AI Is Fast. Engineering Is Still Human
AI writes code faster than ever, but building good software is still about making the right decisions. Here's how I think about engineering in the age of AI.
A few months ago I caught myself asking AI the wrong questions.
Instead of asking it to solve problems, I was asking it to write code.
It worked.
Most of the time.
But the more projects I shipped, the more obvious something became.
Writing code isn't the difficult part anymore.
Making good decisions is.
AI Is Incredibly Fast
Today I can scaffold a project in minutes.
Generate UI.
Write tests.
Refactor components.
Even document APIs.
Things that used to take an afternoon now happen before I finish my coffee.
That's an incredible productivity boost.
But speed alone doesn't build good software.
Engineering Starts Before The First Line Of Code
Clients don't pay for code.
They pay for solutions.
Before opening an editor, I usually spend more time thinking than typing.
Questions like:
- Should this feature exist at all?
- What's the simplest architecture that will survive six months from now?
- What happens when this grows ten times larger?
- What part should never depend on AI?
Those questions still don't have one correct answer.
That's engineering.
AI Doesn't Know My Product
Claude can generate a beautiful React component.
Cursor can refactor an entire folder.
ChatGPT can brainstorm implementation ideas.
But none of them know my users.
They don't know why a feature exists.
They don't know what my client is worried about.
They don't know which shortcut will become technical debt next month.
Context is still human.
Judgment is still human.
The Best Use Of AI
These days I rarely ask AI to "build the feature."
Instead, I ask it things like:
- What options do we have?
- What trade-offs am I missing?
- Can you review this implementation?
- Is there a simpler approach?
- What would break if requirements change?
The conversation becomes much more valuable.
AI becomes another engineer in the room instead of an autocomplete tool.
My Job Changed
AI didn't make engineering less important.
It made good engineering more valuable.
Because implementation became cheaper.
Decisions became more expensive.
Anyone can generate code.
Not everyone can decide what should be built, what should stay simple, and what should never be built at all.
That's where experienced engineers create value.
The Future Isn't AI vs Engineers
The developers who benefit most from AI aren't the ones writing the longest prompts.
They're the ones making the best decisions.
AI will keep getting faster.
Models will keep improving.
New tools will appear every month.
But software has never been just about writing code.
It's about solving problems for real people.
And that's still a very human job.
