The old model went something like this: have an idea, refine it, refine it again, run it past people, refine it further, maybe ship it. Most ideas died somewhere in that process—not because they were bad, but because the effort required to find out if they were good was too high.
LLMs change that equation. The cost of making a rough thing has collapsed. You can go from notion to prototype in an afternoon. Not a polished prototype, a scrappy one, full of gaps, but real enough to learn from.
And yet.
Most of what I see people doing with these tools is compressing the refinement stage. Getting to polished faster. Which is fine, useful even, but it misses the more interesting opportunity, compressing the commitment stage. Making things before you're sure they're worth making. Sketching in code instead of in decks.
I've been doing this A LOT more lately. Small projects that exist to answer a question rather than to ship. Prototypes that might go nowhere. The kind of making I used to do when I was younger, before I learned to be sensible about where I put my effort. Some of it has led somewhere. Some of it hasn't. That's the point.
The permission slip here isn't "AI will do the work for you." It's closer to the cost of trying things is now low enough that you can afford to try more things. You can be less precious. Less strategic. More willing to make something just to see what happens.
We used to call this play. Somewhere along the way, the industry got serious. Briefs got serious. Conversations about metrics, alignment, and measurable outcomes. The craft of making something delightful became secondary to making something that performs.
Maybe the tools are offering us a way back. Not to abandon rigour, but to remember that rigour was never the whole point. The point was to make things worth making. And sometimes you don't know if something's worth making until you've made it.
Play more. Sharpen less. See what emerges.