Using Artificial Intelligence to Meet Regulatory Reporting Demands

ater systems generate and collect large volumes of data through multiple systems, including detailed records of infrastructure, customer interactions, supervisory control and data acquisition systems (SCADA), billing systems, and maintenance logs.Although some of these data sources are digitized and readily accessible, many are siloed-stored in legacy systems or saved in handwritten logs.For example, a utility may want to optimize its pipe replacement schedule by analyzing historical records on breaks, leaks, and remaining useful-life estimations.However, critical data like installation dates and maintenance logs might exist only in decades-old paper files or in different electronic systems.Organizing such data into a usable format is time-consuming, but doing so can help utilities make decisions in their operations, long-term planning, and regulatory compliance.A representative example is the US Environmental Protection Agency's Lead and Copper Rule Improvements (LCRI), which require water systems to perform the following tasks: Develop a comprehensive inventory of the service lines connected to their distribution systems Identify those that have the potential to release lead Replace all service lines that contain lead These tasks require a thorough review of historical records related to service line installations and replacements, which can pose challenges for utilities that do not have the resources to manually digitize and review paper records containing this information.Without an asset inventory, it is difficult to effectively answer regulatory compliance and capital planning questions such as the number of customers at risk from lead service lines (LSLs) and how much investment is needed for an LSL replacement program to eliminate any public health risks and meet all regulatory obligations.In this context, advanced technologies like optical character recognition (OCR) and large language models (LLMs) can extract and organize data from decades of inspections recorded on paper and digital forms.