Neura Market
traceAI logo

traceAI

Free

Open-source OpenTelemetry-native tracing framework for LLM applications, with native LangChain instrumentation and support for 20+ frameworks. [![GitHub Repo stars](https://img.shields.io/github/stars/future-agi/traceAI?style=social)](https://github.com/future-agi/traceAI)

FreeFree tier
Type
Open Source
Company
Future AGI
Visit Website GitHub (40 stars)

About traceAI

traceAI is an open-source, OpenTelemetry-native observability framework for AI applications. It captures every LLM call, prompt, token count, retrieval step, and agent decision as structured traces. With drop-in instrumentation for over 50 AI frameworks across Python, TypeScript, Java, and C#, traceAI sends traces to any OpenTelemetry-compatible backend (Datadog, Grafana, Jaeger, Future AGI, etc.)—no new vendor or dashboard required. It provides standardized tracing, rich context (prompts, completions, tokens, model parameters, tool calls), and production-grade features including async support, streaming, error handling, and low overhead.

Key Features

OpenTelemetry-native tracing with semantic conventions for LLM calls, agents, tools, retrieval, and vector databases
Drop-in instrumentation (3 lines of code) - no refactoring needed
Multi-framework support: 50+ integrations across Python, TypeScript, Java, and C#
Vendor agnostic - works with any OpenTelemetry-compatible backend (Datadog, Grafana, Jaeger, Future AGI, etc.)
Rich context capture: prompts, completions, tokens, model parameters, tool calls, and more
Production-grade features: async support, streaming, error handling, and low-overhead tracing

Pros & Cons

Pros
  • Fully open source and free to use
  • Vendor agnostic - no vendor lock-in; works with existing OTel backends
  • Supports 50+ AI frameworks across 4 languages (Python, TypeScript, Java, C#)
  • Easy drop-in setup with minimal code changes
  • Captures rich context beyond basic traces (prompts, tokens, model params, tool calls)
  • Production-grade with async, streaming, and error handling support
Cons
  • Requires setup and maintenance of an OpenTelemetry-compatible backend for visualization and storage
  • Initial configuration of API keys and environment variables may be non-trivial for newcomers
  • Tracing overhead may impact performance in extremely high-throughput scenarios
  • Documentation and community support are still evolving (open-source project)

Best For

Debugging LLM application errors by tracing each call, retrieval step, and agent decisionMonitoring and optimizing performance of AI pipelines with detailed token usage and latency dataEnsuring observability in production AI systems with structured traces and alertsTracing complex agent workflows involving multiple tool invocations and LLM interactionsAuditing and compliance for AI applications by capturing full prompt and response history

FAQ

What is traceAI?
traceAI is an open-source, OpenTelemetry-native observability framework designed for AI applications. It captures every LLM call, prompt, token count, retrieval step, and agent decision as structured traces, sending them to any OTel-compatible backend.
What backends does traceAI support?
traceAI works with any OpenTelemetry-compatible backend, including Datadog, Grafana, Jaeger, Future AGI, and others. No new vendor or dashboard is required.
Which programming languages and frameworks does traceAI support?
traceAI supports Python, TypeScript, Java, and C# with consistent APIs. It offers drop-in instrumentation for 50+ AI frameworks including OpenAI, LangChain, Anthropic, and more.
Is traceAI free to use?
Yes, traceAI is fully open source and free to use. The source code is available on GitHub under the repository future-agi/traceAI.
How do I instrument my application with traceAI?
Instrumentation typically requires adding 3 lines of code: installing the appropriate package (e.g., traceai-openai), setting environment variables for API keys, registering a tracer provider, and calling the instrumentor for your framework.