SDLC for Agentic AI Engineering

Source: Brett Luelling, cited in Medium Daily Digest of 2026-04-07 (Email 23). 3 min read, 186 claps.

Key takeaways

  • Adapts traditional SDLC (Software Development Life Cycle) practices for agentic AI workflows.
  • Bridges the gap between conventional software engineering process and the new realities of AI-agent-driven development.
  • Relevant for any organisation rolling out Claude Code / autonomous orchestrators to teams with existing SDLC governance.

Why it matters

Harris / Constellation teams have established SDLCs, code review processes, release gates, and compliance checkpoints. Most don’t yet have a written adapted-for-agentic-AI version.

Luelling’s article is a starting framework — adapt existing process rather than replace it. Relevant items:

  • Where does /bad fit in your sprint cycle?
  • How do you code-review AI-generated PRs in bulk?
  • Where does adversarial review (Codex plugin) sit in your SDLC?
  • How does Managed Agents change your deployment pipeline?

Built by Force Information Systems · Harris Computer · Constellation Software. Licensed under MIT.