What can an SRE agent do safely?
An SRE agent can assist incident investigation by gathering operational evidence, proposing hypotheses, and suggesting remediation. Begin with read-only access and an accountable responder. Permission to inspect telemetry is different from permission to change production, restart services, or alter customer data.
What does the source say?
Google Cloud Assist investigations documents AI-supported root-cause investigation across infrastructure and applications. Its existence demonstrates a concrete investigation workflow, not that every generated diagnosis or remediation is correct.
How can you apply this?
Take a resolved incident and prepare a redacted evidence set. Ask an assistant for a timeline, ranked hypotheses, supporting signals, and missing evidence. Compare the answer with the known incident review. Keep remediation in a separate proposal that states expected effect, blast radius, rollback, and the human approval needed to act. Make the final artifact a hypothesis table with evidence, confidence limits, and the next diagnostic query a responder should run.
What should you watch out for?
Do not grant broad production write access simply because a diagnosis sounds confident. Missing telemetry and correlated events can produce plausible but incorrect explanations.