How is vibe coding changing the way we work and make decisions? In this interview, Tobias Mihalek, Senior Project Manager at BARMER.KI, talks about the potential of the current AI Vibe and how highly regulated industries can bridge the gap between legal guardrails and innovative technology.
Mr. Mihalek, you have been working on large projects for 20 years. How did you become one of the driving forces behind BARMER.KI?
I was lucky enough to be able to choose this path deliberately. I’d already worked with a lot of new technologies before. In my private life, too, I’ve long been an early adopter, whether it’s electric cars or home automation. This intrinsic urge to explore helps me see the opportunities in new things, even if it means investing up front. At BARMER.KI, the mission has two sides: trying out new technology and finding a way to anchor it in the organization. That takes someone who isn’t just convinced by the technology but also gets that conviction across internally. I know the business from customer service all the way to the legacy systems, and that builds trust. And I believe you can only carry transformative initiatives forward with decisions that we make and stand behind together.
How do you use AI, and which of those uses would no one have imagined a year ago?
In my day-to-day work, I use AI as an advisor, for image generation, and increasingly as a coding agent in projects. What I personally couldn’t have imagined a year ago? That I’d be able to get web applications off the ground on my own without traditional coding expertise. It used to take several weeks and a team to build a mockup or a first prototype. Today, I can vibe a vision into shape with AI as my sparring partner and make it visible and tangible within a few hours.
Has vibe coding changed the way you conceive and prioritize products?
The question of whether an idea or a project can work can be answered much faster. Vibe coding keeps the barrier to simply trying things out to a minimum. In my previous work, there were often long discussions about whether something was even feasible. By nature, I’m someone who only believes “that won’t work” once I’ve failed at it myself. One example: For a current project, I had to check whether an optical character recognition (OCR) workflow for analyzing receipts was technically feasible. Over a weekend, I set up a private server, used n8n and Claude, and tested the core process. It worked. With this kind of micro-mocking, which shows that the core of the requirement can be met, you win over business departments and decision-makers much faster than with any presentation.
How do you reconcile this drive to explore with concrete business goals without losing focus?
Research is…
Want to read more? You can find the full interview with Tobias Mihalek in our whitepaper, “Owning the Vibe.”