Uploaded September 2026 | Updated November 2026, 14 hours ago
Add tracing to an AI agent in minutes with Arize AX. In this quick walkthrough, see how a coding agent can instrument an existing application, verify that traces are flowing, and get you ready to debug, evaluate, and improve your agent.
Tracing gives you visibility into what happens between an agent’s input and output, including its model calls, tools, and execution flow. It also provides the foundation for building evals and finding production issues as your agent evolves.
In this demo, you’ll see how to:
• Start AI agent tracing with a single command
• Use Claude Code to instrument an existing application
• Sign in to or create an Arize AX account
• Automatically detect the application you want to trace
• Add the required instrumentation and configuration
• Run the application and verify that traces reached Arize AX
• Inspect the resulting agent traces
• Use those traces as the foundation for evals and Signal
Once tracing is in place, you have the data you need to understand agent behavior, identify failures, build evaluations, and continuously improve your application.
Chapters:
00:00 Why AI agent tracing should come first
00:22 Start tracing with one command
00:48 Let Claude Code instrument your app
01:18 Choose the application to trace
01:49 Run the instrumented agent
02:18 Verify traces in Arize AX
02:45 What to do next with traces, evals, and Signal
🔗 Explore Arize AX: arize.com/ax
🔗 Learn more about Arize: arize.com
🔔 Subscribe for more videos about AI agents, tracing, evals, and observability:
youtube.com/@arizeai?sub_confirmation=1
#AIAgents #AgentTracing #AIObservability
Add tracing to an AI agent in minutes with Arize AX. In this quick walkthrough, see how a coding agent can instrument an existing application, verify that traces are flowing, and get you ready to debug, evaluate, and improve your agent.
Tracing gives you visibility into what happens between an agent’s input and output, including its model calls, tools, and execution flow. It also provides the foundation for building evals and finding production issues as your agent evolves.
In this demo, you’ll see how to:
• Start AI agent tracing with a single command
• Use Claude Code to instrument an existing application
• Sign in to or create an Arize AX account
• Automatically detect the application you want to trace
• Add the required instrumentation and configuration
• Run the application and verify that traces reached Arize AX
• Inspect the resulting agent traces
• Use those traces as the foundation for evals and Signal
Once tracing is in place, you have the data you need to understand agent behavior, identify failures, build evaluations, and continuously improve your application.
Chapters:
00:00 Why AI agent tracing should come first
00:22 Start tracing with one command
00:48 Let Claude Code instrument your app
01:18 Choose the application to trace
01:49 Run the instrumented agent
02:18 Verify traces in Arize AX
02:45 What to do next with traces, evals, and Signal
🔗 Explore Arize AX: arize.com/ax
🔗 Learn more about Arize: arize.com
🔔 Subscribe for more videos about AI agents, tracing, evals, and observability:
youtube.com/@arizeai?sub_confirmation=1
#AIAgents #AgentTracing #AIObservability










