AI Testability

Evaluation planning, before model testing

Find the gaps in your AI evaluation plan.

AI Testability is a focused worksheet for turning an AI product change into a plan your team can inspect, discuss, and improve before running an evaluation.

Open the plan linter
For

AI product, quality, and engineering teams preparing a model or agent change.

Use it to

Check whether an evaluation plan covers the decisions and evidence reviewers need.

Boundary

It inspects plan text and metadata only. It does not call a model or judge model outputs.

How it works

A short path to a stronger test plan.

  1. 01

    Describe the change

    Write down what changed, who it affects, and the behavior you need to evaluate.

  2. 02

    Lint the plan

    Check whether the plan names coverage, evidence, baselines, and review controls.

  3. 03

    Review what is missing

    Save a clear plan-quality record before your team runs model tests.

Start with a blank plan

Make the evaluation reviewable before the run begins.

Open AI Testability