Agentic SRE: How DriveWealth wired Grafana Assistant, Investigations, and IRM for triaging alerts before anyone gets paged

At DriveWealth, a global B2B financial technology platform providing brokerage-as-a-service, every alert used to mean waking an engineer — not because every alert represented a critical issue, but because there was no triage layer between the alert and the on-call rotation. Serge Sheynkman, VP, Head of SRE and IT Operations, set out to change that by creating a control plane in Grafana Cloud, where Grafana Assistant and Grafana Cloud IRM triage alerts before a human ever sees them. The result? If and when an on-call engineer needs to be paged, the context is already assembled.

Join us for a practitioner's account of building agentic SRE into a regulated, real-time brokerage environment (front, middle, and back office), where generic monitoring has fallen short and response time is measured in dollars. In this session, Serge walks through what the team built, what didn't work, and what it means to design an on-call workflow around an agent doing the first-pass investigation. He also shares some early results: DriveWealth's client-support team now self-serves 25% of its questions through Grafana Assistant, on the way to a 60–70% target, while the on-call engineers are sleeping better.