Product Development
The practices, processes, and team structures used to take ideas from concept to shipped product.
Last updated: 2026-04-13
Overview
Product development conventions are being disrupted by AI. The classic stage-gate process (brainstorm → mocks → PRD → code → launch) made sense when iteration cycles were long and AI capabilities were stable. Neither is true anymore.
The emerging model favors rapid parallel experimentation, small high-agency teams with blurred functional roles, and ruthless pruning over upfront planning. “Good taste” — the ability to distinguish excellent from mediocre — is becoming the scarce resource as AI lowers the cost of execution.
Core Concepts
| Concept | Summary |
|---|---|
| prototype-and-prune | 5-stage AI-era model replacing stage-gate planning with rapid experimentation |
| product-operating-model | Marty Cagan’s framework: outcome > output; empowered teams; continuous discovery |
Key People
- julie-zhuo — former VP Design at Facebook; Prototype-and-Prune framework; The Looking Glass
- teresa-torres — Product Talk; Continuous Discovery Habits; empirical AI product team researcher
- marty-cagan — SVPG; product operating model; empowered teams
Recurring Themes
- Working prototypes beat mocks — ship something testable rather than refine static designs
- Tiny parallel teams — 1–3 people exploring multiple ideas simultaneously outperforms larger coordinated groups
- Generalists over specialists — blurred functional roles (engineers who design, designers who write copy) suit AI-era pace
- High agency over consensus — bias toward action without waiting for permission
- Good taste is the bottleneck — as execution gets cheaper, distinguishing excellent from mediocre is what compounds
Connections
- ai-agents — agents are the execution layer that makes rapid prototyping tractable; the prototype loop collapses with agent assistance