"Code might get you 90% of the way there, but for that 10%, humans are not just valuable, but kind of essential. This whole thing doesn't happen without that last 10%."
That's from a Wall Street Journal podcast this week where two non-technical columnists used Claude Code to create an interactive article. Their in-house developer said it got them about 80% there — impressive, but unpublishable without humans guiding it to the finish line.
That last 10% is the craft now. Knowing when something is right. When the architecture holds up. When the experience clicks for a real user.
AI can't know that. We can.
So what separates builders who feel real gains with AI from those who see mediocre results and feel behind?
Today's builders understand that the value we bring has moved to the bookends: the up-front thinking and the final push across the finish line. The messy middle — the grunt work of implementation — is where AI thrives (if we let it).
I'm spending at least 80% of my time on ideation, planning, and architecting. The actual building takes a fraction of what it used to. The rest comes down to shipping the right solution to the right problem (also something AI can't figure out for us.)
