Project configuration
Set the LLM use case type for your evaluation project.
Project configuration
This tab sets what type of LLM application your project evaluates. The choice determines which scorers and metrics are available when you run experiments.
Use case types
Each project targets one of three use case types:
RAG
Evaluate retrieval-augmented generation. Scorers measure recall, precision, relevancy and faithfulness.
Chatbot
Evaluate single and multi-turn conversations. Scorers focus on coherence, correctness and safety.
Agent
Evaluate AI agents for planning, tool usage, task completion and step efficiency.
Setting the use case
- Open your project and click the Configuration tab in the sidebar.
- Select RAG, Chatbot or Agent from the radio buttons.
- Click Save changes.
Use case locking
Once you create your first experiment, the use case becomes locked. This prevents inconsistencies between your experiments and the configured scorers.
If you need to evaluate a different use case, create a new project. The configuration page shows a Create new project button when the use case is locked.