MasterClass is developing AI teaching agents designed to reduce the costs and staffing requirements of personalized one-to-one tutoring. The company's new program, MasterClass Executive, utilizes a multi-agent system that adapts lessons based on student engagement, including monitoring for cognitive overload and motivation levels.
The platform runs approximately 10 agents for every student interaction. To manage these systems at scale, MasterClass has adopted the W&B Weave platform from CoreWeave to monitor, trace, and improve agent performance in production environments.
MasterClass built a custom agent using the Model Context Protocol interface to analyze interaction logs nightly and identify potential issues. The company reports high demand for its programs, with the first cohort receiving 30 000 applications for 500 spots, while the second cohort has reached nearly 50 000 applications.
