Cisco Live Melbourne: AI agents will break today’s enterprise network, and Cisco wants to rebuild it
Network World · · g2158177
To date, AI conversations have centered on GPUs, AI factories, and data centers. At Cisco Live APJC in Melbourne this week, Cisco shifted the focus to the campus, the branch, and the network that connects everything. The opening keynote set a realistic tone for where the industry is. Agents don’t behave like people, and networks built for people won’t hold up when thousands of agents work around the clock. Ben Dawson , Cisco’s president for Asia Pacific, Japan, and Greater China, highlighted the numbers network teams need to understand. He said agent token consumption surpassed human consumption for the first time in February, and agents now consume tokens at roughly seven times the rate of humans. “Each agent is generating vastly more traffic and data than a human performing the same task,” Dawson said. That matters to network professionals. The chatbot era was bursty and mostly north-south: A prompt went to the cloud, and an answer came back. Jeff Schultz , Cisco’s SVP of portfolio strategy and marketing, explained how agents change that. They still send heavy north-south traffic to reach models and context, but they also generate large volumes of east-west traffic as agents talk to other agents across desktops, campuses, branches, and data centers. “All of this again comes back to the idea of needing to fundamentally reimagine the architecture for networking in the campus and branch,” Schultz said. Wired isn’t dead, and the network isn’t ready One of the more interesting points came when Dawson discussed the wired network’s renaissance. “AI is going to put new demands on wired infrastructure, making it even more critical,” he said. I agree, as many organizations have treated the access layer as a commodity for a decade, and agents will test that decision. The readiness gap is significant. Oliver Tuszik , Cisco’s chief sales officer, cited IDC data showing that only 9% of organizations in the region believe their networks are fully ready for new AI and digital initiatives. He also cited research indicating that 85% of organizations globally are experimenting with agents, but only 5% have deployed them to production. In my conversations with CIOs, the network is often what keeps them stuck in that gap. Cisco’s solutions span its entire stack. In the data center, Schultz highlighted 102.4 Tbps systems based on the Silicon One G300 for scale-out, with air- and liquid-cooled options and a choice of NX-OS or SONiC. For scale-across, the deep-buffered Silicon One P200 connects data centers across metro, long-haul, and even transcontinental distances. On the campus side, the news that matters most to network operators concerns telemetry. Cisco is now integrating ThousandEyes into its Wi-Fi 6E and Wi-Fi 7 access points at no extra cost for customers with eligible Networking Essentials and Advantage licenses. Schultz admitted that ThousandEyes has been “a little bit hard and difficult to license and deploy” and is making it a default fix. Agentic operations depend on the telemetry beneath them, and putting assurance data on every new access point gives Cisco’s AgenticOps agents something real to work with. Operations has to move at agent speed Cisco Cloud Control is Cisco’s unified operations layer and, as Schultz put it, the “single front door” to its portfolio. It combines cross-domain telemetry, purpose-built models, and both Cisco and third-party agents. Two details stood out to network engineers. First, Schultz said no new Cisco product will ship unless it is integrated with Cloud Control. Cisco has needed that discipline for years. Historically, Cisco has built products in silos and left integration to its customers. Under Chief Product Officer Jeetu Patel , Cisco has committed to absorbing complexity and simplifying things for customers, and Cloud Control is a strong example. Second, availability remains limited. Cloud Control is currently available only for US-based infrastructure. Australia, the EU, and the UK are scheduled for February 2027, with Japan, India, and Canada following in May. Closing the trust loop For network engineers, another key point in the keynote was Schultz’s assertion that “the line between observing an agent and securing an agent is beginning to blur.” He cited an agent that suddenly deleted 10,000 files. Was it compromised, or did it decide on its own to free up disk space? Neither a security tool nor a monitoring tool can answer that question alone. In previous conversations with Patel, he has described this as a “continuous trust loop” and has told me that AI can’t scale without security, governance, and observability. The loop has three parts. The first is testing. Cisco AI Defense is adding Anthropic’s Claude Mythos to its adaptive red-teaming