← Home

RAG Chatbot for Flower Shop Recommendations

Role

AI Engineer

Keywords

RAG

MongoDB

Vector search

Flask

API

Python

Year

2026

← Home RAG Chatbot for Flower Shop Recommendations Table of Contents About Introduction How it works Process Other Projects Let’s Work Together

About

Github

GitHub - maithtruong/flowershop_rag: A Naive RAG system that answers questions about products for a flower shop. This is an assignment for the RAG class at ProtonX.

A Naive RAG system that answers questions about products for a flower shop. This is an assignment for the RAG class at ProtonX. - maithtruong/flowershop_rag

https://github.com/maithtruong/flowershop_rag

Tech Stack

OpenAI

OpenRouter

MongoDB

ngrok

Flask

Python

Introduction

A simple, naive RAG chatbot that recommends products based on flower shop data. It uses a free LLM, vector search using a Vietnamese embedding model, and simple UI using Streamlit.

Translation of the above image:

User: I would like to buy flowers for by boyfriend. I prefer a cheap one, below 20 USD.

Chatbot: Hi! We currently have a few bouquets such as Baby M107 …

How it works

Basically, the system works as follows:

Extract texts about product information, then store in MongoDB Atlas

Create vectors for each product using MongoDB Atlas’s native function

The chatbot is actually an LLM Wrapper. On inference, the system will:

Extract the user’s input prompt, vectorize it, then perform vector search across the MongoDB database.
Retrieve the related product(s) information.
Modify the user’s input prompt by adding the retrieved information and a command for the chatbot to answer according to the new retrieved product context.
Repeat the process for new inputs.