Anything I would’ve said in this section two weeks ago is already out of date. And I’m going to leave this right here as proof that this is now an evergreen statement.
The problem with traditional product development isn't the process—it's the speed. While teams spend months debating requirements and refining wireframes, markets shift and opportunities disappear. Nothing matters until you get a real concept in a consumer's hands. Nothing.
Early in my career at Digitas, I trained teams on Dan Roam's "Back of the Napkin" philosophy: your ideas are ephemeral and will get lost if you don't make them tangible immediately. Sketch quickly. When in doubt, sketch it out. The goal was moving from abstract concept to something you could point at and improve.
That principle remains, but the tools have transformed what's possible. What started with napkin sketches now extends to functioning prototypes built over lunch, not over 5-6 sprints. Tools like Claude Code let you flesh out an idea into working software faster than most teams can write a Product Requirements Document.
This changes everything about the design process. Instead of incrementally progressing through discovery → design → delivery, you loop rapidly from insight to shippable prototype multiple times, increasing fidelity with each cycle. Treat this as adding a y-axis to the traditional triple diamond process, where increasing position on the y-axis represents increased completeness. You can get all the way to testable code in one iteration, so why pause anywhere along the way?
At Panasonic, we prototyped Umi's core conversation model in weeks, getting real family feedback on working AI interactions rather than static wireframes. The speed advantage was decisive—we learned what worked and what didn't before competitors even understood the problem we were solving.
The lesson: in the AI age, the fastest iteration wins. Think in shallow terms because consumers will tell you exactly what needs to change. Speed of learning beats perfection of planning.