Does Arize work with my stack?

Arize integrates with 40+ models, frameworks, and AI tools, including OpenAI, Anthropic, Google, LangGraph, LangChain, LlamaIndex, CrewAI, and more. Built on OpenInference and OpenTelemetry standards, it works across Google Cloud, AWS, Azure, and self-hosted environments.

What about data privacy?

Your data stays under your control. Arize supports flexible deployment options (including self-hosting) and is certified to leading security and compliance standards, including SOC 2 Type II, ISO 27001, PCI DSS, HIPAA, and GDPR.

What’s the difference between Arize products?

Phoenix is the open-source platform for local AI observability and evaluation. Arize AX is the managed platform built on the same standards, adding scaled infrastructure (adb), managed agents, online evals, and continual improvement workflows.

How does the 'Signal' agent work?

Signal automatically analyzes production traces, clusters recurring failures into ranked issues, provides trace evidence for each, and generates proposed fixes tied to your repository code. It surfaces what matters without manual triage.

Directory/Coding/AI Development/Arize AX

Arize AX

Arize AX is an AI engineering platform for improving AI agents and applications. It captures traces of real behavior, automatically surfaces recurring issues with Signal, evaluates quality with LLM-as-a-judge or code evaluators, and runs controlled experiments to prove improvements before deployment. Built on OpenInference and OpenTelemetry standards, it integrates with 40+ models and frameworks and supports flexible deployment, including self-hosted options. The platform is designed for AI engineers, ML engineers, and founders who need to debug, monitor, and continuously improve production AI systems. Core capabilities include auto-instrumentation, trace exploration, custom dashboards, online and offline evaluations, and prompt experimentation with versioning.

Coding

Visit website

Added on

Sep 22, 2026

Agent ObservabilityAI EvaluationLLM TracingOpen StandardsProduction Debugging

Product information

Everything worth knowing about Arize AX

The complete picture — from everyday features to the technical detail builders and IT folks dig for.

Core features

What it actually does

Trace & Monitor Agent Behavior

Automatically capture detailed traces of every LLM call, tool use, and agent run to understand exactly what happened during execution.

Automatic Issue Detection (Signal)

AI-powered system that scans production traces to surface recurring failures, group them into ranked issues, and propose root cause fixes.

Comprehensive Evaluation Framework

Run unlimited online and offline evaluations using LLM-as-a-Judge, agent-as-a-judge, custom code evaluators, or human annotations to score agent quality.

Experimentation & Prompt Testing

Build datasets, run controlled experiments comparing prompts, models, or agent changes, and use playgrounds to test before deploying to production.

Alyx AI Engineering Agent

An in-product AI agent that helps you run evaluations, debug issues, and improve your agents by answering plain language questions about your data.

Technical capabilities

OpenTelemetry & OpenInference Standards

Built on open standards (OpenInference, OpenTelemetry) to ensure vendor-neutral instrumentation and compatibility with 40+ models, frameworks, and AI tools.

ADB Datastore & Data Fabric

Proprietary GenAI trace datastore (ADB) storing data in open formats, enabling native integration with BigQuery, Databricks, or Snowflake via Data Fabric.

Must watch videos

Tutorial

How to Evaluate AI Agents: Full Course

AI Agents Mastery is a popular hands-on course we are updating for 2026 for AI engineers and ambitious builders who want to go beyond toy chatbots and ship real, production-ready agents. Over seven modules and a series of guided labs, you’ll learn how to design agent workflows, wire up tools and the Model Context Protocol (MCP), build RAG and agentic RAG, add deep observability, and evaluate your agents with robust metrics and LLM-as-a-judge patterns. You’ll work directly in Python with a GitHub lab repo, instrumenting and improving a single agent system as you go—from the very first “hello world” request all the way to post-deployment monitoring, cost tracking, and guardrails—so you leave with both conceptual understanding and a working, extensible agent stack.

Tutorial

Arize Observe 2026

The AI Agent Evals Conference | June 4, 2026 in the San Francisco Ferry Building

Tutorial

Rise of the AI Engineer

If LLMs were the spark, AI and agent engineers are the ones building the power grid. Blending a unique set of skills – part SWE, part systems/infra, part AI researcher – they are the heroes who actually make agents work in the wild. This series spotlights agent engineering thought leaders to dissect their use cases, best practices, and what they see on the bleeding edge.

Who uses it

Real use cases, no hype

Agent Debugging

Use 'Signal' and tracing to identify and fix recurring failures in production AI agents.

LLM Evaluation

Run comprehensive span, trace, and session evaluations with LLM-as-a-Judge or code evaluators.

Prompt & Model Experimentation

Test changes to prompts, models, or retrieval strategies with controlled experiments before shipping.

Production Monitoring

Track token usage, latency, costs, and create custom dashboards and alerts for AI performance.

Coding Agent Workflow

Set up and manage observability directly from within AI coding assistants like Claude Code or Cursor.

Multi-modal AI Analysis

Trace and evaluate complex interactions involving text, images, voice, and PDFs.

Compliance & Security

Deploy on-premise or use SOC 2, HIPAA, and GDPR compliant cloud infrastructure.

What's great

Free tier is generous with unlimited users, evals, and experiments.
Open-source core (Phoenix) with OpenTelemetry/OpenInference standards.
Integrated AI agent (Alyx) for debugging and improvement workflows.
Enterprise tier supports self-hosted deployment for data sovereignty.

Technical strengths

Agent-first debugging designed for coding agents like Claude and Cursor.
Built-in 'Signal' agent automatically surfaces recurring production issues.

Where it falls short

Free tier limits are restrictive: only 10 signals and 25k trace spans/month.
Key advanced features like custom code evaluators are Enterprise-only.
Complex feature matrix with many tiers can be confusing to navigate.
Pro plan has a 10 GB storage cap, which can be tight for high-volume traces.

Technical limitations

Self-hosted deployment is not available until the custom Enterprise plan.
Data retention is fixed (15-30 days) on managed SaaS tiers.

For technical folks

The deep-dive specs

FAQ

Includes technical Q&A

Connect your agent or application to Arize and send your first trace. Traces show what happened inside each request, giving you the data you need to debug, evaluate, and improve your agents. Start setting up traces for your agent.