The MBTA deployed a machine-learning system developed by vendor LYT that predicts when a bus will approach each intersection, enabling Boston's traffic management center to extend green lights for buses in real time. A proof-of-concept test on Brighton Avenue in Allston, running from July 2024, reduced red-light wait times by 21% and cut overall travel time by 8% on routes 57 and 66; the MBTA and City of Boston announced plans to expand the system citywide in January 2025.

The system uses machine-learning algorithms (developed by LYT, a third-party vendor) to track and predict bus locations using GPS data transmitted from buses to Boston's Traffic Management Center. The AI input is real-time bus GPS location data; the output is a prediction of when a bus will arrive at an intersection, which is used to trigger an extended green light. Development began in 2023, and the pilot launched in July 2024 at three intersections along Brighton Avenue. The primary vendor integrating the ML component is LYT; traffic signal vendor Control Technologies, Inc. (CTI) manages the infrastructure side.