From machines to marketing

I trained as a mechatronics engineer: mechanics, electronics, and control systems in one degree. The habit that stuck is seeing everything as a system, with inputs, outputs, and feedback, and knowing it will usually break at its cleverest part.

I still like working with my hands. In 2025, two friends and I wired electrical cabinets for local businesses on the side. But what I enjoy most is designing the procedure: the drawing that lets the next person run the machine without me in the room.

In 2021 my best friend Miha and I started a writing agency. I set up my own business, and within a few years I was running the whole production side: SEO audits, topic research, writing, editing, and the day-to-day work of a content manager. On paper it was a strange jump. In practice it was the same discipline: a search engine is a machine with no manual, and content production is a production line.

So I ran it like one. I built and led a fully remote team of about eight writers from several countries, wrote the SOPs, and turned our training into an internal course. We specialized in what we called “comparative content” for SaaS companies, hundreds of articles written by hand, and I kept fixing the stations where quality leaked.

When language models got good enough to work on that line, most teams did one of two things: banned them, or let them run unsupervised. Both backfire. The first wastes the machine, and the second burns the reader’s trust.

In 2025 we did neither. I rebuilt our production into a pipeline that’s now about 90% automated. Twelve stages of LLM agents handle the research and writing, and people step in at the end to add images and check quality, verifying every source and catching hallucinations. It has produced 400+ long-form articles, and it follows each client’s messaging and positioning instead of churning out generic AI slop. Today I bring the same approach to other businesses, a couple of projects at a time.