A product often starts as a repeated complaint
My independent projects do not look like one neat category. PromptCoach helps people learn to ask better questions of AI. Güügli helps people find Swiss chicken-grill stands. Emboss3D explores turning an image into a layered physical object. Their markets, users, and technology differ.
What they share is a particular kind of beginning: somebody is trying to do something ordinary, and the available tools make it needlessly hard. I like the moment when that frustration becomes a concrete product question.
Different problems, same discipline
PromptCoach came from seeing people dismiss AI after treating it like a search box. The product uses guided practice and examples to teach a more useful interaction. Güügli grew from a very local discovery problem: where is a mobile grill today, and when is it open? Emboss3D sits at a different boundary, between an image on a screen and a thing you can hold.
Each asks for a mix of product judgment and engineering restraint. The visible interface should make the next useful action obvious. Behind it, the data, workflows, and operational details still have to hold up. I enjoy owning both sides of that equation.
The portfolio is a laboratory
These projects let me keep learning close to users and close to the code. They also keep me honest as a leader: it is harder to speak in abstractions when I have recently had to debug a broken flow, explain a confusing screen, or answer a real person who expected the product to work.