Skip to main content

​
1. Sign up

If you don’t have an Adaline account yet, create one by signing up at app.adaline.ai. After you sign up and log in, Adaline creates a sample project called Get Started. Open the project picker dropdown and select Get Started to follow these quickstarts. The project includes starter resources you can use to run a prompt and try evaluations.

​
2. Setup an AI provider (optional)

New Adaline accounts include up to three free Playground and evaluation runs, so you can try the product before adding provider credentials. This step is optional for the first few runs, but highly recommended if you plan to keep using Playground, run evaluations, or enable continuous evaluations. After the free runs are used, add your own AI provider credentials so Adaline can call the models you choose. Open workspace settings, choose Providers, and add the provider credentials you want Adaline to use. For the full walkthrough, see Configure AI provider. Configure an AI provider in Adaline

​
3. Explore the Evaluators

Evaluators are the scoring functions that assess your prompt’s output. Each evaluator measures a different dimension of quality, giving you a quantified view of how your prompt is performing. Open the sample prompt, then review its evaluators. You can also use the project-level Evaluators library to see evaluator definitions across prompts. View evaluators

​
4. Explore the Dataset

A dataset is a collection of test cases. Each row in the dataset represents one test case, and each column maps to a variable in your prompt or evaluator. Click on the sample Dataset in the sidebar. View dataset Inspect the rows and columns. Notice how the column names correspond to the variable placeholders in the prompt — these values will be substituted into the prompt at runtime for each test case. There is also an extra column called correctResponse which is used to store the expected output for the evaluator to compare against, not directly used in the prompt.

​
5. Run your Evaluation

Now that you understand the building blocks — a prompt, a dataset, and evaluators — here is how they all come together when you run an evaluation.
1
Open the prompt’s evaluation workflow and click Evaluate.Run evaluation
2
When an evaluation runs, Adaline takes each row in your dataset and uses it as a test case:
  1. The variable values from the row are substituted into the prompt’s placeholders.
  2. The prompt is sent to the configured model, and the model generates a response.
  3. The response is then passed through every evaluator attached to the prompt.
  4. Each evaluator produces a quantified score for that test case.
This process repeats for every row in the dataset, giving you a complete matrix of scores across all test cases and evaluators.
3
Once the run completes, review the results table. Each row shows the test case inputs, the model’s output, and the score from each evaluator.View evaluation resultsUse these results to compare prompt quality and performance over time. As you improve your prompt - adjusting instructions, switching models, or tuning parameters - re-running evaluations gives you an objective, quantified measure of whether your changes are improving or degrading output quality.
Congratulations! You just ran your first evaluation in Adaline. Read more about Evaluators and Datasets.