diff --git a/img/integrations/bronto-llm-usage.png b/img/integrations/bronto-llm-usage.png new file mode 100644 index 0000000..ae6c223 Binary files /dev/null and b/img/integrations/bronto-llm-usage.png differ diff --git a/img/integrations/bronto.png b/img/integrations/bronto.png new file mode 100644 index 0000000..6f1fdd8 Binary files /dev/null and b/img/integrations/bronto.png differ diff --git a/mint.json b/mint.json index 4d5a7c7..903b2fc 100644 --- a/mint.json +++ b/mint.json @@ -118,6 +118,7 @@ "openllmetry/integrations/azure", "openllmetry/integrations/braintrust", "openllmetry/integrations/bmc", + "openllmetry/integrations/bronto", "openllmetry/integrations/dash0", "openllmetry/integrations/datadog", "openllmetry/integrations/dynatrace", diff --git a/openllmetry/integrations/bronto.mdx b/openllmetry/integrations/bronto.mdx new file mode 100644 index 0000000..ccc4082 --- /dev/null +++ b/openllmetry/integrations/bronto.mdx @@ -0,0 +1,27 @@ +--- +title: "LLM Observability with Bronto and OpenLLMetry" +sidebarTitle: "Bronto" +--- + + + Bronto trace view of an LLM agent trace, with a span's GenAI input and output messages shown in the detail panel + + +[Bronto](https://www.bronto.io) is an intelligent data platform for observability, built on BrontoDB — a polymorphic database designed to handle high-cardinality logs, traces, and metrics efficiently. Because you don't have to sample workloads or shorten retention to control cost, you can send full-fidelity LLM traces and keep months of them immediately queryable with sub-second search — no rehydration step before a root cause analysis. + +OpenLLMetry spans follow the [OpenTelemetry GenAI semantic conventions](https://github.com/open-telemetry/semantic-conventions-genai/tree/main/docs/gen-ai), so attributes such as `gen_ai.provider.name`, `gen_ai.request.model`, and `gen_ai.usage.input_tokens` / `gen_ai.usage.output_tokens` are searchable and aggregatable on arrival — group token usage by model, chart cost trends, or inspect prompt and completion content. Attribute names vary by instrumentation and release — some report the provider as `gen_ai.system` rather than `gen_ai.provider.name`, and some use `gen_ai.usage.prompt_tokens` / `gen_ai.usage.completion_tokens` for token counts — so check the fields your version emits. Prompt and completion content is captured by default — as span attributes or span events, depending on the instrumentation — and can be turned off with `TRACELOOP_TRACE_CONTENT=false` (see [Privacy](/openllmetry/privacy/traces)). Bronto's MCP server exposes the same search and analysis as tools, so agents can investigate your LLM traces directly. + + + Bronto LLM Usage dashboard charting input and output tokens by model, plus time to first token and token usage per trace + + +Route your traces to Bronto's OTLP endpoint: + +```bash +export TRACELOOP_BASE_URL="https://ingestion.eu.bronto.io" +export TRACELOOP_HEADERS="x-bronto-api-key=" +``` + +Use `https://ingestion.us.bronto.io` for the US region. Do not append `/v1/traces` — OpenLLMetry adds the signal path automatically. + +For the full setup, including the OTel Collector path and the complete attribute reference, see the [Bronto OpenLLMetry documentation](https://docs.bronto.io/integrations/openllmetry). diff --git a/openllmetry/integrations/introduction.mdx b/openllmetry/integrations/introduction.mdx index 33f7742..2fd7dc2 100644 --- a/openllmetry/integrations/introduction.mdx +++ b/openllmetry/integrations/introduction.mdx @@ -17,6 +17,7 @@ in any observability platform that supports OpenTelemetry. > +