Objectives
The primary objective was to optimize the utility payment process by creating a solution that could work with various meter types, fit within budgetary constraints, and provide customers with reliable proof of accurate readings. This required using AI to solve issues of accuracy, consistency, and customer trust in the meter reading and billing process.
Challenges
Three main challenges hindered the adoption of digital technology for utility payments in Ukrainian residential complexes:
"Zoo" of Technologies: There are millions of meters of different types, technologies, and generations in buildings. A lack of standardization has led to a fragmented landscape that makes connecting all types of meters into a unified system extremely challenging.
Limited Budgets: In non-premium housing segments, budget constraints led developers to use simple, affordable meters without modems or smart technology, making the adoption of smart technology more difficult.
Lack of Trust in Digital Technology: When meter data is transmitted directly to a server, it often sparks distrust among residents, especially when meters are located in public areas. There is no clear mechanism for verifying the readings, resulting in disputes and a lack of trust.
Process
To address these challenges, KPI Creatives developed the Zezman Meters app, an Android-based solution for meter reading automation. Using tens of thousands of meter photos, a neural network was trained to recognize meter numbers and readings from photographs. The app was designed to provide a simple, photo-based method for meter reading, allowing electricians and plumbers to quickly and efficiently record data.
In cases where the neural network failed to recognize the reading correctly, users were prompted to confirm or correct the data, ensuring accuracy. The app also generates route sheets to prevent any meter from being missed, providing complete coverage.
Solutions
The Zezman Meters app utilized AI to instantly recognize meter numbers and readings from photos, minimizing the need for manual intervention. This solution drastically improved efficiency in meter reading, cutting down the average reading time to just 15 seconds per meter. The app also automated the export of readings to accounting software, ensuring fast and accurate billing.
Results
Improved Efficiency: One individual can now collect readings from 3,000 meters in two working days—approximately 10 times the daily norm of traditional utility controllers.
Labor Savings: The total labor cost for meter reading was reduced by 6 times, from 200 to 30 working hours.
Customer Satisfaction: Customer service managers now have access to photographic proof of meter readings, allowing them to quickly resolve billing disputes and enhance customer trust.
Easy Payments: Bills are delivered via a Telegram bot, and customers can make payments in one click through the bot, making the payment process seamless.
