Why Prompt-Level AI Safety Isn't Enough
Amir Khan, Co-Founder and COO of PureLogics, speaks with Harold Byun, CEO of BlueRock, about why AI agent security has to go beyond the prompt. They discuss runtime guardrails, agent-to-agent risk, shadow AI, and who should own AI governance in the enterprise. Harold also breaks down how enterprises can prevent agents from going rogue, manage runaway token costs, and where liability lands when something goes wrong.PureLogics Pulse Episode Chapters:(00:00) Introduction and Episode Setup(02:06) What BlueRock Does and the Agentic Execution Layer(04:07) Key Capabilities: Guardrails, MCP Registry…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.