AI-SDLC GOVERNANCE PLAYBOOK
I am on:
Enterprise Standard Operating Procedure · v1.0

The AI-SDLC Governance Playbook

Human-approved, AI-accelerated delivery — from requirement to root cause analysis.

This defines how Product, Engineering, QA, DevOps, Security, and Support use Claude and ChatGPT across the software delivery lifecycle. Tap "I am on" above to highlight exactly where your role owns a decision.

01

AI assists humans; it does not replace engineering ownership.

02

Every architectural, security, and production decision requires human approval.

03

AI-generated output is treated as a draft until reviewed.

04

Every phase has defined inputs, outputs, and approval gates.

The Gate Rail

10 phases, 9 mandatory human checkpoints. Click any node to jump in.

Responsibility Matrix (RACI)

Filter by role above, or read straight across a phase to see who owns what.

R — Responsible A — Accountable C — Consulted I — Informed

Golden Rules

Recommended Next Steps

This SOP is concise enough for day-to-day use while keeping accountability explicit. To operationalize it beyond this page:

  1. Publish a Confluence space with one page per SDLC phase.
  2. Create reusable Claude Project instructions / ChatGPT project prompts per phase.
  3. Attach role-specific checklists to JIRA workflows and pull request templates.
  4. Export a designed PDF with diagrams for offline/print distribution.
knowledgewala.com