Testing Systems That Think: How We Evaluate AI Agents

CIRCLE LEARNING

Vaibhav Sharma

Knowledge Host
Sunday, 15 Mar 2026

About the session  A hands-on walkthrough of how quality assurance must evolve when the system under test is non-deterministic.

Key takeaways

01

Why traditional testing approaches don’t work for AI agents

02

How to evaluate hallucination, jailbreak attempts, bias and guardrails

03

How to check response consistency in AI systems

04

A practical way to build a free, local evaluation setup using Ollama