v5.0 — The
Decision System Release

A Decision System,
Not a Methodology

A product organization is a decision system. The quality of its decisions is what makes it world-class. Shipped output is the consequence of good decisions — not the goal itself. v5 reorients the entire OS to optimize for decision quality and provenance, so the work product you can actually trust is the reasoning trail, not just the artifact.

• ✓Decisions become the unit of product leadership — recorded, not lost in chat
• ✓Output is treated as the downstream consequence of a decision, not a substitute for one
• ✓The OS optimizes for decision quality and provenance end to end
• ✓Surface skills: /decision-record, /decision-charter, /strategic-bet

Written from the top seat of the product organization, the OS speaks to every altitude: the Chief Product Officer in the seat, the VP of Product on the line to it, the Product Leadership Team and Directors, the C-suite peers across the table, and the Director track rising toward it.

Custodianship Stays
Human-Held

The OS keeps custodianship human-held. The agent records a decision, but the record is only closed when the named accountable owner affirms it. AI drafts the decision; the human owns the call — none of it works without a human in the seat.

• ✓The agent drafts the decision record; only a named human affirms it to closed (Layer-4 attestation)
• ✓Every decision is surfaced to you for sign-off — never self-marked "accepted" by the agent
• ✓Custodianship stays human-held as AI participation deepens
• ✓Records stay affirmable, auditable, and resumable — whether a human or a model produced the analysis

The Decision
Provenance Standard

This is the headline of v5. The book is the architecture; the Standard is the bridge; the Product Org OS is the reference implementation. The Standard bridges human custodianship and the AI tooling participating in the decision system, through records that are affirmable, auditable, and resumable — with a named-human attestation gate. v5 meets the Decision Provenance Standard's highest conformance level (Level 3).

• ✓Meets the Standard's highest conformance level (Level 3)
• ✓Custodianship stays human-held, with a named-human attestation gate
• ✓Decision records are affirmable, auditable, and resumable
• ✓The Decision Interface Charter is the central decision instrument — examples include Data, Engineering, PMM, and Customer-Outcome interfaces

Onboards Any Coding Agent.
One Link.

The agent guide now makes the OS operate itself. Point any coding agent — Claude Code, Codex, Cursor, Copilot, Gemini CLI, or Windsurf — at one URL. It installs the OS, shows you the inventory, offers to wire itself into your CLAUDE.md, and spawns the right agent when you ask.

• ✓Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and Windsurf
• ✓Installs, then shows you the full skill and agent inventory
• ✓Offers to wire itself into your CLAUDE.md
• ✓Spawns the right agent when you ask — no manual routing

Bigger.
Deeper.

The Decision Provenance Standard is what makes v5 a major version. The depth is what makes it land: v5 ships with 143 skills (up from 133) and 35 knowledge packs (up from 28). The new packs go deep on AI and specialist domains, so agents reason from current, authoritative sources instead of stale training data.

• ✓143 skills — up from 133 in v4
• ✓35 knowledge packs — up from 28 in v4
• ✓New AI packs: value-score-design, mcp-architecture, agentic-security
• ✓New specialist packs: generative-ui, ai-agent-supervisor, ai-research-synthesis