How to build reliable AI Agents?

Estimated read time 1 min read

Post Content

​ Join Ivan and Aparna Dhinakaran, CPO of Arize AI, as they discuss the challenges of moving AI agents from prototype to production. This tutorial demonstrates how to integrate the Google’s Agent Development Kit (ADK) with Arize Phoenix to enable deep observability, tracing, and evaluation for your agents. Learn how to instrument your ADK agent with OpenTelemetry standards, visualize live traces to debug logic errors, and run evaluations to ensure your agent is reliable at scale.

Chapters:
0:00 – Introduction
0:40 – Why observability matters for agents
1:35 – Callbacks in ADK
2:05 – Arize Phoenix
2:45 – The Financial Advisor agent
3:16 – Installation and setup
4:30 – Demo: Live tracing
5:20 – Error analysis and debugging
7:09 – Scaling debug with evaluations
8:46 – Wrap up

Resources:
Learn more about the Agent Development Kit → https://google.github.io/adk-docs/
Check out the ADK Samples repo → https://github.com/google/adk-samples

Subscribe to Google for Developers → https://goo.gle/developers

Products Mentioned: Google ADK, Google AI   Read More Google for Developers 

You May Also Like

More From Author