Show notes
Sid Mangalik and Andrew Clark explore the unique governance challenges of agentic AI systems, highlighting the compounding error rates, security risks, and hidden costs that organizations must address when implementing multi-step AI processes. Show notes: • Agentic AI systems require governance at every step: perception, reasoning, action, and learning • Error rates compound dramatically in multi-step processes - a 90% accurate model per step becomes only 65% accurate over four steps •...

