RAG Chatbot Demo Video - CloudWorld 2024

In a recent blog post we gave an overview of Retrieval Augmented Generation (RAG) as a way to leverage the natural language querying capabilities of Large Language Models (LLMs) on private or proprietary stores of unstructured data. A major use case for RAG is in the creation of chatbots which allow users to have a conversation with their documents, avoiding the potentially time-consuming task of searching through hundreds or thousands of files for a piece of information. As a demonstrator for this, Rittman Mead have used python and RAG to create a chatbot which searches a vector database of our own blogs, allowing the user to query hundreds of posts for information about the subject of their choice.

The video below was produced for CloudWorld 2024, and should be familiar to those who visited our stand. It gives a top-level summary of the steps followed to produce our RAG model, and shows the corpus of blogs, some glimpses of the code written to tokenise, chunk and embed those blogs as a datastore, and the chatbot in action from its early days on the command line to the working web interface. 

If your organisation could also benefit from the ability to converse or otherwise query your unstructured data using natural language, or you are interested in any of our other Data Science Services, then please feel free to get in contact with us at info@rittmanmead.com.