Arize Python SDK Quickstart Guide

This notebook contains runnable code examples that accompany the three-part Arize AX Python SDK tutorial. It walks through the complete workflow:

First, you will instrument a multi-agent application to send traces to Arize AX & evaluate those traces to measure performance. Then, you will create datasets and run experiments to systematically improve your application.

Use this notebook alongside the documentation - run each cell as you follow along with the guides to see the SDK in action and build a working financial research agent from scratch.

The notebook is designed to be executed sequentially, with each section building on the previous one, so you'll have trace data to evaluate and then use those evaluations to guide your experiments.

Send Traces to Arize AX

Send Traces to Arize AX

Run and Log Evals to Measure Performance

Iterate with Experiments