Why governing AI agents starts with the identities and permissions they inherit.
August 20 is National Bacon Day, so we’re leaning in! Join P0 Security for a practical 30-minute session on the layers behind agentic access control: the agent, the invoking identity and the permissions already sitting underneath.
Every AI agent action has three layers: the Bot, the Launcher and the Target. That is the BLT problem of agent access (see what we did there?).
You cannot govern the agent without understanding who or what invoked it and what system it is trying to access.
If those identities have standing privilege, stale access or poorly scoped permissions, your agent governance starts with a bad baseline.
We’ll walk through why every agent action is a blended identity event, how identity hygiene for users and NHIs shapes what agents can safely do and how teams can build toward agentic runtime governance without treating it as a totally separate program.
Also: the first 10 people who attend live will get Snake River Farms bacon on us! Must attend the live session to qualify. Limit one gift per person. No purchase necessary. Void where prohibited or restricted by employer policy.
What you will learn:
- Why every agent action is a BLT event: bot, launcher and target
- How to evaluate the agent, the invoking identity and the target system together
- How standing privilege, stale access and unmanaged lifecycle issues weaken agent governance
- Where attribution breaks down when users, NHIs and agents are governed separately
- How just-in-time access, approval workflows and ZSP create a cleaner model for agent authorization
- How to assess whether your current identity baseline is strong enough for enforceable agent policy
Who should attend:
- Security and identity teams deploying or preparing to deploy AI agents in production who want to understand how human and NHI identity hygiene strengthens — and enables — effective agentic access control.
