At Block, we have spent the past several months creating an LLM driven mobile testing framework, “Trailblaze”. It has enabled Square and Cash to run natural language based instrumented tests for Android both locally and in our continuous integration pipeline. This has allowed us to reduce manual effort for authoring and executing tests. This talk goes deep into the inner workings of Trailblaze as we share our lessons learned while building it. Learn the strategies we use to communicate effectively with the LLM and have it decide how to drive the user interface. Mobile AI driven tests are no longer science fiction, it’s here, works well, and will only get better. Come learn from our experiences and find inspiration to leverage AI to make your testing efforts more effective.