What is AI-DLC? AWS’s AI-driven lifecycle
AI-DLC is the AI-Driven Development Lifecycle, a methodology that Raja SP and his AWS team introduced on 31 July 2025. AI drafts a plan, asks clarifying questions and implements only after the team validates it, in sessions AWS calls “Mob Elaboration” and “Mob Construction”, and sprints become “bolts” of hours or days. AWS says these rituals keep AI’s suggestions from being blindly accepted; vibe coding accepts them unread.
Origin Raja SP, a Principal Solutions Architect at AWS, introduced AI-DLC on 31 July 2025 in “AI-Driven Development Life Cycle: Reimagining Software Engineering” (opens in a new tab), and AWS open-sourced its adaptive workflows on 29 November 2025 in “Open-Sourcing Adaptive Workflows for AI-Driven Development Life Cycle (AI-DLC)” (opens in a new tab).
- In Polish
- AI-DLC
- Closest ladder level
- Level 4, free guide
- Also searched as
- AIDLC, AI DLC, AI-Driven Development Lifecycle, AI-Driven Development Life Cycle, aidlc-workflows
Then $19.99 a month or $99.99 a year. Cancel from your account page. 30-day refund on a first purchase.
Why it matters.
AI-DLC gives a team a vocabulary and an open-source runtime, and the part a CTO should read is the audit trail. The v2.11.0 README (read on 10 October 2026) lists 5 phases, 33 stages and a 118-event audit trail, and this site’s framework comparison finds it the only framework in its group that pairs human approval gates with an append-only audit trail. Bolts and Units of Work are vocabulary; the mob sessions are where suggestions get checked. AWS’s own posts describe three phases, so say which description you mean.
AI-DLC vs vibe coding
AWS never uses the words “vibe coding” in its three AI-DLC posts (searched on 10 October 2026); its nearest contrast is with AI-autonomous development.
| Question | Vibe coding | AI-DLC |
|---|---|---|
| What it names | A way of working: accept what the model writes without reading the diff | A named AWS methodology with its own phases, rituals and vocabulary, plus an open-source runtime |
| Who writes the code | The model | AI, after the plan is validated |
| Who checks it | Nobody reads the diff; the human watches the behaviour | The team, in Mob Elaboration and Mob Construction sessions |
| What stops the loop | The app appears to work | Human validation before the AI implements; the open-source workflow gates every stage except initialization |
| Where the name came from | Andrej Karpathy, post on X, 2 February 2025 | AWS, Raja SP, blog post of 31 July 2025 |
| Where it breaks | Beyond throwaway work, nothing replaces the reader | Rituals without verification: AWS warns that developers “often drift into passive execution” |
AI-DLC phases and rituals, in AWS’s words
AWS’s 2025 description has three phases, each with a point where the team validates what the AI proposed.
| Phase | What the AI does | What the team does |
|---|---|---|
| Inception, “Mob Elaboration” | “transforms business intent into detailed requirements, stories and units” | “actively validates AI’s questions and proposals” |
| Construction, “Mob Construction” | “proposes a logical architecture, domain models, code solution and tests” | “provides clarification on technical decisions and architectural choices in real time” |
| Operations | “applies the accumulated context from previous phases to manage infrastructure as code and deployments” | Oversight: “with team oversight” |
awslabs/aidlc-workflows has five phases (Initialization, Ideation, Inception, Construction, Operation) and 33 stages at v2.11.0, read on 10 October 2026.How each tool does it.
Cursor, Claude Code and Codex each name this their own way. Each note carries the date it was checked.
Cursor
aidlc config --harness cursor installs 14 personas as subagents, skills, .cursor/hooks.json and an AGENTS.md section; start with /aidlc.
AI-DLC and the other frameworks with audit trails (Cursor), 2026-09-26
Claude Code
aidlc config --harness claude installs the runtime; /aidlc starts the workflow in the session.
AI-DLC and the other frameworks with audit trails, Claude Code 2.1.283, 2026-09-26
Codex
aidlc config --harness codex needs codex-cli 0.145.0 or later; start with $aidlc, not /aidlc.
AI-DLC’s gates depend on hooks, and Codex never runs untrusted project hooks: at the hooks dialog choose Trust all and continue, or merge the [hooks.state] block from the generated .codex/trust-seed.toml into $CODEX_HOME/config.toml.
AI-DLC and the other frameworks with audit trails, Codex CLI 0.157.1, 2026-09-26
Questions about AI-DLC.
How is AI-DLC different from vibe coding?
AWS does not use the words “vibe coding” in its AI-DLC posts. Its nearest contrast is “AI-autonomous development, where AI is expected to generate entire applications without human intervention”, and it says its mob rituals “ensure that AI’s suggestions are not blindly accepted.” Vibe coding in Karpathy’s original sense is blind acceptance, and this site reads AI-DLC as built to stop it.
Is AI-DLC the same as AI SDLC?
No. AI SDLC is the generic label for any software lifecycle redesigned around agents, and nobody owns it. AI-DLC is a named methodology from AWS with its own vocabulary (bolts, Units of Work, Mob Elaboration, Mob Construction) and an open-source implementation, awslabs/aidlc-workflows. Put simply, AI SDLC is the category and AI-DLC is one AWS-branded method inside it.
What are bolts, Mob Elaboration and Mob Construction?
They are AWS’s terms. “Bolts” replace sprints: “shorter, more intense work cycles measured in hours or days rather than weeks”. In “Mob Elaboration” the whole team validates the AI’s questions and proposals during Inception; in “Mob Construction” the team clarifies technical and architectural choices in real time during Construction.
How many phases does AI-DLC have?
Three in AWS’s 2025 description: Inception, Construction and Operations. The open-source workflow is larger: its v2.11.0 README, read on 10 October 2026, counts 5 phases and 33 stages, and its introduction guide names the phases (Initialization, Ideation, Inception, Construction, Operation). The first reading is the method as AWS introduced it on 31 July 2025; the second is the tool as it stood on 10 October 2026.
The vocabulary of AI-driven development
Every name below is defined against vibe coding: who writes the code, who checks it, and what stops the loop.
Read the long-form guide: agentic engineering vs vibe coding
Sources.
The primary sources outside this site that this page relies on.
- AI-Driven Development Life Cycle: Reimagining Software Engineering (opens in a new tab) Raja SP, AWS DevOps & Developer Productivity Blog
- Open-Sourcing Adaptive Workflows for AI-Driven Development Life Cycle (AI-DLC) (opens in a new tab) Will Matos, Raj Jain, Siddhesh Jog and Raja SP, AWS DevOps & Developer Productivity Blog
- awslabs/aidlc-workflows: README of release v2.11.0 (opens in a new tab) AWS Labs, GitHub
Keep reading.
The guides that go deeper, and the terms and comparisons next to this one.
In the docs
Related terms
Comparisons
Read the guides in the same words.
Open every guide with the 7-day free trial. Each term is defined once here, and the guides use it the same way.
Then $19.99 a month or $99.99 a year. Cancel from your account page. 30-day refund on a first purchase.