Stanislav Kapustin

As a kid I liked drawing schemes in my notebooks — lines and blocks, little diagrams of how something worked. I liked that the angle of incidence equals the angle of reflection. Before university I was choosing between two paths, computer science or art, and I went with art — studied at an art school, and I can actually draw.

Form pulled me in, and the process behind how something comes together. Computers stayed with me too, quietly, the whole time. After school I went into marketing, websites, and illustration.

In 2017, LinguaPolis started — a language-learning game, still running today. I built it first as a game designer: a new learning system, a form, mechanics. Illustration, promo videos, marketing, sales — all of it was mine to figure out. A small side project turned into years of my life. I ended up at conferences, meeting teachers, building contacts across the language-learning industry.

The part I kept coming back to was marketing and process. With illustration, the moment I liked most was when a drawing actually solved a problem.

That pull eventually took me into a marketing role at a Dutch tax firm, where I also ran their internal learning platform. Somewhere in there I found n8n, and it looked like exactly the kind of system I used to draw in notebooks as a kid — boxes and lines, actually running. I started with simple workflows and kept wanting to go one level deeper: what happens if this connects to that, what if the whole thing could catch its own errors. That curiosity is probably what kept me going past the first workflow that worked, toward the ones that actually got good.

Working inside a tax firm gave me something tutorials don't: context. I understood what the numbers meant, why an invoice needed to match a specific ledger entry, why a process existed before I tried to automate it. Understanding process is what lets me walk into a different niche now and figure out fast what's actually worth fixing.

I kept going in the direction I'd found — automation. These days that's The Automation Detective.

I like digging into numbers until an actual insight shows up, then making the one change that comes from it. Before automating something, I check whether it's even worth automating in the first place — half the job is figuring that out, not the building itself.

These days that mostly means n8n, Python, SQL, and Claude and OpenAI APIs. If you want to talk about a project, get in touch.