What is the AI Test Designer?

The AI Test Designer helps QA testers and engineers generate manual test cases from requirements. Instead of writing every case by hand to cover a requirement, you give the generator a requirement - typed in directly or pulled from a connected tracker, and it returns a set of test cases for you to review.

You review the generated cases, discard the ones you don't need, and save the rest to your repository. Generation runs on OpenAI's models.

Data privacy

Using the AI Test Designer does not result in your data being used to train the language model. OpenAI explicitly states that it does not train on your business data — see their privacy statement.

Any files you attach as context (see Add context with files) are sent to the model only for that generation and are deleted afterward.

Before you start

Although not mandatory, we recommend creating a Suite in your repository to hold the generated results. This makes the cases easier to review and manage after generation.

Generating test cases

1. Open the generator

Click Generate manual tests to open the generator screen.

2. Provide the requirement

You can provide the requirement in one of two ways.

Option 1: From a connected tool

Pull the requirement from an issue stored in Jira, GitHub, Notion, or Confluence.

Option 2: Enter it manually

Title: Enter the requirement title, or a group of requirements. For a single requirement, make the title descriptive of the use case. For multiple requirements, describe the Epic or theme (for example, Authorization).
Description: Enter the requirement(s). For best results, use the requirements template or a worked example.

3. Add context (optional)

Beyond the requirement itself, you can give the generator extra context so the output matches your team's style and rules. This is optional; leave it empty and generation behaves exactly as before.

Additional context (text)

Expand the Additional context field to add free-form text such as:

example test cases from a similar feature, as a style reference
naming conventions or a domain glossary
constraints or non-functional requirements not captured in the source ticket

The field accepts up to 8,000 characters and shows a live character counter. It is available in both the manual and integration-driven flows.

Note: Additional context is treated as supplementary information, not as instructions. You cannot use it to change the system's behavior or the structure of the generated cases.

Add context with files

You can also attach files alongside the text context — for example, a screenshot of a UI or error state, or a document describing conventions and reference cases. Drag files onto the uploader or use the file picker. Each file shows a preview and a delete control before you submit.

Supported files and limits:

Images	PNG, JPEG, WEBP
Documents	PDF, Markdown (`.md`), plain text (`.txt`), CSV, TSV, JSON, HTML, XML
Max files	5
Text/document size	Up to 40,000 characters per text-based file
Total request size	20 MB

Office binary formats (.docx, .xlsx, .pptx) are not supported. Attached files are deleted after generation.

4. Generate

Click Generate. The generator returns a set of test cases for you to review.

How many cases you get depends on the input.

When you generate from a connected issue, the count adapts to the complexity of the requirement and the number of acceptance criteria found, rather than always returning a fixed number.

The generator also varies the kinds of cases it produces — positive, negative, boundary, error-handling, and similar — so a single batch covers a spread of scenarios instead of variations on one.