Success case

Automation of bank transaction

categorization and financial

indicator calculations

The challenge

Unnax is a fintech that helps businesses to easily and securely access their customers’ financial data, as well as enrich it to facilitate decision-making.


The main goal is to make the process of enriching individuals’ financial data more efficient by automating the categorization of bank transactions and the calculation of financial indicators through behavioral patterns.

How have we done it?

  • Statistical Learning
  • Language Models
  • Ensemble Models
  • Rule-based models
  • Machine Learning Operations automated pipelines
  • Active Learning, DB design

Results provided

  • Cost reduction in infrastructure (databases and APIs).
  • Reduction in efforts dedicated to repetitive tasks (through automations).
  • Reduction in categorization response time by approximately 20%.
  • Improvement in categorization quality from approximately 67% to over 85% accuracy → Increased customer satisfaction and acquisition of new customers.