Coming from an engineering and agency background, our default mindset was always to write clean, fully modular, and perfectly scalable code before pushing anything to production.
However, building our own Micro-SaaS products quickly shattered that mindset.
We learned the hard way that spending weeks over-architecting a feature that users might not even want is the fastest way to waste momentum. Shifting to "shipping fast and listening to user feedback" forced us to make uncomfortable trade-offs between technical debt and speed.
When you are in the early days of building a product, you don't have the budget for massive operational bloat. You are forced to design every touchpoint from first principles. You sit down, map out what a perfect, magical experience looks like, and you build backward from there.
But as companies scale, they compromise.
Right now, I see a lot of founders making a classic mistake with AI and automation: they are taking broken, bureaucratic, legacy processes and simply layering software on top of them.
They ask: "How do we make our manual data entry 20% faster?"
There s a common assumption that job applications are mobile-first now. Before moving into building a mobile app, we wanted to test that assumption with data rather than rely on intuition..
First, we looked at real usage. In our GA4 data, mobile accounts for ~22% of active users. But desktop dominates engagement and core actions, longer sessions, more events, and more consistent usage over time.
Everyone focuses differently. And for a lot of people, where and how they work makes a huge difference.
Some need silence. Others need background noise. Some can t stand the cold, others lose focus if it s too warm. Coworking spaces? Only if they have the right vibe. Offices? Only if colleagues aren t interrupting every five minutes.
Most complaints I get are not new. They are old, quiet, and only turn into a complaint once someone finally runs out of patience. By the time I hear it, the person has usually been annoyed for weeks and just never said anything, or said something small that I did not register as a warning. Curious whether that pattern is just me, or whether the actual complaint almost always arrives later than the moment the person first felt it.
AI is becoming a bigger part of software development every day. It can generate code, write tests, review pull requests, and even help with debugging.
But I don't think every part of development should be automated.
Understanding user needs, making architectural decisions, solving complex business problems, and thinking creatively still seem to require human judgment.
