What is Lumen

Lumen is an AI-powered assistant designed to facilitate quick queries of AWS documentation using Retrieval Augmented Generation (RAG), MCPs, and Amazon’s Nova Large Language Models (LLMs).

Use cases

Use cases for Lumen include quickly retrieving AWS documentation answers, assisting developers in understanding AWS services, and providing support teams with instant documentation access.

How to use

To use Lumen, you need an AWS account, Docker, ‘uv’, and ‘git’ installed. Set up your AWS account and ensure you have access to the required Bedrock models. Create a new IAM policy for Lumen to interact with AWS Bedrock foundation models.

Key features

Key features of Lumen include its ability to leverage advanced AI models for efficient documentation queries, integration with AWS services, and support for multiple AWS regions.

Where to use

Lumen can be utilized in various fields such as cloud computing, software development, and IT support, particularly for teams that frequently reference AWS documentation.

lumen

lumen is an AI-powered assistant designed to help you quickly query AWS documentation and get the answers you need. It uses Retrieval Augmented Generation (RAG), AWS Model Context Protocol (MCP) servers and Amazon’s Nova Large Language Models (LLMs).

Prerequisites

• An AWS Account
• Docker installed and running
• uv installed
• git installed

AWS Setup

To run lumen, you will need an AWS Account. If you don’t have one, you can find how-to steps here.

[!NOTE]
If you plan to use a different region from N. Virginia (us-east-1), don’t forget to change the selected region in the AWS Console, update the region part (e.g., us-east-1) in all resource ARNs within the IAM policy JSON, and the .env file. Double-check if the foundation models mentioned below are available in the selected region.

Bedrock Model Access

You need Access granted status for the following models:

• Titan Text Embeddings V2
• Nova Pro
• Nova Lite
• Nova Micro

More info on how to modify Bedrock model access can be found here.

Create new IAM Policy:

This IAM policy will be attached to a new IAM User, which lumen will use to make calls to AWS Bedrock foundation models.

1. Open the IAM Console
2. Click Policies on the left sidebar
3. Select Create policy
4. Switch the view from Visual to JSON
5. Paste the JSON from below and create the policy

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid": "AllowInvokeBedrockModels",
      "Effect": "Allow",
      "Action": "bedrock:InvokeModel",
      "Resource": [
        "arn:aws:bedrock:us-east-1::foundation-model/amazon.nova-micro-v1:0",
        "arn:aws:bedrock:us-east-1::foundation-model/amazon.nova-lite-v1:0",
        "arn:aws:bedrock:us-east-1::foundation-model/amazon.nova-pro-v1:0",
        "arn:aws:bedrock:us-east-1::foundation-model/amazon.titan-embed-text-v2:0"
      ]
    }
  ]
}

IAM User setup

At this point, you should have access to the four foundation models and an IAM policy ready.

1. Open the IAM Console
2. Click on Users on the left sidebar
3. Select Create user
   1. No need to check Provide user access to the AWS Management Console
4. Attach the policy you created in the previous step
5. Create the user and go to the new user’s overview page
6. Open the Security credentials tab
7. Create and save the access key

.env file

Now you have your Access key and Secret access key.

Rename the .env.template file to .env and update the following variables:

• AWS_ACCESS_KEY_ID=your_access_key
• AWS_SECRET_ACCESS_KEY=your_secret_key
• AWS_REGION=your_chosen_region (e.g., us-east-1)

Quick Start

Install uv. Then, follow these steps:

1. Clone the repository:

   git clone https://github.com/kaumnen/lumen
   cd lumen
   

2. Install dependencies using uv:

   uv sync --frozen --refresh
   

3. Run Qdrant:
   The docker run command for Qdrant mounts a volume for data persistence, meaning you will lose the vectors only if you explicitly delete them.

   docker run -p 6333:6333 -p 6334:6334 \
       -v "$(pwd)/qdrant_storage:/qdrant/storage:z" \
       qdrant/qdrant
   

4. Run lumen:
   From the root directory of the project, execute the following command:

   uv run streamlit run app.py
   

You can access the lumen app by navigating to the local URL provided in your terminal (usually http://localhost:8501) in your browser.

License