Infrastructure Identity Moves Into the Agent Era
Infrastructure identity is becoming a higher priority as enterprises prepare to let AI agents interact with production systems. In this Techstrong TV interview from Black Hat 2026, Diana Jovin, CMO at Teleport, explains why identity can no longer stop with human users and traditional machines. Modern environments now include workloads, services and autonomous agents that need secure access without static credentials.
Teleport’s approach is built around a unified identity layer for humans, machines, workloads and agents. That model is designed to reduce identity fragmentation while helping engineering teams move faster. It also supports zero trust by removing standing privileges and replacing long-lived credentials with short-lived, cryptographically backed access.
Zero Trust Needs to Evolve for Agents
The conversation explores why traditional zero trust principles still matter, but need to be extended for agentic systems. Jovin argues that “verify explicitly” becomes continuous enforcement when agents are involved. Since agents can run quickly and persistently, organizations need visibility into what they are doing at all times.
Infrastructure identity gives teams a stronger foundation for that control. It helps define what agents can access, how long they can operate and which actions should be reviewed. That becomes critical when many agents work together, because individual actions may appear harmless while the combined effect creates risk.
Teleport Beams Creates a Trusted Runtime
Teleport Beams is presented as a trusted runtime for infrastructure agents. Jovin describes it as an ephemeral microVM that embeds identity, access control and auditability into the agent’s operating environment. When the task ends, the runtime expires and cleans itself up.
That model gives organizations a way to apply least privilege, continuous monitoring and audit trails to agent behavior. It also supports a broader shift from assuming breach to assuming misalignment. If an agent drifts from its intended objective, teams need evidence, context and intervention options.
Security Becomes the AI Adoption Gate
The discussion closes with a practical view of AI adoption. Jovin notes that many organizations are still early in the transformation, even if the AI conversation feels crowded. The next phase is less about plumbing and more about putting AI-native practices into real operational environments.
For security and engineering teams, infrastructure identity can help make that transition safer. As agents begin deploying software, accessing databases and taking operational action, identity-backed controls will become essential. Teleport’s message is that trusted agent adoption starts with a foundation that treats every actor as a first-class identity.