Installing the Helm chart now deploys a per-node sensor that observes every service on the cluster — eBPF traces, RED metrics and log collection with no SDK, no sidecar, no app changes.

The Services screen is live: every service seen in the selected window — auto-discovered zero-code on Kubernetes or sending OTLP directly — with its rate, error percentage and latency percentiles in one sortable table.

Our north star — a fresh Kubernetes cluster showing a live service map in under five minutes, zero app changes — is now a hard CI gate, and the gateway ships as a purpose-built minimal collector.

The trace explorer moved to a split layout: the trace list stays on the left while the detail opens on the right, so hopping between traces is one click. Clicking a span now opens a side panel (no more inline expand), and a maximize button pops the whole thing to a full-window view.

Open any trace in a full-window explorer and switch between six different views of the same trace — no context lost.

The trace search grew a proper filter panel — so you can slice by what actually matters.

The service map went from disconnected dots to an actual topology.

A new System Status tab (under Settings) gives operators an at-a-glance view of the backend's health.

Logs are searchable and wired to traces.