Why AI matters at this scale

Public Service Company of Oklahoma (PSO), a subsidiary of American Electric Power, is a regulated electric utility serving over 562,000 customers across Oklahoma. Founded in 1913 and headquartered in Tulsa, PSO operates and maintains a vast distribution grid, delivering electricity generated from a diverse mix of sources, including wind and natural gas. As a mid-sized utility with over 1,000 employees, PSO balances the demands of reliable service, regulatory compliance, infrastructure modernization, and integrating renewable energy.

For a company of PSO's size and in the utility sector, AI is not a futuristic concept but a practical tool for addressing existential pressures. The grid is aging, severe weather events are increasing, and customer expectations for reliability and digital engagement are rising. At the 1001-5000 employee scale, PSO has the operational complexity and data volume that makes manual processes inefficient, yet it likely lacks the vast R&D budgets of mega-utilities. This makes targeted, high-ROI AI applications crucial for maintaining competitiveness and regulatory standing. AI enables moving from reactive operations to predictive and proactive management, which is essential for a capital-intensive business where unplanned outages are enormously costly.

Concrete AI Opportunities with ROI Framing

1. Predictive Asset Maintenance: By applying machine learning to historical sensor data, weather patterns, and maintenance records, PSO can predict failures in transformers, breakers, and other critical assets. The ROI is clear: preventing a single major substation failure can avoid millions in equipment damage, emergency labor, and regulatory penalties, while enhancing reliability metrics that influence rate cases.

2. Dynamic Load and Renewable Forecasting: AI models excel at analyzing complex, multivariate data (e.g., weather, economic activity, calendar events) to forecast electricity demand and renewable generation. More accurate forecasts allow PSO to optimize power purchases, reduce reliance on expensive peaking plants, and seamlessly integrate more wind and solar—directly lowering fuel costs and supporting state clean energy goals.

3. Automated Vegetation and Inspection Management: Using drones or satellite imagery combined with computer vision, PSO can automatically identify vegetation encroachment and structural damage on thousands of miles of lines. This transforms a labor-intensive, cyclical manual process into a targeted, data-driven program. The ROI comes from reducing the frequency of costly trimming cycles, preventing vegetation-caused wildfires and outages, and improving field crew safety and productivity.

Deployment Risks Specific to This Size Band

PSO's size presents unique deployment challenges. The organization is large enough to have legacy IT (e.g., old SCADA systems, customer information systems) and operational technology that may not be designed for real-time data sharing, creating significant integration hurdles. There is likely a mix of modern and outdated data infrastructure, requiring careful data governance and middleware investments. Furthermore, while PSO has dedicated IT staff, it may lack deep in-house AI/ML expertise, creating a reliance on vendors or consultants. Cybersecurity risks are paramount—any AI system connected to grid operations becomes a high-value target, necessitating robust security by design. Finally, as a regulated entity, PSO must justify AI investments to regulators, requiring clear business cases and potential pilot programs to demonstrate value before large-scale deployment.