wayfinder
A step-by-step implementation guide to AI evaluation systems using a flight search example.
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
Wayfinder is a reference implementation for learning and building AI evaluation systems. It extends the same flight search application to demonstrate rule-based evaluation, human evaluation, LLM judging, online evaluation, and experiment comparisons. Its examples use OpenAI models to generate AI responses and LangSmith for tracing and some evaluations.
KEY FEATURES
Key features
- Uses an AI flight search app as an evaluation example
- Provides rule-based and human evaluation examples
- Provides LLM-judge evaluation examples
- Provides online evaluation and experiment comparison examples