A sample of an AI Driven Development Lifecycle (AI-DLC) platform built on AWS. It takes a user's project idea and autonomously guides it through the full SDLC — from requirements gathering and architecture design through code generation and testing — using a multi-agent system built on AWS Bedrock AgentCore and the Strands Agent SDK.
Users interact via a React web UI. They describe a project, answer clarifying questions (Human-in-the-Loop), and the system produces structured artifacts: requirements docs, architecture designs, user stories, code, and test results — all stored in S3 and optionally synced to Atlassian Confluence/Jira.
This sample project showcases the capabilities of the AI Driven Development Lifecycle (AI-DLC) methodology and how it can be integrated into a platform. It demonstrates how to leverage Agentic AI to revolutionize both software architecture design and development — generating standardized patterns, optimizing configurations, ensuring compliance, and creating documentation.
AI-DLC is an AI-centric transformative approach to software development that emphasizes two powerful dimensions:
- AI-Powered Execution with Human Oversight: AI systematically creates detailed work plans, actively seeks clarification and guidance, and defers critical decisions to humans. This is critical since only humans possess the contextual understanding and knowledge of business requirements needed to make informed choices.
- Dynamic Team Collaboration: As AI handles routine tasks, teams unite in collaborative spaces for real-time problem-solving, creative thinking, and rapid decision-making. This shift from isolated work to high-energy teamwork accelerates innovation and delivery.
💡 For a deep dive on the concepts, see the AI Driven Development Lifecycle blog post.
Who is this for? Product owners, project managers, architects, and engineering leaders who want to reduce the operational burden of a standard development lifecycle and explore how Agentic AI can accelerate cloud-native software delivery.
The AI-DLC Platform follows a layered architecture that separates concerns and enables scalability. User interactions flow from the Portal (or an external IDE like Kiro) through an orchestration layer to specialized AI agents, which leverage MCP (Model Context Protocol) tools to perform specific tasks. All artifacts live in Amazon S3 as the central repository for project state and generated work products.
Key components:
- Orchestrator: 14-node Strands GraphBuilder graph covering inception and construction stages
- Agents: Stage-specific specialist agents (requirements, design, code generation, build & test, etc.)
- Gateway: AWS Bedrock AgentCore Gateway with MCP protocol — wraps Lambda tools for file ops and integrations
- Memory: AWS Bedrock AgentCore Memory (summarization + semantic)
- Frontend: React 18 + Vite + TypeScript, deployed to S3 + CloudFront
- Auth: Amazon Cognito (PKCE for UI, client credentials for agent ↔ Gateway)
See code/docs/ARCHITECTURE.md for full architecture details.
- AWS account with administrator access
- AWS CLI configured (
aws configure) - Node.js 20+
- Yarn (
npm install -g yarn) - Docker running locally (for building the ARM64 agent container)
- AWS CDK v2 (
npm install -g aws-cdk)
-
Clone the repo
git clone <repo-url> cd AIDLC-Platform-Solution
-
Install dependencies
yarn install
-
Bootstrap CDK in your AWS account (first time only)
cd cdk cdk bootstrap cd ..
-
Deploy the stack
yarn deploy-infrastructure
This packages all assets (agent container source, UI build, Lambda zips), uploads them to an S3 assets bucket (
aidlc-assets-<account>-<region>), and deploys theAIDLCPlatformWorkshopCloudFormation stack. Full deployment typically takes 15–20 minutes (the CodeBuild job for the ARM64 agent container is the long pole).Optional: link an external IDE. If you want the UI to display an "Open IDE" button (pointing to VSCode Server, Cloud9, or any URL), pass it via CDK context:
yarn workspace cdk cdk deploy \ --require-approval never \ -c assetsBucketName=aidlc-assets-<account>-<region> \ -c ideUrl=https://your-ide.example.com
Omit
-c ideUrl=...(or use the defaultyarn deploy-infrastructure) to hide the button. -
Grab the outputs
When deployment completes, the stack prints:
Output What it is AIDLCPlatformURLCloudFront URL for the web UI TestUserEmail/TestUserPasswordPre-provisioned Cognito test user
yarn delete-infrastructureThe assets bucket (
aidlc-assets-<account>-<region>) is created outside the stack, socdk destroywill not remove it. Delete it manually if you no longer need the packaged zips.
Once the stack is deployed, open the AI-DLC Platform and create your first project.
Open the AIDLCPlatformURL output in your browser — this is the CloudFront-hosted portal.
Click Sign In and enter the TestUserEmail and TestUserPassword values from the stack outputs.
After logging in, click Create Project, enter a name and description, and follow the on-screen prompts. The agents will walk you through requirements gathering, user stories, architecture design, and code generation — pausing at each stage to let you review and approve artifacts before moving on.
Once the Inception phase is complete, the platform will have generated a full set of design artifacts — requirements, user stories, workflow plans, and application architecture — all stored in S3 and visible in the portal's document viewer.
Take these artifacts to your preferred IDE — Kiro, Claude Code, or any AI-assisted development environment — and start the Construction phase. The design documents provide the structured context that coding agents need to generate implementation code, infrastructure templates, and tests aligned with the decisions made during Inception.
For a guided, step-by-step walkthrough — including recommended prompts, expected Kiro CLI responses, and deeper context on each AI-DLC phase — check out the full workshop:
Contributions are welcome. Please open an issue to discuss substantial changes before submitting a pull request.
See LICENSE for details.