AWS AI-DLC threw out the sprint

Everyone's bolting AI onto SDLC. AWS threw out the sprint instead. 👇

AI-DLC (AI-Driven Development Lifecycle) - AWS's methodology, open-sourced under MIT-0. Not a tool. A lifecycle.

SDLC: Plan → Design → Develop → Test → Deploy → Maintain Two-week sprints exist because human coordination is expensive.

AI-DLC restructures it into 3 phases - and every phase has explicit stages:

INCEPTION

Workspace Detection - greenfield or brownfield?
Requirements Analysis - AI asks clarifying questions
Workflow Planning - which stages this project actually needs ↳ then as needed: Reverse Engineering · User Stories · Application Design · Units Generation

CONSTRUCTION

Functional Design → NFR Requirements → NFR Design → Infrastructure Design (as needed)
Code Generation - business logic, API layer, data layer, tests, docs
Build & Test - build/packaging instructions + testing layers

OPERATIONS

Deployment & monitoring, with team oversight

The three details that make it a methodology, not a wrapper

It's adaptive - stages are skipped for simple projects, included for complex ones.
Humans validate in real time - "Mob Elaboration" in Inception, "Mob Construction" in Construction, instead of async backlog grooming and review queues.
Every decision, input and response is logged in an audit trail. And the vocabulary shift says it all: Sprint (2 weeks) → Bolt (hours to days) · Epic → Unit of work

The bottleneck moved: it's no longer writing code - it's how fast humans can validate what AI proposes. Accountability didn't move, though. You still can't defend a check-in with "the AI did this."

Sprints or bolts - where's your team today? 👇

Sources

AWS DevOps blog - building with AI-DLC: aws.amazon.com/blogs/devops/building-with-ai-dlc-using-amazon-q-developer
Open-source workflows (MIT-0): github.com/awslabs/aidlc-workflows
Concept: Raja SP, AWS