Explore the concept of self-driving databases and their potential implementation in PostgreSQL through this keynote talk by Andy Pavlo at Citus Con. Delve into the state-of-the-art in autonomous DBMSs, including current possibilities and remaining challenges. Learn about Carnegie Mellon's efforts to create a true self-driving DBMS through the NoisePage and OtterTune projects. Discover how PostgreSQL's existing instrumentation and extension hooks can be leveraged for ML-based planning components. Gain insights into Pavlo's "wish-list" of features for PostgreSQL that could bring self-driving capabilities closer to reality. Explore topics such as workload capture, system telemetry, what-if API, and action APIs in the context of autonomous database systems.