06-reference

every claude code for product managers

Thu Apr 30 2026 20:00:00 GMT-0400 (Eastern Daylight Time) ·newsletter ·source: Every (`p/` slot, not Source Code) ·by Marcus Moretti
everyclaude-codeproduct-managementagent-deployercompound-engineeringslash-commandsmcpsponsor-convexmarcus-morettispiral

Claude Code for Product Managers — Marcus Moretti (Every, 2026-05-01)

Why this is in the vault

Adjacent-but-not-identical to our HQ work. RDCO operates on Claude Code as the COO harness; our skills/ directory IS the load-bearing surface. Every is now publishing a PM-flavored variant of the same playbook (slash commands, MCP, agent fan-out) — relevant for two reasons: (1) borrowable workflow patterns we could lift into our skills, and (2) reinforces the agent-deployer evidence cluster (a major AI publication is teaching its own GMs to operate this way and packaging it as a thought-leadership piece, not a how-to).

Sponsorship

Convex ad in-line (“Building blocks that help your agents compose elegant backends”). Convex is a backend-as-a-service positioning itself for the agent-native era. Note for hq-business-stack: Convex is a recurring agent-era backend sponsor across the Every property — worth tracking as a competitive signal vs whatever we end up using for HQ persistence.

Sister-publication promos within the paywall: Spiral, Monologue, Sparkle, Cora. Standard Every cross-promo, no surprise.

Core argument

Moretti, GM of Spiral (an Every Studio app), argues solo PMs can collapse the multi-tool busywork stack — Linear, Jira, Notion, dashboards, research notebooks — into a single Claude Code thread by treating “the conversation as the work” (verbatim). One human-authored roadmap doc; everything downstream (PRDs, GitHub issues, project board state, metric briefings, anomaly triage, literature reviews) is generated or maintained by Claude via slash commands and MCP integrations.

Concrete pieces he describes:

Posture: “AI should open up product management to more people” (verbatim) — flattening the access curve, same beat as Klaassen and the rest of the Every Source Code stable.

Format: opinion essay with embedded tutorial. He cites Cat Wu (Claude Code head of product), Jensen Huang, and Paul Graham/YC for color but the meat is his own workflow.

Mapping against Ray Data Co

Strength: medium. Adjacent and reinforcing, not new evidence and not load-bearing for an open question.

Specifically:

  1. Borrowable workflow patterns — yes, two:

    • The /pulse daily-metric-briefing pattern is exactly what we don’t have for HQ surfaces yet. We have check-board for Notion-backed work, morning-prep for calendar, but no daily plain-English briefing on the actual product/business metrics (Sanity Check open rates, Squarely traffic, MAC funnel, RDCO site analytics). Worth queuing as a Notion item: “design /pulse skill that collates per-surface KPIs into a morning briefing.” See also: existing morning-prep skill — /pulse would be its product-metrics sibling.
    • The “single roadmap doc, everything downstream is generated” pattern is structurally the same as hq-business-stack — but Moretti’s version is much tighter than ours because he forces one canonical artifact. Our equivalent today is split across the Notion Task Board, ad-hoc vault notes, and project-specific PRDs. Worth a self-review: is the critical-component-field discipline functioning as the “single roadmap” forcing function we want, or is it just decoration on a fragmented stack?

  2. Compound-engineering positioning — reinforcing. This piece sits next to Klaassen’s 2026-04-13-the-folder-is-the-agent and the rest of the Source Code corpus as more evidence for the agent-deployer thesis. Notable that Every is now extending this from engineering → PM → (presumably next) GM/founder roles. Cross-link: 2026-03-13-compound-engineering-camp-every-step-from-scratch, 2026-04-13-the-folder-is-the-agent.

  3. Tracked-author candidate. Marcus Moretti is new to our author graph. He’s GM of Spiral (Every Studio’s writing-style app). He writes from a product/operator lens, not engineering. Worth a low-friction add to the tracked-authors list — he’ll publish more of this and it’s directly in our adjacent-discipline beat. (Decision item below.)

  4. No contradictions. Nothing in this piece pushes against our existing positioning. The “skills over commands” feedback note we already operate by (feedback_skills_over_commands) is consistent with his slash-command-heavy approach.

  5. Anti-pattern flag — minor. He writes the piece as if the multi-platform busywork stack is the only alternative. That’s a strawman for solo operators but rings true for actual mid-size product teams. We should not lift his “consolidate everything into one chat” framing wholesale — that pattern works for a solo GM with one product, but breaks the moment you have two surfaces (which is already RDCO today: Sanity Check, HQ, Squarely, MAC). Treat his pattern as a building block, not the whole architecture.

Decisions surfaced