<h1>knowledgebase</h1><h2>Features</h2>It is basically a RAG App. In this app you can upload your Website &amp; Documents and have conversation.Checkout this demo:- https://knowledgebase-8u9.pages.dev/chat<h2>For Developers</h2>Directories <code>/api</code> - Express js app backend to handle documents parsing &amp; website parsing. <code>/ewc</code> - Botpress v12 chat embedding example for websites integration. (embedded web chat) <code>/ui</code> - Web ui to upload documents &amp; website + Chat<h2>Technical process</h2>Data Preparation<ol><li>Extracting text out of documents. Get html of the website &gt; cleaning html document &gt; extracting text &gt; cleaning text.</li><li>Embedding text to vectors with Gemini text embedding api.</li><li>Storing embedded vectors to Qdrant Vector store.</li></ol>Conversation<ol><li>Creating a function to get relevant information from vector store. Write function definition of the function &amp; feed to the LLM.</li><li>Prompt engineering the LLM to behave in a particular way &amp; handle basic queries like greeting without searching database.</li><li>From here LLM will handle the conversation.</li></ol>Handling complex queries / Fine tuningworking on it....

Explore the enhanced capabilities of a V2 full-stack Retrieval-Augmented Generation (RAG) application integrated with a knowledgebase for improved AI-powered interactions.

V2 Full stack RAG app with knowledgebase