What is JetBrains Tracy and how does it trace LLM calls?

Tracy is JetBrains' open-source AI observability library for Kotlin/Java. A single annotation (e.g., @TraceAI) automatically traces the full LLM call chain—inputs, outputs, latency, token consumption, and cost—integrating with APM tools like Grafana and Datadog.

What unique advantages does Tracy offer Kotlin/Java developers?

Tracy is designed for the JVM ecosystem with seamless Spring Boot and Micronaut integration. Its annotation-driven approach matches Java development conventions, making it far more suitable for JVM environments than Python-first tools like LangSmith.

Why is AI observability critical for production AI applications?

LLM calls have inherent uncertainty—same inputs may yield different outputs, costs are unpredictable, and latency fluctuates. Tracy's full-chain tracing helps teams identify anomalous outputs, optimize prompts, control costs, and satisfy enterprise security and compliance auditing.

Python Has LangSmith for LLM Tracing. What About JVM? JetBrains Tracy Does It With One Annotation

JetBrains released Tracy, an open-source AI observability library for Kotlin/Java. Developers can trace LLM calls, tool executions, and custom logic with a single @Trace annotation. Built on OpenTelemetry, compatible with Langfuse and W&B Weave, supporting OpenAI, Anthropic, and Gemini SDKs.

JetBrains Tracy: Filling the JVM Ecosystem's AI Observability Gap

The Problem

Python ecosystem has relatively mature AI observability tools (LangSmith, W&B Weave, Langfuse). The JVM ecosystem (Java/Kotlin) was almost completely empty—no purpose-built AI tracing library, forcing Kotlin developers to cobble together custom solutions.

Tracy (Released March 11, 2026, Apache 2.0)

An open-source AI observability library for Kotlin/Java. Kotlin 2.0.0+, Java 17+ compatible.

The Simplest Usage—One Annotation

@Trace
suspend fun analyzeDocument(doc: String): AnalysisResult {
// All LLM calls inside are auto-traced
}

What Tracy Automatically Captures

Complete call chains
All LLM calls: model, prompt, token usage, latency, cost estimate
Tool call parameters and results
Custom business logic execution time and state

Technical Architecture

**OpenTelemetry-compliant**: Follows OpenTelemetry Generative AI Semantic Conventions
**Compatible backends**: Langfuse, W&B Weave, Jaeger, Zipkin, any OTLP-compatible system
**Supported LLM SDKs**: OpenAI, Anthropic, Google Gemini; compatible with OkHttp and Ktor

Four Core Values

1. Debug failures: Locate where AI agents fail and what prompt/context was involved

2. Cost monitoring: Track token consumption and estimated costs per call

3. Latency analysis: Identify bottlenecks (slowest LLM? timeout tool call?)

4. Unified LLM usage view: All AI activity in one dashboard

Strategic Importance

Large enterprise AI applications run on JVM (banking, insurance, manufacturing, telecom). Tracy enables these organizations to confidently deploy AI in production with Python-level observability support.

JetBrains AI Stack

Tracy + Koog (AI agent framework) + IntelliJ IDEA/Junie (AI-assisted coding) = JetBrains' comprehensive Kotlin AI development stack.

In-Depth Analysis and Industry Outlook

From a broader perspective, this development reflects the accelerating trend of AI technology transitioning from laboratories to industrial applications. Industry analysts widely agree that 2026 will be a pivotal year for AI commercialization. On the technical front, large model inference efficiency continues to improve while deployment costs decline, enabling more SMEs to access advanced AI capabilities. On the market front, enterprise expectations for AI investment returns are shifting from long-term strategic value to short-term quantifiable gains.

However, the rapid proliferation of AI also brings new challenges: increasing complexity of data privacy protection, growing demands for AI decision transparency, and difficulties in cross-border AI governance coordination. Regulatory authorities across multiple countries are closely monitoring these developments, attempting to balance innovation promotion with risk prevention. For investors, identifying AI companies with truly sustainable competitive advantages has become increasingly critical as the market transitions from hype to value validation.