How AES Uses Claude on Vertex AI to Automate Safety Audits

AES is a global energy company undergoing a major strategic shift from traditional power generation to renewable energy sources such as wind and solar. This transition fundamentally changed the operational complexity of managing energy assets: whereas a single traditional turbine generates hundreds of megawatts, a wind turbine generates only about 1.5 megawatts, meaning AES must now manage dozens of assets to match equivalent output. This explosion in asset count placed enormous pressure on operational and safety teams.

The core challenge was AES's safety audit program. The company conducts approximately 1,550 safety audits annually, and each traditionally required up to two weeks to complete—an unsustainable pace given the growing asset base. The manual, resource-intensive nature of audits threatened to become a critical bottleneck, limiting AES's ability to scale its renewable energy operations safely.

To address this, AES's data science team implemented Claude, integrated through Google Cloud's Vertex AI platform, using a multi-agent framework to automate audit result generation. The choice of Vertex AI was deliberate: it streamlined the internal security approval process for handling sensitive operational data and provided the infrastructure needed for high API call volumes across synchronized agents.

The results were transformative. Audit result generation became 96% faster, audit accuracy improved by 10–20%, and audit costs fell by 99%. The system enabled AES to keep pace with its expanding renewable portfolio without proportional increases in audit staffing or time.