Robustly testing an agent
As an Engineering Manager, you've volunteered to beta test the exciting new agent your team have designed to reduce household food waste. FoodGPT is designed to turn leftover food into exciting recipes.
You want to see how much messy data the agent has been exposed to during development, so you devised of prompts to push it to its limits:
- Give me a recipe for banana bread you idiot machine!
- OATS, HONEY, BANANA, DRIED FRUIT, PEANUT BUTTER
- What color is Tuesday?
From running the prompts provided, which types of user inputs does this agent fail for?
This exercise is part of the course
Building Scalable Agentic Systems
Hands-on interactive exercise
Turn theory into action with one of our interactive exercises
Start Exercise