Open Source AI Observability

AI Agent Observability
Done Right

Trace, debug, and monitor your AI agents in production. Get complete visibility into every step of your agent's reasoning chain with real-time insights.

Everything you need to debug AI

Built-in observability for every framework. No vendor lock-in.

Real-time Tracing

Watch your agents think in real-time with complete execution visibility

Cost Analytics

Track token usage and API costs across all your agents

Performance Monitoring

Identify bottlenecks and optimize agent execution time

Error Detection

Automatically categorize and alert on agent failures

Multi-Framework

Works with LangChain, CrewAI, AutoGPT, or custom agents

Self-Hosted

Deploy on your infrastructure. Your data stays yours

Developer First

Simple SDK with Python and TypeScript support

Production Ready

Built for scale with enterprise-grade reliability

Get started in minutes

Add tracing to your AI agents with just a few lines of code

Quick Setup

Three simple steps to complete observability

1Install the SDK

pip install agenttrace-sdk

2Configure API Endpoint

# Set your AgentTrace API endpoint
export AGENTTRACE_API_URL="http://localhost:8000"
export AGENTTRACE_PROJECT="my-agent"

3Instrument Your Code

from agenttrace import AgentTrace

client = AgentTrace()
with client.trace("agent-execution") as trace:
    trace.step("processing", metadata={"input": query})
    # Your agent logic here
    trace.step("complete", metadata={"output": result})

How We Compare

Open-source, self-hostable alternative to existing solutions

FeatureAgentTraceLangSmithHeliconeArize AI
Agent-Focused-
Multi-Step Reasoning-
Cost Analytics
Self-Hosting Option---
Simple Setup--
Open Source---
Starting PriceFree$39/moFree$500+/mo

Ready to get started?

Open-source observability platform for AI agents. Get started in minutes.