Cross Selling Products using LLM driven Machine Learning

Cross Selling Products using LLM driven Machine Learning

The Challenge

A financial company wanted to develop a Customer Targeting system for campaigns like attrition, upselling, and cross-selling services such as credit cards. Tatras developed a solution that leverages AutoML and GenAI capabilities, where users simply upload customer’s past data on transactions, investments, banking, etc., and select the desired campaign use case. The system then effectively manages the campaign, targeting customers based on the patterns found from the data.

Hypothesis

  • GenAI could enhance data preprocessing through generic data enrichment and feature engineering.
  • AutoML can train optimized ML models using past data for better predictions.
  • GenAI can generate targeted content for customers classified by the ML model.

Execution

  • Ensure the accuracy and integrity of data uploaded by end users.
  • Enhance uploaded data by identifying additional features using GenAI.
  • Train an AutoML model on the enhanced data.
  • Generate targeted content for advertising or campaigns to the classified customers.
  • Libraries used: Scikit-learn, Langchain, PyTorch, Llama Index Transformers.

Outcomes

  • AutoML trains models on enriched data, leading to improved accuracy and efficiency in predictions.
  • The system generates campaign content tailored to classified customer segments, enhancing marketing effectiveness and customer engagement.
  • GenAI-based intelligent feature engineering effectively enriches data.
  • AutoML and GenAI together create a versatile system that addresses various customer targeting use cases.

Project Highlights

  • Massive increase in marketing team efficiency to drive consumer engagement.