Designing for the decade, not the demo.
Notes from ten years of building software that had to keep working long after the launch announcement fell off the front page.
A demo is optimised for the first ninety seconds. A production system is optimised for the ten thousandth day. These are different design problems, and confusing them is the single most common failure mode we see.
Reversibility over cleverness
Every architectural decision has a half-life. The ones that survive are the ones that were reversible — small blast radius, clear seams, no assumption baked so deeply that removing it requires a rewrite.
We design under the assumption that every choice we make today will be revisited by someone who has never met us. That constraint is disciplining. It kills a lot of clever ideas early, which is exactly the point.
Boring is a feature
The best long-lived systems we've shipped are aggressively boring. Standard stacks. Standard patterns. Standard deployment. Novelty concentrated only in the parts of the product that genuinely require it.