Artificial Intelligence & Machine Learning
,
Events
,
Governance & Risk Management
Menlo Ventures’ Rama Sekhar on Securing AI Agents and Non-Human Identities
Artificial intelligence agents are transforming enterprise cybersecurity from a tool-based model to an autonomous operating layer. Unlike generative AI interfaces, agents act independently with access to enterprise systems, introducing non-human identities that must be governed and secured.
See Also: Taming the Rise of Shadow AI Agents
Rama Sekhar, partner at Menlo Ventures, said agents introduce new risks because they operate with memory, autonomy and a defined blast radius. As organizations deploy them across workflows, visibility and permission control become critical to preventing overprovisioned access and unmanaged shadow AI.
“The problem is no longer finding issues; it’s fixing and prioritizing them,” Sekhar said. Enterprises must move beyond detection toward automated remediation, where agents can identify and patch vulnerabilities.
“We have to use AI to fight AI,” Sekhar said, noting manual approaches cannot keep pace with AI-driven attacks.
In this video interview with Information Security Media Group at RSAC Conference 2026, Sekhar also discussed:
- Why non-human identities are becoming the defining enterprise security challenge;
- How visibility and governance, knowing where agents are and what they can access, are critical in securing agentic AI;
- How just-in-time access reduces risks from overprovisioned agents.
At Menlo Ventures, Sekhar focuses on investments in cybersecurity, AI and cloud infrastructure. He partners with founders to build the next generation of cybersecurity and infrastructure. Before joining Menlo, he spent nearly 15 years at Norwest Venture Partners leading investments in Veza, Harness, Agari, SourceClear, InfluxData and Dremio.

