Cisco has unveiled a suite of agent‑centric security innovations designed to let organisations deploy AI agents at scale while preserving control, visibility and auditability — developments of acute interest to CISOs across Asia as enterprises move from pilots to production.
“AI agents aren't just making existing work faster; they're a new workforce of co‑workers,” said Jeetu Patel, President and Chief Product Officer at Cisco. “The only limit is imagination, and security teams are the key to unlocking this opportunity by making the agentic workforce safe enough to trust.”
Key moves centre on four practical themes for security leaders: identity and zero trust for agents; pre‑deployment hardening; runtime policy enforcement; and SOC automation at machine speed.
Identity and Zero Trust for agents
Cisco extends Zero Trust Access to non‑human identities with agent discovery in Cisco Identity Intelligence and agentic IAM in Duo. Agents can be registered, mapped to accountable human owners and given fine‑grained, time‑bound permissions routed through a Model Context Protocol (MCP) gateway.
For Asia, where hybrid estates and third‑party integrations proliferate, this approach helps close visibility gaps that often leave automated workloads untracked and unaudited.
Hardening agents before production
Cisco AI Defense: Explorer Edition offers self‑serve red‑teaming tools that let developers and AppSec teams test models and agent workflows against prompt injection, jailbreaks and multi‑turn attacks before deployment.
That capability addresses a common regional pain point — rapid AI adoption without sufficient security testing — and supports compliance evidence gathering for regulators in markets such as Singapore, Hong Kong and India.
Runtime controls and secure frameworks
The open‑source DefenseClaw framework automates security scanning, inventory and sandboxing of agent “skills”, with planned integration into NVIDIA OpenShell to simplify secure runtime.
Embedding policy enforcement via an Agent Runtime SDK for major agent frameworks (LangChain, AWS Bedrock AgentCore, Google Vertex Agent Builder, Azure AI Foundry) is particularly relevant for APAC organisations balancing cloud sovereignty and vendor diversity.
Agentic SOC and machine‑speed response
Cisco’s Splunk innovations — Exposure Analytics, Detection Studio, Federated Search and specialised agentic SOC agents (Triage, Guided Response, Malware Threat Reversing, etc.) — aim to shift SOCs from reactive triage to proactive, automated response.
As Ryan Morris, president of Blackwood, put it: “This is exactly the innovation required to help security teams stay ahead of constantly increasing and evolving SOC workloads.”
