Cromax

Cromax

Cromax : An AI service powered by LLMs, enabling businesses to interact with real-time data for demand forecasting and shop setup, while providing consumers with personalized product recommendations.

Created on 2nd October 2024

Cromax

Cromax

Cromax : An AI service powered by LLMs, enabling businesses to interact with real-time data for demand forecasting and shop setup, while providing consumers with personalized product recommendations.

Describe your project

Cromax is a dual-utility platform offering business and customer-focused services. On the business side, an AI-powered chatbot leverages natural language processing (NLP) and machine learning algorithms for shop setup assistance. It extracts location and domain from user inputs, analyzes product demand, competition, and demographics by integrating real-time data from platforms like Reddit and YouTube. Deymax ranks potential locations based on demand scores, competitor density, and customer demographics. It also visualizes optimal locations using geospatial APIs, displaying dynamic maps with zoom functionality for specific regions. Users can export market data, including demand scores and coordinates, for deeper offline analysis. The Product Interest Forecasting application employs real-time social media data for predicting brand or product demand. By integrating platforms like Reddit and YouTube, the app generates interest forecasts over a user-defined period using Facebook’s Prophet model. The forecasting system processes and normalizes engagement data, visualizing historical and projected trends in a graph, enabling data-driven decisions regarding inventory and marketing strategies. In the Analytics Tab, customer purchase probability is modeled using key attributes such as loyalty points vs. transaction frequency. A histogram visualizes transaction frequency based on loyalty point ranges, identifying trends in purchase behavior. This aids businesses in optimizing loyalty programs and customer retention strategies. On the customer side, Cromax utilizes Google’s Gemini AI SDK to power a chatbot-based product recommendation system. Product data scraped from the Croma website, including names, prices, and links, is processed through Gemini AI, generating personalized recommendations based on user inputs like price range or product type. The system provides real-time responses, with links to relevant products, ensuring a streamlined, data-driven shopping experience

Challenges we ran into

Developed during a hackathon by Croma and Google, this platform helps users establish new businesses by offering insights on optimal store locations, competitor analysis, and warehouse listings.

  1. Tech Stack Discrepancies
    Our team faced a split in tech stacks, with the front-end using Node.js/React.js and the back-end in Python. We used Flask APIs to integrate them seamlessly.
  2. Chatbot & Interactive Map
    The chatbot provides business location recommendations based on user queries, while interacting with an interactive map (powered by Geoapify and Leaflet.js) to display relevant markers and adjust zoom levels.
  3. Competitor Analysis
    We developed a competitor analysis feature that identifies similar businesses in a selected area, but full integration was postponed due to time constraints.
  4. Data Collection
    Data for warehouse and plot listings were scraped from Magicbricks.com and stored in Supabase. For product recommendations, we collected data from the Croma website and stored it in CSV format.
  5. Optimization
    During testing, we noticed performance issues with large datasets. Our future plan is to move from CSV to Supabase for efficient querying and adopt a multimodel approach for better data processing.
  6. Demand Forecasting
    Using Python’s Prophet library, we built a demand forecasting tool to predict future product trends based on social media sentiment analysis from platforms like YouTube and Reddit.
    7.Analytics Dashboard
    We developed an analytics dashboard for shopkeepers to track customer behavior, with future plans to add user profile generation and deeper insights.
    8.Unfinished Features
    Some features, such as full competitor analysis, sentiment-based forecasting, multiprocessing models, and distinct user/shopkeeper dashboards via Firebase, are still in the pipeline.
    Despite challenges, we created a functional prototype with multiple features and learned valuable lessons in tech integration and project management.

Tracks Applied (1)

12. Problem statement shared by Croma

Our solution leverages Google’s Gemini API for Generative AI (GenAI) to revolutionize retail operations by integrating r...Read More

Discussion

Builders also viewed

See more projects on Devfolio