Snap&Find

• 

  logo

Inspiration

We wanted to see if we can build a convenient way to check for products in a supermarket chain

What it does

It checks for availability of products in different stores of a supermarket chain based on an image input. It use a CSV RAG tool to search for the items based on the text output of the image to text feature of the LLM. It also provides facts about that product based on internet search.

How we built it

Using a Databricks notebook, with data in a csv and chroma vectordb stored in Databricks volume. Google Generative AI for the model. CrewAI and Langchain for the agents and the tools.

Challenges we ran into

Find the right dataset and getting an agent and the tools to work seamlessly.

Accomplishments that we're proud of

Building an app that works

What we learned

The Databricks Platform

What's next for Snap & Find

Apply it to an eCommerce platform so that customers can add to cart the products of their choice using their images from their phones.

Built With

• crewai
• gemini
• gradio
• langchain
• python