Relvy (YC F24) – Automating On-call Runbooks with AI
Relvy is an AI-powered platform that automates on-call runbooks for software engineering teams, helping them debug and resolve production issues faster.
Why it matters
Relvy's AI-powered approach to automating on-call runbooks can significantly reduce the burden on software engineering teams, helping them resolve production issues more efficiently.
Key Points
- 1Relvy uses specialized tools to analyze telemetry data and code at scale, detecting anomalies and identifying problem areas
- 2The AI agent is anchored around runbooks, leading to less exploration and more deterministic steps that reflect experienced engineers' actions
- 3Relvy can be configured to automatically respond to alerts and run mitigation actions with human approval
Details
Relvy is tackling the challenge of autonomous root cause analysis, which has proven difficult for AI systems. The company has identified three main reasons for this: the volume of telemetry data can overwhelm the model, data interpretation is enterprise-context dependent, and on-call is a high-stakes, time-constrained problem where AI errors are not easily forgiven. To address these issues, Relvy has built specialized tools for telemetry data analysis, anomaly detection, log pattern search, and reasoning about span trees. By anchoring the AI agent around runbooks, Relvy aims to provide faster analysis and less cognitive load on engineers. The platform can be installed locally or accessed through the cloud, and it integrates with observability and code repositories to automate various investigation and mitigation steps.
No comments yet
Be the first to comment