-
Split AI Work by Decision Authority, Not by Artifact Type
Most teams split AI work by role—PM owns prompts, engineer owns code. This creates bottlenecks without clarifying who's accountable for outcomes. Here's a better model.
-
When PM Prompt Ownership Becomes a Bottleneck (And How to Hand Off Safely)
Owning prompts made sense when your AI feature was a prototype. But there are specific conditions where that ownership quietly turns into a liability—and recognizing them is the actual skill.
-
PMs Should Define What 'Good' Means for AI Features—Not Measure It
Most PMs assume they need to run evals because they own the product spec. That assumption creates bottlenecks, false confidence, and slower iteration. Here's where the line actually belongs.
-
Prompts Are Product Specs: Why PMs Can't Delegate Prompt Iteration to Engineers
Most PM guidance says 'collaborate' on prompts, but that framing hides a real accountability gap. Prompt tuning is product iteration—and delegating it to engineers is how AI products end up optimized for technical correctness instead of user outcomes.