warrenshi

The thing that changed the most for me wasn't the coding setup.

It was the order.

A couple years ago I'd think:

architecture database implementation deployment

Now I spend much more time before that asking:

does this map to a workflow people actually repeat?

AI made implementation so much cheaper that I've become much more cautious about building the wrong thing quickly.

I think they overestimate less than it looks.

The actual coding part compressed dramatically.

The unpredictable part is still everything around the code.

Ironically I use AI heavily for structuring thoughts, but often delete the most polished parts.

Overly optimized writing now triggers the same feeling as overly optimized SEO pages used to.

I usually talk more about the decisions/tradeoffs than who typed the code.

Using AI changed implementation speed for me way more than it changed product direction.

Honestly the API key part ended up being the easy part for me.

The messy stuff was abuse prevention, retries/fallbacks, and sudden cost spikes once real users started using different workflows.