Inspiration
Stock control in large retail chain businesses within the e-commerce sector can be cumbersome and time-consuming especially as the size of such business grows. Retrieving relevant information from stock inventory and sales records, often stored in CSV documents, by old and mundane methods often presents a challenge. Most evident is the lack of proactivity in restocking out-of-stock products which leads to loss of potential revenue and eventually loss of customers to competitors. All as a result of constant oversight caused by inefficiency in FairyTale PLC's current stock control system. Thus the need for a more efficient and user-friendly solution.
What it does
To address this issue, the "FairyTale PLC AI_based stock management system" demo (minimum viable product) application leverages AI through google's latest LLM -gemini 1.5 pro and the LangChain library to simplify the old process of stocking taking and retrieval. Takes it a notch higher by merging and comparing sales data with inventory data and allows for efficient stock control by parsing through each and every record of the products in the store, allowing client query records in conversational user convenient format, and receive meticulously accurate responses in a streamlined manner.
How we built it
It is a python application built that leverages AI by enabling users to upload a stock-taking inventory record CSV alongside sales records, query the records by asking questions in a conversational format to the records in the document, and receive responses using the LangChain library. It is implemented using google large language model as its backbone - 'Gemini 1.5 latest' and streamlit to build its user interactive UI.
Challenges we ran into
- Conceptualization
- Managing large output tokens
Accomplishments that we're proud of
- Conceptualization and eventual implementation of minimum viable product
What we learned
When it comes to LLMs, I am particularly biased towards using a particular one for all the projects I have; but with this hackathon, I have learnt that other LLMs are fast rising and are in constant competition in terms of performance. I have learned to look beyond my bias and pick a suitable language model for my projects based on my feature tradeoffs.
What's next for FairyTale PLC AI_based stock management system
While working on this minimum viable product, I also had lots of upgrade ideas and new features to make the product even more appealing and scalable in the market. Whats next? Product improvement!
Log in or sign up for Devpost to join the conversation.