Skip to main content
Available via: Kong’s OpenTelemetry plugin (any language calling through the gateway). This is not a Zespan SDK wrapper — see Send traces from any OTel-instrumented app for how the underlying endpoint works.
If you already route LLM traffic through Kong AI Gateway, enable Kong’s OpenTelemetry plugin on the relevant route or service and point it at Zespan — every call proxied through Kong is traced with GenAI-specific span attributes (model, tokens, cost), independent of which app made the call.

​
Setup

Enable the opentelemetry plugin on your AI Gateway route (declarative config shown; the same fields apply via the Admin API or kubectl/Konnect):
Kong sends OTLP over HTTP with Protobuf by default — this matches what Zespan’s /v1/traces endpoint accepts, no protocol change needed.

​
What gets captured

Kong AI Gateway’s LLM-specific spans already carry GenAI semantic-convention attributes — model, token usage, cost, latency — the same fields Zespan reads from any OTel source. Every request proxied through the gateway is traced regardless of the calling app’s language or framework, which is the main reason to route through a gateway in the first place.