So we're talking about self tuning database servers?
Indexes that are auto-created based on data access pattterns like statistics.
Deployments that will auto-rollback in the event of an error, without creating one big transaction, something similar to Oracle's RESTORE POINT and FLACHBACK technology.
Intrusion detection and mitigation based on ususual data access patterns (ie: D R O P TABLE SalesHistory; or SELECT * FROM EmpoyeeSalary;).
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho