On 10 November 2026, the conversation surrounding enterprise artificial intelligence shifted starkly from data centre processing power to the vulnerability of the local networks connecting it all. Speaking at the Cisco Live Asia-Pacific, Japan, and Greater China (APJC) summit in Melbourne, networking executives delivered a blunt warning: traditional enterprise architectures are fundamentally unprepared for the era of autonomous AI agents.

According to Ben Dawson, Cisco’s president for the APJC region, the core failure mode lies in traffic behaviour. Networks engineered to accommodate human work patterns will buckle when thousands of AI agents begin operating autonomously around the clock. While generative AI tools currently act as digital advisors waiting for human direction, the transition toward “agentic AI”—systems that independently execute complex operational workflows—promises a radically different, continuous operational load.

The Alert Backlog Bottleneck

The implications for network observability and security are severe. Research previously released by Cisco indicates that 95 percent of enterprises find their existing IT operations tools cannot keep pace with current demands. If an organisation were to rely on manual intervention, it would require dozens of IT specialists merely to clear the daily network alert backlogs generated by basic automated systems.

To safely harness this next generation of enterprise AI, organisations must pivot from reactive monitoring to adaptive security. In response to this impending bottleneck, Cisco unveiled a suite of new tools on 10 November 2026 aimed at securing the “agentic enterprise.” The newly announced framework includes Frontier AI Red Teaming—an adaptive security testing mechanism integrated with Anthropic’s Claude—and enhanced agent observability metrics powered by Splunk.

These tools are designed to provide security and operations teams with a unified feedback loop, allowing them to proactively identify risks, measure trust, and validate the genuine operational value of AI agents before they trigger system-wide performance degradations.

Relevance for Aotearoa and the Asia-Pacific

For businesses in Aotearoa New Zealand and Australia, the message from the Melbourne summit represents a critical juncture in AI stewardship. The regional rush to deploy AI agents as a productivity multiplier is well underway, but doing so on legacy networks introduces profound operational risks. If an enterprise network cannot securely manage the continuous, high-volume traffic of automated software agents, the promised efficiency gains will instead materialise as catastrophic outages and security blind spots.

The pivot from advisory AI to autonomous agents is no longer theoretical, with more than half of enterprises already trialling agentic systems in production environments. As these tools rapidly scale across the Asia-Pacific, robust infrastructure and stringent observability will dictate which organisations successfully automate their operations, and which are left overwhelmed by their own digital workforce.

Sources


AI disclosure: Automated systems assisted with research and drafting. Gemini independently peer-reviewed this article before publication. Sources are linked for verification.