How do PostgreSQL connection pooling and autoscaling work in a managed service?

How do PostgreSQL connection pooling and autoscaling work in a managed service?