What's shipping this week

Status: drafted · Time: 3 min · Audience: everyone Outcome: Understand that the playbook is alive, what the current version contains, and what is still coming.

This is a living program

The playbook is not a frozen manual. It is closer to a product surface: versioned, reviewed, changed when the tools change, and sharpened when real cohorts hit friction.

That matters because AI tooling moves quickly. Claude Code changes. The Compass plugin changes. Skills appear. Connectors improve. Security guidance tightens. A static manual would be stale before the second cohort finished reading it.

So every version of the playbook should answer one question clearly: what can I trust in this version?

What the current version gives you

The playbook is drafted end-to-end and hosted as a live site at razorpay.github.io/ai-playbook.

Drafted and readable now:

Part 0 — Foundation: Tech 101 and Ops 101 end-to-end.
Prologue: twelve chapters including the 5-layer mental model, the 9-layer Enablement Stack, the people directory, operating principles, the safety brief, and this page.
White Belt: twelve modules, Quest W-0, Quest W-1, Boss Fight W-B, and the badge.
Yellow Belt: fourteen modules, Quest Y-1, Quest Y-2, Boss Fight Y-B, and the badge.
Green Belt: twenty-eight modules across Craft, Practices, and Guardrails, both quests, Boss Fight G-B, and the badge.
Black Belt: sixteen modules across Platform, Craft, and Org, both quests, Boss Fight B-B, and the badge.
Staff+ Council: six chapters and the charter template.
Appendices A through N, with Appendix E kept as a drafted skeleton and Appendix M still planned. The drafted set includes the live Slack channel directory (F), the named role-holder roster (in §0.6), the Day-1 quick reference card (H.7), and the root changelog as Appendix K.
Seven Claude Code reference definitions: playbook-course, setup-verify, pre-ship-check, blade-compliance-reviewer, production-compiler, design-intel, security-review-subagent. They document intended workflows; confirm the current runnable distribution in #ai-help.
The Astro Starlight hub at razorpay.github.io/ai-playbook, auto-deployed on every merge to master with PR-preview builds.
Nine signature SVG diagrams, two legacy SVG aliases, and seven hand-drawn Excalidraw companion illustrations.

Still in motion:

Audio summaries (~11 podcast-style chapter walkthroughs) — slated for a future release, see ROADMAP.md.
Validation sweep with security, engineering, and design reviewers — runs as calendar work, not authoring.
Appendix M — visible scaffold for the Ship-to-Learn operational template, not drafted yet.
The eight open program questions in ROADMAP.md — each gets an explicit answer before v1.0.

The version history lives in CHANGELOG.md. What is coming next is in ROADMAP.md.

How to read changes

When a future version lands, look for three things:

New drafted chapters or reference workflows. These are safe to read end-to-end.
Changed safety guidance. Treat this as required reading.
New quests, boss fights, or scoring criteria. These change what counts as evidence.

Everything else is supporting structure.

What you should carry forward

The playbook is drafted end-to-end and live on a hosted hub; it is not the final playbook. Versions iterate; see CHANGELOG.md for the running log.
The Markdown in this repo is the source of truth for the hub at razorpay.github.io/ai-playbook and the in-repo playbook-course reference definition. A source definition and a runnable distribution are separate contracts.
Safety changes (Prologue §0.11, Green Belt Part C) matter more than ordinary wording changes.
If you spot something stale, ping #ai-help with the stale line and the better source — staleness is a bug, not a fact of life.

Previous: ← 0.11 The safety brief · Next: → White Belt