Easiest Way to Use Agentic RAG in Any Enterprise - RAGapp - Install Locally

---

---

AI Nuggets

Instructions for Setting Up RAG Locally

Prerequisites:

• Operating System: Ubuntu 22.4
• GPU: At least 16 GB of VRAM recommended for AMA models (22 GB used in the video)
• Docker: Ensure Docker is installed (version 26.1.3 mentioned)
• Conda: Recommended for creating a virtual environment

Steps:

1. Create a Conda Virtual Environment:

   conda create -n <environment_name>  

   • Press y to confirm.
   • Activate the virtual environment:

     conda activate <environment_name>  

2. Clone the GitHub Repository:
   • Clone the repository (URL provided in the video description):

     git clone <repository_url>  

   • Navigate to the cloned directory:

     cd <cloned_directory>  

3. Run the Docker Command:
   • Execute the Docker command to run the rag app:

     docker run -p 8000:8000 <docker_image_name>  

   • Ignore any warnings if the image is already downloaded.
4. Access the Application:
   • Open the application by navigating to localhost:8000 in your web browser.
5. Configure the Application:
   • Select the desired AI provider (OpenAI, GPT-J, or AMA).
   • Enter your API key from the provider’s platform (e.g., platform.openai.com for OpenAI).
   • Click on “Update” to save the configuration.
6. Upload Your Data:
   • Provide a prompt or context for the model.
   • Upload documents or files that contain the data you want the model to use.
   • Optionally, you can use tools to fetch data from Wikipedia, DuckDuckGo, or other websites.
7. Test the Application:
   • Enter a query and observe the retrieved context and generated response.
   • The application will display the sources used for generating the response.
8. Distribute the Application to Users:
   • Click on “Start the app” to open it in another browser window.
   • Share the URL with users for them to access the application.

Additional Tips:

• Kubernetes Deployment:
  • The application is container-based and can be run on Kubernetes.
  • It can be exposed through a load balancer in cloud environments like AWS EKS, Azure AKS, or Google GKE.
  • Note: The application does not include authentication or authorization, which needs to be implemented separately.
• Security Considerations:
  • To secure the application, you can use a load balancer, CloudFront, API Gateway, or integrate with identity access management systems.

Video Description Links:

• The video description contains the link to the GitHub repository for the rag app.
• The API key for OpenAI can be obtained from platform.openai.com.

Upcoming Content:

• A subsequent video will demonstrate how to use the application with AMA once integration issues are resolved.

Call to Action:

• The video encourages subscribing to the channel and sharing the content with others.

(Note: The exact URLs and commands are not provided in the transcript and are expected to be found in the video description or by following the instructions in the video.)