Arize is now part of Dynatrace

See your AI, your apps and your infra as one chain.

Arize AI brings AI-native tracing, evaluation and experimentation. Dynatrace brings full-stack monitoring of applications, infrastructure and user experience. Together, they show you exactly what your AI agents do and everything underneath them.

Start monitoring today See how it fits together →

Infrastructure · Applications · AI agents · End to end

trace 7f3a…c21 · support-agent1.24s

agent.plan182ms
└ llm.completion640ms
└ retriever.query95ms
└ checkout-api GET /cart41ms
└ postgres SELECT12ms
└ k8s node gpu-pool-3CPU 71%

Eval score

0.94

p95 latency

1.3s

Infra health

The entire chain

When an agent gives a bad answer, the cause can sit anywhere in the stack.

A hallucination might start with a weak prompt, a stale retrieval, a slow API or an overloaded GPU node. Monitoring one layer at a time leaves gaps. Arize AI by Dynatrace follows the signal all the way down.

L3 · AI

AI agents & LLM apps

Every model call, retrieval, tool use and agent step traced and scored — not just the final answer, but how the agent got there.

LLM spansEvalsHallucination rateToken cost

L2 · Apps

Applications & services

Distributed traces across microservices, APIs and user sessions, so a slow answer is linked to the exact service that caused it.

TracesLatencyErrorsUser experience

L1 · Infra

Infrastructure & cloud

Hosts, Kubernetes, GPUs, networks and cloud services, with dependencies mapped automatically from the bare metal up.

HostsK8sGPUNetwork

What you get

AI observability that understands the whole lifecycle

AI-native tracing

Capture the full execution path of agents and LLM workflows: prompts, completions, retrievals and tool calls as structured spans.

Evaluation

Score outputs and intermediate steps with LLM-as-judge and custom evaluators to catch hallucinations, drift and regressions.

Experimentation

Compare prompts, models and agent designs against real datasets before shipping, then keep measuring in production.

Full-stack context

Dynatrace connects AI behavior to the applications, infrastructure, user experiences and business processes underneath it.

Open source + enterprise

Arize Phoenix stays open source, and Arize AX remains the enterprise platform, both on a path into one Dynatrace experience.

Built for production

Observability across the whole lifecycle, from the first experiment to agents running at scale.

How it works

From blind spots to root cause in three steps

Instrument

Deploy Dynatrace for infrastructure and apps, and add OpenTelemetry-based tracing to your AI workloads.

Correlate

Agent steps, service traces and host metrics land in one view, linked by context instead of by guesswork.

Evaluate & improve

Score quality, find root causes across layers, and ship better models and agents with confidence.

The story

Why Arize joined Dynatrace

AI agents create far more telemetry than any human can read, and their failures rarely stay in one layer. Dynatrace acquired Arize to pair Arize's AI-native tracing, evaluation and experimentation with Dynatrace's monitoring of applications, infrastructure, user experiences and business processes.

Arize founders Jason Lopatecki and Aparna Dhinakaran joined Dynatrace, with the Arize team continuing to build Phoenix and AX while capabilities come together into a single AI observability offering.

Announced: Aug 2026
Completed: Oct 2026
Deal value: $915M
Platforms: Phoenix · AX