineup of high-quality speakers. The fact that the next edition is already in preparation shows that there is a demand for business-first applied AI in Romania. And we continue to push the bar every year with our partners and collaborators here.
