Essential Energy is preparing to deploy a new AI application that will help it to spot safety issues across its distribution network sooner and more accurately. The utility company, which operates a network serving 900,000 regional and remote premises in NSW and southern Queensland, said that the new AI system uses algorithms to analyse notes its field staff record on iPads to spot patterns that can be used to generate safety insights. Principal data scientist Andrew Slack-Smith said that while early iterations of the system were expected to involve analysis of structured data, the utility provider realised it was missing important human intelligence returning from the field. “We were going through all the structured data when we realised that all the juicy data was under the comments section in the system, which isn’t that unusual. It’s where people were writing the notes that really make a difference because there were real insights,” Slack-Smith said. The electricity company said that the AI-based system’s underlying algorithm was expected to spot safety-related data in 76 percent of cases, compared to only 59 percent of cases for its current rules-based automated checks. Slack-Smith said that the system had the potential scale to “hundreds of other data sources”. “All we need to do is find the data source we want to analyse and run an algorithm over it,” Slack-Smith said. Essential Energy said that it would eventually evolve the system to free field staff from entering text into their devices by hand, instead using voice-to-text input to provide a speed improvement. “Crews in the field will be able to simply talk about what they are seeing and AI will not only help build out the full picture of the issue by using its knowledge of our assets, but also give them three succinct paragraphs that will fill in any gaps for them,” Slack-Smith said. Essential Energy isn’t alone in wanting to test AI’s potential use in monitoring infrastructure assets. Last month iTnews revealed that Yarra Valley Water is planning to use generative AI to predict failures across its water supply infrastructure drive cost out of its maintenance operations. In that case the water authority is currently working on a proof-of-concept for a new system that would rely on large language model inference engines to analyse data coming from sensors embedded in its infrastructure. Yarra Valley Water’s cloud and devops lead Murali Manohar Shunmugaraja told iTnews that the system could be in operation as soon as next year.