For me, AI is not an island. It is an additional layer that makes sense when it improves a real working or decision-making process.
I could have said this myself. Digitising an unnecessarily complicated or fundamentally flawed workflow will not automatically improve it. That is why I first ask: What problem are we trying to solve, what causes it, and what solution will help in everyday work?
From the real problem to measurable impact: clarify the need, define the value, set boundaries and test results in everyday use.
Clarify the workflow, people involved, causes and actual needs.
Make the intended benefits specific: less workload, better quality, greater speed or better decisions.
Consider data, responsibility, approvals, compliance and security from the start.
What matters is practical usability and measurable improvement, not just a successful demo.
Since the early waves of generative AI, I have been exploring its practical use in business processes in depth.
I start with the people who use a process every day. Together, we make workflows, bottlenecks and expectations visible — for example, in an AI Design Sprint. This creates a shared understanding before we start discussing tools.
Not every idea needs to become a project. I assess potential use cases by value, feasibility, available data and effort. We start where a manageable solution can make a noticeable difference.
A small, working first version makes assumptions testable. A clearly defined use case and early feedback help us discover what works before investing substantial time and budget in a larger solution.
Sometimes a language model helps. Sometimes an interface, conventional automation or a better workflow is enough. I choose the technology to fit the problem, with expert review, data protection and clear responsibilities built in.
A good demo is not enough. Time savings and quality must justify the effort of implementation, use and support. Ongoing model costs, integrations and human rework also belong in the calculation.
I test new tools, question results and learn from practical use. Feedback from the people involved and measurable impact tell us what to improve, expand or deliberately leave behind.
I do not want to introduce as much AI as possible. I want to find the right problems and develop solutions that work in everyday use, scale economically and genuinely improve operations.
What type of solution fits the need?
My further training, completed in 2026, adds Azure AI, Azure OpenAI, APIs, automation, governance, the EU AI Act and GDPR to many years of project and business application experience.
From MVP design and AI Design Sprints to generative AI tools, agent workflows and API integration: choosing the right application remains the priority.
Explore AI expertise →