AI for identifying coding inaccuracies of pipeline inspections

Challenge

Seattle Public Utilities CCTV Operators have inconsistencies with their PACP (Pipeline Assessment Certification Program) coding practices. PACP coding inaccuracies can significantly affect the decision-making process for pipeline maintenance and rehabilitation. These inaccuracies have the potential to lead to costly repairs, missed critical interventions, or unnecessary maintenance.

Solution

Utilizing Azure AI, Kopius developed an AI system to compare PACP codes generated by operators during CCTV inspection of pipes with those produced by SPU’s benchmark system.

To visualize the difference between operator performance and benchmark standards, Kopius produced a Power BI dashboard that highlights the discrepancies to help operators and leadership alike understand where training intervention can make the biggest impact.

To improve training, Kopius developed an Azure AI Agent on top of the digital benchmark system manual that allows users to ask questions and rapidly receive relevant responses to enhance PACP coding education.

Results

After a successful Phase 1 of the project, Kopius is engaging in a Phase 2 that transforms the initial AI comparison model into a fully integrated QA/QC ecosystem with additional interactive dashboards and a human-in-the-loop feedback loop between staff and the AI model.