No Server RAG

No Server RAG is a full-stack, serverless Retrieval-Augmented Generation (RAG) application deployed entirely on AWS. It enables secure, scalable, and maintenance-free AI-powered Q&A over your private documents using large language models (LLMs) from Amazon Bedrock.

Built using AWS CDK, Lambda, S3, LanceDB, and a modern React frontend, this sample provides a low-cost, production-ready foundation for enterprise-grade document chat experiences — with zero server management.

---

Demo

---

📐 Architecture

A high-level view of the system components:

This system uses:

• Amazon Bedrock for LLM inference and embeddings
• Amazon S3 for document storage
• LanceDB (in-serverless mode) for vector search
• Amazon Cognito for authentication
• AWS Lambda + API Gateway for compute and messaging
• Amazon CloudFront for global frontend delivery

---

✨ Features

• chatplayground – Ask questions and view LLM responses alongside retrieved document chunks.
• chathistory – View and manage past chat sessions, forward selective messages to test/debug prompts.
• serverlesskb – Store and retrieve documents via LanceDB + S3, eliminating the need to manage vector DB infrastructure.
• promptmanagement – Allow users to override system prompts in the UI.
• dynamicprompt – Change default prompt templates globally by editing prompt-templates.yml and running a script.

---

⚙️ Prerequisites

Ensure the following are installed:

• Node.js v18+ (use nvm)
• Docker
• AWS CDK v2 (>= 2.142.1)
• AWS credentials (configured via aws configure)
• Access to Amazon Bedrock in your target region

---

🌍 Supported AWS Regions

This project supports AWS regions where Amazon Bedrock and embedding models are available:

us-east-1, us-east-2, us-west-2, eu-central-1,
eu-west-2, eu-west-3, ap-south-1,
ap-southeast-2, ap-northeast-1, ca-central-1, sa-east-1

---

🧾 Embedding Configuration

To set your default embedding model, update lib/llm-config.json:

{
  "us-west-2": {
    "embedding": {
      "model": "amazon.titan-embed-text-v1",
      "size": 1536
    }
  }
}

⚠️ Do not change the embedding model after deployment — it may result in incorrect or inconsistent retrieval behavior.

---

📦 Installation

nvm use        # Ensure Node.js 18 is active
npm install    # Install project dependencies

---

🚀 Deployment

Recommended deployment region: us-west-2

Step 1 – Deploy the stack

cdk deploy

Step 2 – Save output values

After deployment, you'll get several important outputs:

• WebDistributionName: URL to access the frontend
• FrontendConfigS3Path: S3 path for frontend config JSON
• Cognito credentials and authentication details

---

🧪 Running Frontend Locally

1. Fetch Frontend Configuration

./fetch-frontend-config.sh ServerlessRagOnAwsStack
# OR manually:
aws s3 cp s3://.../appconfig.json ./resources/ui/public/

2. Start Local Dev Server

cd resources/ui
npm install
npm run dev

This will run the Vite-powered React app on http://localhost:5173.

---

🔁 Changing Default Prompt Templates

To update the default system prompt used by all users:

1. Open and edit prompt-templates.yml
2. Save changes
3. Run:

npx ts-node update-default-prompt-templates.ts

⚠️ You may need to pass AWS region or profile using environment variables, depending on your setup.

---

🧠 Example Configuration File

This file (appconfig.json) is used by the frontend to connect to backend resources:

{
  "inferenceURL": "https://xxxxxxxxxxxxx.lambda-url.us-west-2.on.aws/",
  "websocketURL": "wss://xxxxxxxxxx.execute-api.us-west-2.amazonaws.com/Prod",
  "websocketStateTable": "ServerlessRagOnAwsStack-websocketStateTable-xxxxxx",
  "region": "us-west-2",
  "bucketName": "ServerlessRagOnAwsStack-documentsbucket-xxxxxxxxx",
  "auth": {
    "user_pool_id": "us-west-2_XXXXXXXXXX",
    "aws_region": "us-west-2",
    "user_pool_client_id": "XXXXXXXXXX",
    "identity_pool_id": "us-west-2:XXXXX-XXXX-XXXXXX",
    "standard_required_attributes": ["email"],
    "username_attributes": ["email"],
    "user_verification_types": ["email"],
    "password_policy": {
      "min_length": 8,
      "require_numbers": true,
      "require_lowercase": true,
      "require_uppercase": true,
      "require_symbols": true
    },
    "unauthenticated_identities_enabled": true
  },
  "version": "1",
  "storage": {
    "bucket_name": "ServerlessRagOnAwsStack-documentsbucket-XXXXXXXXXXX",
    "aws_region": "us-west-2"
  }
}

---

📄 License

This project is open-sourced under the MIT License.

---

🙌 Acknowledgments

Built with love using Amazon Bedrock, AWS CDK, and LanceDB.

---

No Server RAG – A zero-server RAG application template for private, production-ready document Q&A.