CavPoint develops software tools to reduce the costs of testing automated vehicle technologies. Automated/autonomous vehicles must operate in environments with extremely high levels of variability and it is in practice impossible to test all of these variations before launch. There is the risk that the deployed system will encounter an untested scenario in the field, and not behave appropriately or operate in an unsafe manner. Our machine learning based software actively searches for the combinations of variations where the system has undesirable behaviour so that the issue can be identified and addressed during the development phase. This reduces the overall time and cost of testing, improves the robustness of the deployed system, and accelerates the time to market.