Coval is a simulation and evaluation platform for AI voice and chat agents. Inspired by autonomous-vehicle testing, it simulates end customers across realistic scenarios, personas, accents, and background noise, then scores agent behavior with built-in and custom metrics. - View it on GitHub