Simple code I wrote a while back to build a conversational chatbot using any LLM API which uses historic conversation for context and intent
Uses standard RAG model with cosine similarity, SQLite for history, FAISS for vectordb, html chatbot served using REST API
Uses Cosine similarity threshold for intent matching to determine if the user query matches a predefined intent in the FAISS vector database
Chat history is summarized in reverse chronological order
Just to show how easy it is to build these apps PoC MVP, functionality using LLMs and coding tools. I used OpenAI API, Grok and Gemini for coding
Done before Cursor, Claude and other AI coding IDE’s existed, so even more easier now with just prompting them to write the entire code and deploy. Doesn’t always work fully so requires manual installation, stitching together code, etc
Code in repo – https://github.com/datawisdomx1/conversationalchatbot
Hope it helps to give some basic ideas and code logic for PoC and MVP
