The agent trace, not just the LLM call.
Most observability tooling stops at the model boundary. Trefur extends one layer up — to the decisions agents make, the tools they call, the retries they fire, and the outcomes they produce. That is where the interesting questions live.
What changes when you turn it on.
These are the outcomes engineering teams shipping production agents see in the first week — no re-platforming, no rip-and- replace.
Stop debugging in the dark
Every decision, every tool call, every retry — captured the moment it happens. Drill from a failing customer interaction to the exact step that went wrong.
See cost the day it drifts
Tokens and dollars attributed to the step that emitted them. Catch a 40% bill spike before it lands on the CFO's desk.
Find regressions before customers do
Alert envelopes on retry rate, error rate, p95 latency, cost-per-run. Routed to your existing oncall tooling with the trace already attached.
Swap model vendors without rebuilding observability
One canonical trace shape across OpenAI, Anthropic, Bedrock, Azure OpenAI, Gemini. Vendor decisions become data, not lock-in.
Adopt without re-platforming
OpenTelemetry-native. Add Trefur as a second exporter alongside Datadog, Honeycomb, Grafana. Your existing pipeline stays in place.
Keep your data on your network
Run the collector inside your VPC. Batch, buffer, and redact before anything leaves. Self-hostable from day one.
A connected trace, end to end.
A single root span per agent run. Every LLM call, tool invocation, sub-agent spawn, retry, and outcome is a child span. Schema-validated, OpenTelemetry-canonical, replayable.
Drill from a failing customer interaction down to the specific decision that went wrong, in the specific deployment, on the specific build.
Open standards. Every framework. Your pipeline.
Trefur reads OpenTelemetry GenAI Semantic Conventions natively. You can drop in our SDK for one-line auto-instrumentation, run the trefur collector for multi-process capture, or simply add Trefur as a second OTel exporter alongside whatever you already run.
What engineering teams actually do with Trefur.
Debug failing agents in production
Search by trace_id, agent name, customer interaction, or error class. Drill into the specific decision that went wrong and the prompt context the model saw.
Catch regressions before customers do
Alert on p95 latency, retry-rate spikes, cost-per-run drift, and error-rate envelopes. Route alerts to Slack, PagerDuty, Teams, or any webhook.
Investigate cost spikes
Per-trace, per-agent, per-deployment cost roll-ups. See exactly which step pushed the token count and whether it was the LLM or the tool output.
Track cost and latency across deployments
Roll up cost-per-run, p95 latency, retry rate, and error rate by build SHA and environment. See how a new deploy moved the numbers against its baseline.
Go deeper.
Each capability gets its own page with concrete benefits, the shape of the data, and what teams actually do with it.
Agent traces
Root span per run, child span per decision, tool call, retry, outcome.
Cost tracking
Every token and dollar attributed to the step that emitted them.
Alerts
Retry-rate, cost drift, latency, error envelopes — routed with the trace attached.
Cross-vendor
One trace shape across every model provider you call.
Self-hostable collector
Single Go binary inside your network. Batching, redaction, fan-out.
Replay debug
Reconstruct any failing run from production, exactly as it happened.
AI engineers
Ship production agents you can debug. Same workflow as services.
Platform teams
One layer across every team, every framework, every vendor.
Business & ops
See AI in real workflows in plain English. Not a tracing tutorial.
Agent frameworks
LangChain, CrewAI, AutoGen, Pydantic AI, MCP — drop-in.
Model providers
OpenAI, Anthropic, Gemini, Bedrock, Azure — one trace shape.
SaaS connectors
Google Workspace, Microsoft 365, Slack, Salesforce — and more.
FAQ
Does Trefur work with my existing OpenTelemetry pipeline?+
Yes. Trefur reads OTLP/HTTP and OTLP/gRPC on the standard endpoints. Add a second OTel exporter and fan out — your current backend stays in place.
Do I have to use a specific agent framework?+
No. Drop-in SDKs cover LangChain, CrewAI, AutoGen, Pydantic AI, and MCP. For custom frameworks, emit OpenTelemetry GenAI semantic conventions and Trefur reads them.
Can I self-host the collector?+
Yes. The collector is a single Go binary. Run it inside your network for local batching, redaction, and on-disk buffering before egress.
How does pricing work?+
Free tier is permanent for individual developers — full feature access, no credit card. Pro / Team / Enterprise tiers scale with trace volume, retention, and support. See pricing for details.
Ship agents you can debug.
Free tier. No card required. First trace in under five minutes.