What is the value of a product that is 99% finished, if the remaining 1% still requires 90% of the time?

I keep seeing that soon 90% of code will be generated by AI. But...

When did we start measuring progress by the percentage of code written?
Honestly — I must have missed that meeting. Since when is “how much code is written” a meaningful metric of results? When did this become important?

In nearly 20 years of working across different companies and roles, I’ve never once seen progress measured as “let’s check how many percent of the codebase we have.”

Not a single manager. Not a single lead. No one saying: “Good job, we’re at 63% of the code.”

And yet here we are.

A new generation of CTOs and tech influencers keeps popping up, confidently claiming that AI will generate 90% of the code (or already does).

Okay — and then what?

What problem does this actually solve?

What is the real win here?

If that code were copied from GitHub, Stack Overflow, or internal templates instead of being generated by AI, would anything materially change?

In reality, that “90% of generated code” usually translates to maybe 10% of the total product development time — at best.

I won’t go deep into details here. I’ve commented on this topic many times already, received plenty of reactions, negative comments, and even personal messages.

But I’ll leave you with this: What is the value of a product that is 99% finished, if the remaining 1% still requires 90% of the time?

Think about it.

PS: Yes, I use AI every day. I generate code with it.

No, it does not save me 90% of my total working time.