Myna is an A/B testing service, a type of tool for testing the effectiveness of proposed website changes. It does so by presenting some visitors with the original site and others with the updated version of the site. The impacts of both versions on user behavior are then monitored and compared. Myna employs multi-armed bandit algorithms, a machine learning technique that optimizes experiments as they run, resulting in more conversions and faster tests. Myna can be integrated with the user's website using its RESTful API, which employs calls issued in JSON and JSONP.