It’s working code.
Because once something works, your brain **wants **to _trust _it.
The button clicks.
The page loads.
The dashboard renders.
The demo looks real.
And suddenly, it’s tempting to believe the project is further along than it actually is.
But working is not the same as ready.
Working is not the same as secure.
Working is not the same as understood.
And working is definitely not the same as remembered.
That was the surprising part for me.
AI-assisted coding helped me move fast, but it also created a new problem: forgotten assumptions.
The shortcuts.
The warnings.
The “we’ll fix that later” moments.
The decisions that made sense in one session but got fuzzy in the next.
The internet is noticing this too
“A lot of security is contextual.”
— Jack Cable, security researcher
“Speed without control is a liability, not an advantage.”
— Manav Khurana, GitLab
That was the surprising part for me.
AI-assisted coding made it easier to build fast.
But it also made it easier to forget why certain decisions were made in the first place.
And when a quick prototype starts becoming real software, that forgotten context starts to matter.
I wrote more about that here:
Empirical - Your vibe-coded app got serious faster than expected
Curious how others are handling this: where do you keep the “don’t forget this before this ships” stuff when working with AI coding tools?
Top comments (0)