From Pilot to Production: Guardrails, Failure Modes and Error Propagation in Agentic AI Workflows

Oct 28, 2026
HPC & AI infrastructure

Analyze how error propagation, tool misuse and cascading failures emerge in multi-agentic AI systems.

Explore the guardrails needed to support reliable deployment, including orchestration controls, human oversight, monitoring and fallback mechanisms.

Assess what organizations need to move agentic workflows from promising pilots into routine use in regulated pharma environments.

Speakers
Arvind Rao
Arvind Rao, Professor - Uni of Michigan