Meet তড়িৎবন্ধু: Bangladesh’s First Bangla Conversational Agent for Electricity Services

Introducing our groundbreaking conversational agent, তড়িৎবন্ধু, meticulously designed from scratch to revolutionize customer interaction for electricity-related service providers in Bangladesh. As the first of its kind, this innovative agent is capable of engaging in fluent Bangla conversations, addressing the unique needs of Bangla-speaking users with remarkable accuracy and cultural relevance. Operating seamlessly in both text and voice modes, তড়িৎবন্ধু offers versatile communication options to suit diverse user preferences, ensuring accessibility and convenience for all.

This project addresses a critical challenge in conversational agent development: user understanding. We conducted a comprehensive human-computer interaction (HCI) oriented user study to inform the agent’s design, gathering valuable insights to enhance usability and effectiveness. A standout feature of তড়িৎবন্ধু is its ability to understand and respond to user emotions, ensuring a personalized and empathetic communication experience that significantly enhances customer satisfaction. Our development process included the creation of an extensive Bangla multi-turn conversational dataset, which enables the agent to handle complex, multi-step interactions with ease and efficiency.

By leveraging advanced natural language processing and machine learning technologies, তড়িৎবন্ধু sets a new standard for conversational agents in Bangladesh. This pioneering solution not only showcases our commitment to technological innovation but also highlights our dedication to improving service delivery and customer engagement in the Bangla language. Through this initiative, we aim to provide an unparalleled user experience, ultimately transforming how electricity-related services are delivered and managed in Bangladesh.

Electricity and Consumption: Analyzing current prepaid and past postpaid consumption data of Dhaka south’s electricity users

  • Measuring the change in payment and consumption behavior due to the replacing of conventional postpaid meters to prepaid meters.
  • Analyzing the consumption data of households before and after prepaid meter intervention and comparing them with households who are still using old meters.
  • We found prepaid meters effectively reduced payment delinquency. Further analysis on consumption is still under progress.

Publication:

S. B. Rabbani, A. A. Ali and M. Zaber, “Does Electric Prepaid Meters Decrease Payment Delinquency? Evidence from Data Centric Analysis of Electricity Consumption in Dhaka, Bangladesh” in IEEE Region 10 Symposium (TENSYMP), Dhaka, 2020

Abstract— Prepaid electricity meters are favored by power distribution companies to improve revenue collection, debt prevention and ensure customer satisfaction. In this paper, we investigate changes in bill payments pattern of customers in Dhaka, Bangladesh who had switched from post to pre-payment meters. We collected postpaid bill and prepaid recharge payment data over there years. We specifically look into payment behavior of residential household customers to understand the effect of prepaid meters on payment delinquency. Analysis of postpaid bill indicate that at least 32% households skipped paying one or more bills over a period of one year. Households consuming 100-300 KWH accounts for almost 51% of the total delinquent households. After prepaid switch, approximately 46% of all households are making prepayments every month. 52% of households who skip paying bill and 45% of households who pay bill after due date, at least once, are now recharging every month. Overall the technological intervention shows considerable reduction of retail debts and a behavioral shift towards on time payment.