Your product's source-of-truth

• QA Wolf has always been about taking the pain out of end-to-end testing.
• But as AI capabilities evolved, we saw an opportunity to rethink the entire platform from the ground up.
• The vision: an AI-native system that creates a living map of your product and automates test coverage based on that map.

• Product owners often don't have visibility into what's actually being tested.
• Test plans live in spreadsheets, wikis, or (worse) someone's head.
• When features change, there's no easy way to know which tests are affected or missing.

We kept hearing the same question from customers: "What does my test coverage actually look like?"

I worked with the CEO and Director of AI to understand where LLMs could add the most value to our workflow. We landed on exploration and mapping.

App mapping

I prototyped an experience where AI explores a product, identifies user flows, and builds an interactive map. That map gives product owners a shared view of what their product does and what needs testing.

From a map to test coverage

In the proposed workflow, product owners review, adjust, and approve the test plan before automation begins. The approved plan becomes the source of truth for test creation, connecting what the product does to what gets tested.

As the prototype grew across Map, Automate, Runs, Bugs, and Insights, I needed a way to document the rationale behind design decisions and help collaborators get familiar with the product.

I created an interactive design guide for humans, with matching markdown documentation for AI context. It covers the patterns, interactions, and details that help engineers review the proposal and start to iterate with me on it.

For example, Map and Insights use less dense layouts for product owners and leadership, while Automate and Bugs are denser workspaces for engineers. I also documented how we distinguish shipped features from ideas we’re still exploring. These details help people understand how to apply the patterns as the product grows.

When I work with Claude Code, I point it to the guide so it has the context behind earlier decisions. I use it myself when revisiting a flow or figuring out how something new should fit into the product.

Shifting mental models. Our existing customers know QA Wolf as a service. Introducing a self-serve, AI-native tool requires rethinking how we communicate value.

Balancing AI autonomy with user control. Product owners need to trust the AI's suggestions but also feel like they're in the driver's seat. Too much automation feels scary. Too little defeats the purpose.