Date: Tuesday, August 18, 2026
Time: 5:30 PM – 7:00 PM (Presentation starts at 6:00 PM)
Featured Speaker: John Franolich consults for Anthropic on security red/blue team AI harnesses and model products. John worked in every industry vertical. Successful Security AI projects John led include securing Google Cloud Platform for an FDA-regulated digital Radiology Type I Medical Device and Financial Modeling in secure Microsoft Azure environments for a large financial institution. John currently works for a global corporation doing cyber threat intelligence investigating cyber criminals and nation state actors.
Topic: Measuring Security AI Agents. Evals, Benchmarks and Frameworks. How did it achieve success? vs. Did It Work?
Details: Discuss a programmatic agentic framework to shift the focus from "did it work?" to measuring "how did it achieve success?" Organizations can use the CLASP Framework to measure success. CLASP stands for Closed‑Loop Autonomous Security Performance. CLASP allows an organization to Measure implementations like different SOC AI Agent Architectures using security efficacy and agentic efficiency metrics for designs like single LLM, multi‑agent graph, LLM Judge, human‑in‑loop augment etc.