Hanzo Telemetry
OpenTelemetry collector and pipeline for AI workloads
Collect, transform, and route traces, metrics, and logs from any source to any backend. Built on OpenTelemetry, tuned for high-cardinality AI inference traffic.
One Pipeline for Every Signal
Vendor-neutral collection, transformation, and routing.
OTLP Native
First-class support for OTLP/gRPC and OTLP/HTTP. Drop-in replacement for any OpenTelemetry collector.
Traces, Metrics, Logs
One pipeline for all three signals. Correlate spans, gauges, and log lines without juggling agents.
Smart Sampling
Tail-based sampling on errors, latency, and cost. Keep what matters, drop the noise.
AI-Aware Processors
Built-in processors for token counts, prompt redaction, model routing, and inference latency.
Any Backend
Export to Hanzo Metrics, Hanzo O11y, Datadog, Honeycomb, Grafana — or all at once.
Privacy First
Redact PII at the edge. Self-hosted by default. Your prompts never leave your network.