I started in 2020 converting PSD designs into WordPress sites. Today I build AI-powered mobile apps and platforms. The path between those two points taught me things a CS degree wouldn't have — and required unlearning a few things too.
What WordPress actually taught me
Clients don't buy code. A restaurant owner doesn't care about your clean architecture; they care that reservations went up. WordPress work is brutally direct about this — you're never more than one conversation away from the business outcome. That instinct transfers to every AI project I scope today: start from the business problem, work backwards.
Constraints breed pragmatism. When the budget is small and the deadline is Friday, you learn to choose the boring solution that ships. Plugin or custom build? Buy or make? That judgment — knowing when not to engineer — is rarer than engineering skill.
Maintenance is the real product. Sites live for years. The first time you revisit your own two-year-old work, you learn more about code quality than any style guide teaches.
What I had to unlearn
The plugin reflex. WordPress trains you to find a plugin for everything. Product engineering often demands the opposite: own your core logic, because your differentiation lives there. The skill is knowing which is which — buy commodity, build the moat.
Treating the database as someone else's problem. WordPress hides the data model. Real products are built on the data model — designing schemas for Postgres, Firestore, or Convex is now where I start, not where I end.
Shipping without instrumentation. A brochure site either works or doesn't. A product needs analytics, error tracking, and feedback loops from day one — you can't improve what you can't see.
The throughline
Every stage — WordPress, full-stack, mobile, AI — has the same job underneath: understand what the business actually needs, then build the simplest system that reliably delivers it. The tools got more powerful. The job never changed.