Everyone's talking about AI coding productivity. 200% gains, 20 PRs a day, anyone can ship software now. We get it.
But here's what we're not talking about enough — everything is actually getting shipped.
For a long time, slow dev cycles acted as an accidental filter. Only the most important ideas made it to production. Everything else sat in the backlog — sometimes for good reason. 'Prioritisation' was often just a polite way of saying no.
That friction was doing quiet, invisible work.
Now that friction is gone. And the result isn't just more software — it's bloated, half-baked, everything-but-the-kitchen-sink software.
The bottleneck used to be "Can we build this?" — which kept teams honest about what actually mattered.
Now that the answer is almost always yes, the harder question becomes: should we?
With development getting faster, saying NO becomes the hardest skill on the team. And unlike coding, AI can't do that, YET!