A SumatoSoft study reveals that approximately 25% of AI agents operate without human oversight, while 80% of deployed AI agents were subsequently withdrawn following their launch. The findings suggest significant challenges in AI agent reliability and post-deployment performance management across organizations.