Migrate to Rius
Rius takes OpenTelemetry traces, so migrating to it is usually a routing change rather than a re-instrumentation. How much you change depends on how your code is instrumented today.
- Your instrumentation already emits OpenTelemetry (OpenLLMetry, OpenInference, or the OTel GenAI conventions): keep it and re-point the OTLP exporter, or your Collector, at Rius. Rius understands these conventions, so models, tokens and cost show up without code changes. See OpenLLMetry, OpenInference and Trace LLM calls with OpenTelemetry.
- You use another tool’s own SDK or proxy: swap it for the Rius SDK, or for an OpenTelemetry instrumentor. The guides below map each tool’s API to its Rius equivalent.
Either way, you can send the same traces to both backends while you evaluate, and cut over only when Rius shows what you need.
Rius has no evals, datasets or prompt management today. If you depend on them, keep those workflows in your current tool. Each guide ends with what does not carry over.
Migration guides
- Migrate from Arize Phoenix: keep your OpenInference instrumentation and point it at Rius, or switch to the Rius SDK, then run both side by side and cut over.
- Migrate from Helicone: add the Rius SDK whether you use the Helicone proxy, AI Gateway or async logger, and map sessions, users and custom properties.
- Migrate from Langfuse: map the concepts, keep your instrumentation or swap the SDK, run both backends side by side, then cut over.
- Migrate from LangSmith: move LangChain,
LangGraph and
@traceablecode to Rius, or point LangSmith’s own OpenTelemetry export at Rius. - Migrate from Traceloop: keep OpenLLMetry and change where it sends traces, or switch to the Rius SDK.
If your tool is not listed, the OpenTelemetry path above still applies.
Last updated on