> ## Documentation Index
> Fetch the complete documentation index at: https://next.gentrace.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Overview of Gentrace's evaluation system.

Gentrace's agent evaluation operates in three steps:

1. Create a [dataset](/evaluation/datasets) with test cases for your AI pipeline
2. Run an [experiment](/evaluation/experiments) using [unit tests](/evaluation/unit-tests) and/or [dataset tests](/evaluation/dataset-tests)
3. Analyze results with [Gentrace Chat](/error-analysis/chat) and [derivations](/error-analysis/derivations) to extract insights and monitor performance

```mermaid theme={null}
flowchart TD
    A[Datasets] --> B[Test Cases]

    C[Experiment Context]
    B --> E[Dataset Tests]
    C --> D[Unit Tests]
    C --> E

    D --> F[OpenTelemetry Traces]
    E --> F

    F --> G[Experiment Results]
    G --> H[Derivations]

    classDef prep fill:#1a2e35,stroke:#18375d,color:#fff
    classDef exec fill:#2e1f25,stroke:#53242e,color:#fff
    classDef traces fill:#312724,stroke:#5e3d2b,color:#fff
    classDef analysis fill:#1d2e33,stroke:#205258,color:#fff

    class A,B prep
    class C,D,E exec
    class F traces
    class G,H analysis
```

## Next steps

* Set up [experiments](./experiments) to run systematic evaluations
* Create [datasets](./datasets) to organize your test cases
* Use [derivations](/error-analysis/derivations) to analyze your results