Yes, coding has accelerated, but that's mostly the tail-end of the cycle.
We still need to:
๐ ๐๐ฒ๐๐ฒ๐น๐ผ๐ฝ ๐ฎ ๐๐๐ฟ๐ฎ๐๐ฒ๐ด๐ โ which opportunities we're pursuing, in which market segments.
๐ ๐ฆ๐ฒ๐ ๐ผ๐๐๐ฐ๐ผ๐บ๐ฒ ๐ด๐ผ๐ฎ๐น๐ โ what success looks like, when we can achieve it.
๐ ๐๐๐ฎ๐น๐๐ฎ๐๐ฒ ๐ฎ๐ป๐ฑ ๐ฐ๐ต๐ผ๐ผ๐๐ฒ ๐ถ๐ฑ๐ฒ๐ฎ๐ (๐ฝ๐ฟ๐ถ๐ผ๐ฟ๐ถ๐๐ถ๐๐ฎ๐๐ถ๐ผ๐ป) โ based on impact, cost, and supporting evidence.
๐ ๐ง๐ฒ๐๐ ๐๐ต๐ฒ๐๐ฒ ๐ถ๐ฑ๐ฒ๐ฎ๐ ๐ฎ๐ป๐ฑ ๐ถ๐บ๐ฝ๐ฟ๐ผ๐๐ฒ ๐ผ๐ฟ ๐ฝ๐ฎ๐ฟ๐ธ ๐๐ต๐ฒ๐บ (๐ฑ๐ถ๐๐ฐ๐ผ๐๐ฒ๐ฟ๐)
๐ ๐๐ฒ๐ ๐ฒ๐๐ฒ๐ฟ๐๐ผ๐ป๐ฒ ๐๐ผ ๐๐ผ๐ฟ๐ธ ๐ถ๐ป ๐ฎ๐น๐ถ๐ด๐ป๐บ๐ฒ๐ป๐ ๐๐ผ ๐ฎ๐ฐ๐ต๐ถ๐ฒ๐๐ฒ ๐๐ต๐ฒ ๐ด๐ผ๐ฎ๐น๐
AI does not eliminate the need for any of these. In fact they are more important than before.
You still need an outcomes-based roadmap, a metrics tree, a GIST board, even if your engineers use Claude Code.
๐โโ๏ธ See courses that teach all of the methods shown in the diagram, with AI, in the comments.
The logic "we used to do these things because coding was expensive, but now we don't have to" doesn't hold water.
We used these methods because launching bad products and features is a deadly anti-pattern. AI lets us sink ourselves in the quicksand of waste and bloat faster and deeper than before.
AI may create a new PDLC paradigm in the future, when it is better understood and more widely used beyond coding. But we're not there yet.
For now, "classic" product skills are as crucial for product peoples' and product leaders' success as they always were.
If you disagree โ let me know why in the comments