Skip to content

Integration

OpenTelemetry logo

OpenTelemetry

Export Hyperpriors traces, metrics, and telemetry over OTLP to any OpenTelemetry-compatible backend your organisation already runs.

Request access to connect

Connections are provisioned in private beta — not authorised from this page.

01 — Permissions

When connected, Hyperpriors is limited to the following — declared up front, revocable at any time. Scope is granted during provisioning, not by a button on this page.

  • 01

    Export agent traces from the Hyperpriors harness over OTLP to your collector or backend

  • 02

    Emit cost, latency, and token-usage metrics as OpenTelemetry metrics

  • 03

    Propagate trace context between your services and Hyperpriors-mediated model calls

  • 04

    Attach guardrail and evaluation outcomes as span attributes and events

02 — Details

Built by
CNCF community
Website
opentelemetry.io (opens in new tab)
Category
Observability
Docs
opentelemetry.io/docs (opens in new tab)

03 — Notes

About OpenTelemetry

OpenTelemetry is the vendor-neutral observability framework hosted by the Cloud Native Computing Foundation. It defines a common standard — APIs, SDKs, semantic conventions, and the OTLP wire protocol — for producing and transporting traces, metrics, and logs. Because nearly every observability backend now ingests OTLP, instrumenting once against OpenTelemetry means never being locked to a single vendor’s pipeline.

What the integration does

Hyperpriors speaks OTLP natively. Every trace the harness produces — agent runs, individual steps, model calls, tool invocations, retries, and fallbacks — can be exported to your OpenTelemetry Collector or directly to any OTLP-compatible backend. Spans carry the attributes that matter for LLM systems: model, token counts, cost, latency, and the guardrail decisions applied along the way.

Trace context propagates in both directions. If your services already emit OpenTelemetry traces, Hyperpriors joins those traces rather than starting parallel ones, so a single distributed trace runs from the inbound request, through your application, through every model call, and back. Cost and latency telemetry is emitted as standard metrics, ready for the dashboards and alerting you already operate.

The practical consequence: your existing observability stack becomes the observability stack for your AI systems, with no second pane of glass to argue about.

Get started

Point Hyperpriors at your OTLP endpoint from the dashboard, or contact us and we will help wire it into your pipeline.