In praise of boring technology
A note on choosing tools you will still understand at 2am in six months.
Every side project is a small referendum on how much novelty you can afford. It is tempting to treat each one as a chance to learn the newest framework. Increasingly, I reach for the boring option instead — the tool I already understand, whose failure modes I can predict, whose documentation is stable.
Boring technology is not a lack of ambition. It is spending your limited novelty budget where it actually differentiates the work. In a project about AI-assisted categorisation, the interesting part is the categorisation — not the database. So the database should be the most predictable thing in the stack.
Choose tools you will still understand at 2am, six months from now, when something breaks.
The reward is compounding. Boring tools free attention for the one or two genuinely hard things a project is really about — which is, usually, the whole reason it was worth building.
Denis Cercasin
Business Informatics student working across software engineering, artificial intelligence, and product design.
Keep reading
Designing for when the model is wrong
Most AI product thinking optimises the average case. The experience is decided by the tail — the moments the model is confidently wrong — and that is a design problem, not a modelling one.
Business Informatics is a translation discipline
The most valuable skill the field teaches is not coding or strategy — it is translation between people who think in systems and people who think in outcomes.