Upcoming · 45 min
Live now
Available on demand soon
On demand · 45 min
Architecting Zero Standing Privilege for AI agents
Vaults, checkouts and scheduled rotations were built around long-lived secrets and human-paced sessions. AI agents request access at machine speed, often many times in a single task. Learn how to rebuild PAM so agents get only the access a task needs, only for as long as it runs.
Learn more about using Venice. Book a demo


PAM was built on a simple assumption: a known identity, a long-lived credential, a predictable session. Cloud strained that assumption. AI agents have broken it.
Access control is still where AI agent risk is decided, but the model has to change from vaults and rotations to ephemeral, policy-driven access for humans, NHIs, and AI agents. Shir and Uriel will cover the architecture, policy, and incident response behind that shift. They’ll also show live how Venice ties each AI agent request to its owner and issues credentials that expire when the task ends.
What you'll learnWhat you'll learnWhat's coveredWhat's covered
- Why PAM, not a separate “AI security” tool, is the right control point for AI agents
- Why humans, NHIs, and AI agents belong under one policy engine, and the architecture choices that make it work
- How to let policy decide routine AI agent requests in real time, and when to bring a human in
- How ephemeral, just-in-time credentials remove standing privilege without becoming a bottleneck
- See how Venice brings human and AI context together to control AI agents
Presenters

Shir Hirshman
Sr. PM Agentic @ Venice

Uriel Zilberberg
Tech Alliances @ Venice