Uploaded August 2026 | Updated November 2026, 4 days ago
AI agents don't leave you a source of truth you can just read, like code does. See how Arize AX helps you trace what your agent did, evaluate whether it worked, and fix what's broken, starting with Alyx, Arize's AI engineer, and Signal, your agent SRE.
Watch this to learn:
• How Arize AX traces every LLM call, tool call, and agent step, so you have a source of truth when code alone can't give you one
• How Alyx, Arize's AI engineer, builds an eval for you from a plain-English request and reads your traces to configure it
• Why evals only catch the failure modes you already know to look for, and how Signal finds the ones you don't
A closer look at tracing, evaluating, and fixing agents in Arize AX.
⏱️ Chapters
0:00 Why verifying AI agents is hard
0:47 How Arize AX traces every agent step
1:08 How to read an agent trace
1:48 How to set up an eval with Alyx
2:04 How Alyx builds a hallucination eval automatically
2:56 How to read an eval's explanation
3:26 Why evals only catch known failure modes
3:50 How Signal finds unknown failures
4:09 A real failure Signal caught
4:34 How Signal opens a PR to fix it
5:04 Recap: trace, evaluate, and fix with Arize AX
🔗 Try Arize AX & Phoenix OSS: arize.com/ax
📚 Docs: docs.arize.com
🔔 Subscribe for weekly content on LLMs, agents, and evaluation: youtube.com/@arizeai?sub_confirmation=1
#ArizeAI #AIAgents #AgentObservability #LLMEvaluation #AIEngineering
AI agents don't leave you a source of truth you can just read, like code does. See how Arize AX helps you trace what your agent did, evaluate whether it worked, and fix what's broken, starting with Alyx, Arize's AI engineer, and Signal, your agent SRE.
Watch this to learn:
• How Arize AX traces every LLM call, tool call, and agent step, so you have a source of truth when code alone can't give you one
• How Alyx, Arize's AI engineer, builds an eval for you from a plain-English request and reads your traces to configure it
• Why evals only catch the failure modes you already know to look for, and how Signal finds the ones you don't
A closer look at tracing, evaluating, and fixing agents in Arize AX.
⏱️ Chapters
0:00 Why verifying AI agents is hard
0:47 How Arize AX traces every agent step
1:08 How to read an agent trace
1:48 How to set up an eval with Alyx
2:04 How Alyx builds a hallucination eval automatically
2:56 How to read an eval's explanation
3:26 Why evals only catch known failure modes
3:50 How Signal finds unknown failures
4:09 A real failure Signal caught
4:34 How Signal opens a PR to fix it
5:04 Recap: trace, evaluate, and fix with Arize AX
🔗 Try Arize AX & Phoenix OSS: arize.com/ax
📚 Docs: docs.arize.com
🔔 Subscribe for weekly content on LLMs, agents, and evaluation: youtube.com/@arizeai?sub_confirmation=1
#ArizeAI #AIAgents #AgentObservability #LLMEvaluation #AIEngineering










