Configure AI-driven test case generation
Set up AI-assisted test case generation end-to-end in SquashTM, from configuring the AI server to the first generations performed by your testers.
Required license
AI-driven test case generation is available with a SquashTM Ultimate license 💎.
Why configure generative AI in SquashTM?
Configuring generative AI lets your testers benefit from the following capabilities, directly from any requirement:
- automatically generate classic test cases (with test steps, prerequisites, importance, reference, and datasets);
- enrich generation with linked requirements and attached documents;
- keep editorial control: every suggestion must be validated by the tester before being saved, and remains editable afterwards.
AI is a writing aid
The tester remains responsible for the quality, relevance, and completeness of the generated test cases.
What you will learn
- Declare an AI server at the instance level (Anthropic, OpenAI, Mistral AI, Azure OpenAI, Google Vertex AI, or a custom server).
- Prepare a prompt set tailored to your needs.
- Enable AI on a project by associating it with a server and a prompt set.
- Verify that the full chain works through a first generation.
Who this guide is for
- SquashTM administrators: configure the AI server, the prompt set, and enable AI at project level.
- Project leaders: enable AI on the project when this configuration is delegated to them.
- Testers: validate the setup through a real generation.
What you need
- A SquashTM Ultimate 💎 instance.
- An administrator account for the global steps.
- A valid API key with one of the supported providers (see the full list in Manage Artificial Intelligence servers).
- A project containing at least one filled-in requirement to test generation.
Workflow overview
The deployment follows a logical order: each step builds on the previous one, and the last step validates the whole chain.
- AI server (global): declared once for the whole instance.
- Prompt set (global): reusable across multiple projects.
- Project enablement: a project leader or an administrator associates a server and a prompt set with the project.
- User-side generation: from a requirement, the tester produces their test cases.
Validation
If the generation at step 4 works, the entire configuration chain is correct. Conversely, if the Generate test cases button is missing on a requirement, one of the steps was likely skipped (see the dedicated section).
Guide structure
- Configure the AI server: declare the provider and test the connection.
- Prepare a prompt set: use the default prompt sets or create a new one.
- Enable AI on a project: link the server and the prompt set to a project.
- Generate test cases: first validation generation on the user side.
Each step links back to the corresponding reference page in the documentation for granular details (advanced parameters, Handlebars syntax, per-provider configuration, etc.).