Versus Incident is a self-hosted AI SRE agent that watches your services, detects anomalies and helps your team investigate, with 897 stars on GitHub.
Versus Incident describes itself as an AI SRE teammate: you do not have to monitor your systems alone. Written in Go and deployed on Kubernetes with Helm, the self-hosted agent keeps track of your services, applications and systems, learns what normal behaviour looks like, and detects new or unexpected problems โ new errors, unusual log patterns and activity spikes. Rather than paging on every repeated event, it groups repeated detections into a single incident episode and gives your team the evidence needed to investigate. A DevOps chat interface lets you ask the agent what changed or why an incident fired, and findings can be sent to Slack, PagerDuty or Telegram for your on-call workflow. Crucially, the agent investigates read-only and never changes infrastructure on its own; engineers review the evidence and decide the response, which keeps the automation accountable and safe to run against production systems.
Versus Incident is open-core. The MIT-licensed Community edition โ log-based detection, AI analysis and DevOps chat โ is free to self-host. A paid Enterprise edition adds standing metrics and trace monitoring, SLI/SLO guidance, advanced alert-fatigue controls and enterprise access controls.
Versus Incident is for DevOps and SRE teams drowning in noisy alerts who want AI help triaging what is actually new. Teams that prefer self-hosted tooling and read-only, human-in-the-loop automation will appreciate its design boundaries.