Single-pane monitoring where you bring the AI keys

What it does Pulse pulls health and metrics from Proxmox VE (and its backup and mail siblings), Docker, Podman, and Kubernetes into one React frontend backed by Go. It stores metrics history locally, fires alerts through Discord, Slack, Telegram, or email, and can auto-discover Proxmox nodes on your network. A built-in AI chat and scheduled “Patrol” health checks let you query infrastructure status using your own LLM API key, with all inference staying on your server.

The interesting bit Most monitoring tools either ignore AI or ship your data to it; Pulse keeps the LLM analysis local by making you BYOK. Even the free tier gets the chat assistant and Patrol, which runs scheduled checks for things like ZFS pool capacity, silent backup failures, and VMs stuck in restart loops. The Pro tier adds alert-triggered AI analysis and an optional auto-fix mode with command safety policies.

Key highlights

• Auto-discovers Proxmox nodes and monitors PVE, PBS, PMG, Docker, Podman, and Kubernetes (the latter via agents).
• BYOK AI chat and scheduled Patrol runs entirely on your hardware; no telemetry leaves the box.
• Credentials encrypted at rest with strict API scoping, plus OIDC/SSO support.
• Persistent metrics history and a backup explorer for visualizing job status and storage.
• One-click updates for supported deployments and a live demo to try before you install.

Caveats

• Maintained by a single developer, so bus factor is low.
• Advanced AI features—alert-triggered analysis, Kubernetes AI deep dives, auto-fix, and SIEM webhooks—sit behind the Pulse Pro paywall.
• Kubernetes monitoring requires agent deployment; it is not agentless.

Verdict Homelabbers and small MSPs running Proxmox or Docker fleets who want a pretty, unified UI without the weight of Prometheus/Grafana should look here. Enterprises needing multi-tenant RBAC or long-term maintainability guarantees should probably keep walking.