a builder's codex
codex · operators · Amol Avasare · ins_pm-should-not-ship-at-scale

At scale, PMs should up-level the why and what; not ship more features

By Amol Avasare · Head of Growth, Anthropic · 2026-04-05 · podcast · Anthropic growth, CASH, and the squeezed PM

Tier B · TL;DR
At scale, PMs should up-level the why and what; not ship more features

Claim

With 20 engineers and 1–2 PMs, the highest-leverage PM use of time is making the "why" and "what" 5% better — not personally shipping the 21st feature. Shipping for learning (a prototype to communicate an idea) remains valid; shipping for output is not.

Mechanism

Engineering output is no longer the constraint. PM time spent shipping is time not spent sharpening direction. A 5% improvement in the "why" multiplies across 20 engineers' work; a single feature ship adds to the pile and degrades the average focus. PRDs largely vanish in this model — direction lives in Slack and in the heads of senior engineers who already know the why.

Conditions

Holds when:

Fails when:

Evidence

"PRDs largely dead — 70–80% of Anthropic ships have no PRD; just Slack + good engineers."

— Amol Avasare on Lenny's Podcast, 2026-04-05

Signals

Counter-evidence

Brandon Chu's "PM as writer" tradition argues the artifact is the work — writing PRDs is how PMs think. Dropping the artifact may hide quality drift. In small teams or new domains, the PM shipping is exactly the leverage point.

Cross-references

Open the interactive view → View original source → Markdown source →