IT Operations Automation · Security and Compliance
AES needed to scale health and safety audits, reduce the time and cost of each audit, and keep a secure human-in-the-loop process while adopting generative AI in a regulated operating environment.
Reported impact
AES doubled audit capacity while cutting time and cost per audit and improving accuracy, proving that secure agentic workflows can accelerate safety operations in the energy sector.
AES built a generative-AI audit environment on Google Cloud using Gemini platform and Anthropic Claude models to automate large portions of safety-review work. Gemini platform provides the secure managed environment for deploying and coordinating the audit agents, while Claude is called through Gemini platform APIs to handle the reasoning-heavy portions of the process. This approach solved security concerns around permissions and service-account management and gave AES a way to synchronize multiple agents across complex audit steps without building a new runtime from scratch. Human reviewers remain in the loop for final examination, but the AI agents now handle much of the repetitive evidence gathering and report generation. The result is a significantly faster, cheaper, and more accurate audit workflow that also lays a foundation for future predictive and maintenance-focused AI use cases across energy operations.
Audit inputs are processed through Gemini platform, where Claude-powered agents execute reasoning and report-generation tasks. Human reviewers then examine the output before audits are finalized, keeping the process controlled and compliant.
Our auditors previously spent 14 days completing each audit process. Now ... the same work is completed in just one hour.
Running Claude on Google Cloud is driving immense improvements, such as a 20% increase in audit accuracy and a 99% reduction in the cost of each audit.
The other half, which involves a ‘human in the loop’ examining the process, remains in place.
We found the solution by enabling Claude to be used through API calls within the Gemini platform environment.
we can double the number of audits conducted
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Google Cloud Sample Modified · Oct 9, 2026 Replenit Legacy marketing tools often treat individual shoppers as spreadsheet averages, relying on broad customer segments and predictable, rigid automation rules. As a result, brands frequently blast customers with irrelevant,… 360% business growth in one quarter with Google Cloud BigQuery · Cloud TPUs · Gemini Google Cloud Sample Modified · Oct 9, 2026 Riafy Technologies Riafy Technologies began its journey with a clear mission: to build adaptable and accessible AI systems that operate natively at a global scale. Today, the company's consumer-end products support 125 million people… Cut incident diagnostics from 40 minutes to under 1 minute Cloud Run · Cloud SQL · Gemini Models Google Cloud Sample Archived · Oct 2, 2026 PT Astra Graphia IT AGIT needed to scale a managed security operations center that could meet Indonesian data-residency requirements, reduce noisy alert handling, and standardize detection engineering for enterprise customers. AGIT scales its enterprise Managed SOC in Indonesia using Google SecOps, Gemini, and Mandiant to drive 15%… Cloud Run · Compute Engine · Gemini 