What it was for
AWS introduced AI-DLC in 2025 as a methodology for AI-native delivery: a three-phase model — Inception, Construction, Operations — in which AI proposes plans, requirements, code, and tests while humans validate at defined checkpoints, with practices like Mob Elaboration structuring the human-validation moments. It has since been open-sourced and wired into AWS's own tooling.
The verdict
EMERGES — as a serious, production-tested entry in the new category, and the one most worth reading. Through the Engine's lens, AI-DLC is a prescriptive implementation of the inner loop plus a disciplined intent-in interface: its checkpoints are gear-mesh engineering, its phases a workflow for keeping humans in the validator seat. Its acknowledged thinness is theory — heavy on prompts and folder structures, lighter on why — which is the gap a charter exists to fill.
What changes
For teams inside the AWS ecosystem, AI-DLC offers something this charter deliberately does not: an opinionated, tool-integrated starting procedure. The two compose rather than compete — the Engine explains the physics; AI-DLC ships one engine model. Its sequential, validate-each-stage structure is also more conservative than the concentric loops described here, which regulated teams may count as a feature.
The strongest objection
A vendor's methodology optimises for the vendor's platform. Conceded as a standing discount to apply — and the methodology's published form is open enough to audit, which is more than most of its predecessors offered.