Sustainable Multilingual AI for User Support in Public Transport: How ATTG Reduced Load, Costs and Environmental Impact with Compressed Models
Public transport authorities face rising volumes of repetitive queries and increasing costs for multilingual customer support. ATTG demonstrates how lightweight, compressed AI models can power a multilingual conversational agent in Spanish, English and Basque, resolving most incidents automatically while escalating complex cases to human staff. By replacing large models with efficient compressed ones, ATTG achieves real-time performance, lowers energy consumption and CO2 emissions and supports on-prem or sovereign-cloud deployment. This case shows how AI can improve customer satisfaction, streamline operations and advance sustainability goals without compromising inclusivity.