Creating test data from plain text instructions
- Last Updated: November 10, 2025
- 3 minute read
- Corticon
- Documentation
The Corticon AI Assistant can automatically generate test data for your rulesheets based on plain text instructions. This makes it easy to create input data that exercises all rule conditions, helping you validate behavior and ensure complete coverage without manually building test cases.
How it works
- The rule definitions in the rulesheet.
- The scoped vocabulary—the subset of vocabulary elements (entities, attributes, and associations) used within that rulesheet.
- Which vocabulary elements are used as inputs versus outputs. This is important as it allows the assistant to only generate data for inputs and thus making the tests set concise and more readable.
Using this information, the AI produces structured and realistic test inputs that align with your model and reflect how the rules operate.
Generating test data automatically
This command generates input records designed to exercise each rule in the rulesheet, performing a form of white box testing—testing that uses knowledge of the rule logic to ensure every condition and action path is executed at least once.
Once the response is received, the generated data is immediately added to the input in the rule test sheet. If the results are not what you intended, you can press Ctrl+Z to undo and remove the generated data. The AI Assistant only adds data—it never deletes existing data in a test sheet. Additionally, you can refine the test data by editing the tests.
Providing more focused instructions
You can guide the AI by providing more detailed or targeted prompts to shape the test generation process. For example:
• “Generate test data for 10 applicants over age 65.”
• “Create 5 inputs that will exercise Rule #3.”
• “Add examples where claims should be denied due to missing coverage.”
This flexibility lets you generate data that emphasizes specific conditions or edge cases. The AI uses your additional criteria to tailor the output while still conforming to the rulesheet and vocabulary scope.
How the scoped vocabulary influences generation
When generating test data, the AI Assistant uses the scoped vocabulary and the rulesheet’s structure to ensure all generated data fits correctly. It matches the input structure to the entities and associations actually referenced by the rules.
A good example is the Cargo sample project:
Reviewing AI Reasoning
Reviewing this output can confirm that the generated test data accurately represents the expected logic and provides insight into why certain combinations were chosen.