engine. Cisco has set firm limits: Mythos will test only targets the customer owns and has registered; each engagement is cryptographically tied to a named operator ; sessions are time-bound and can be stopped; and customers can’t prompt the model directly. It is coming soon to select enterprises, with broader availability for qualified customers later. The second is cost and quality. Splunk Agent Observability’s Tokenomics feature pairs each agent’s spending with evaluations of its work, using Amazon Bedrock and CloudWatch data. It enters private preview in December, with general availability on AWS Marketplace in early 2027. Splunk’s new Eval-G1 small language models run evaluations and guardrails at what Cisco says is up to 98% lower cost than a commercial LLM , making it practical to check all agent traffic rather than a sample. Schultz argued that tokenomics is a security issue too: In agent-driven attacks, the side that spends tokens more efficiently wins. The third part matters most to network teams. Cisco IQ vulnerability, lifecycle, and configuration data now flow into the Splunk Platform via Cisco Data Fabric, enabling a SOC analyst or an AI agent to match a live detection in Splunk Enterprise Security to the exposed switch or router . New MCP and ServiceNow connectors convert those findings into asset-linked problem records. This closes a gap between network state and security workflows that has slowed incident response for years. Speed is the reason. Jacqui Guichelaar , who leads Cisco’s customer experience organization in APJC, said that the share of exploits occurring on or before disclosure day has risen from 22% in 2019 to 78% this year. Cisco also expanded Live Protect, which pushes “shields” that block exploitation of critical vulnerabilities until a patch is applied. What customers said The customer stories made the case more effectively than the product slides. John Kelvie , chief digital and information officer at Asahi Beverages, described a greenfield Melbourne head office for about 1,000 employees, built on Cisco’s SD-Access architecture so the network could be segmented and secured from the start. “We couldn’t build tomorrow’s AI organization and enterprise on yesterday’s technology,” he said, calling the network “the nervous system” of the company. Asahi has since brought OT on the production floor onto a single IP network. The goal was a consistent experience, whether an employee is at the head office, a brewery, a distribution center, or in Malaysia or New Zealand. As Kelvie put it, the technology “just had to work.” Asahi already uses AI for demand sensing, forecast optimization, trade promotion, and back-office automation. Kelvie said the next phase is to shift from team-level experiments to “fewer, bigger initiatives” across the enterprise. His biggest obstacle isn’t people or technology but data. “AI is going to amplify your strengths, but it’s also going to amplify your inconsistencies,” he said. Asahi still runs separate processes in parts of its alcohol and non-alcohol businesses, creating different versions of the truth. Harmonizing those processes is a prerequisite for the fully automated distribution centers the company is building, which promise “one customer, one order, and one delivery.” Moving fast, he said, doesn’t have to mean adding complexity. Yessie Yosetya , CIO of Indonesian operator XLSmart, had a tougher starting point. After the merger that created the company, she inherited two networks with very different security postures, one scoring 3.2 on a NIST-based maturity scale and the other 1.6. The easy option was to compare the two legacy stacks tool by tool and keep whichever was better, 54 times over. Instead, she stepped back and redesigned around five areas: monitoring and detection, identity, network, applications, and an agentic SOC. “We are not choosing two yesterdays,” she said. The result is a reduction from 54 security tools to 14. Her experience with agents serves as a useful warning for teams that build on their own. XLSmart put eight agents into production for red, blue, and purple teaming, then found it couldn’t scale them without standardization and guidance. “We cannot beat machine speed with human speed,” she said. Her advice on choosing a partner was practical: watch how a vendor behaves when a problem arises. The one that focuses on finding a fix is the one that can execute. Final thoughts Cisco’s argument is that the full stack of silicon, systems, optics, operations, security, and observability performs better when it comes from a single vendor. Customers concerned about lock-in will be skeptical, and the open-ecosystem story still has to prove itself. But the thesis is correct. Agents will change traffic patterns, increase the number of non-human identities, and shrink response windows faster than most network teams expect. For IT leaders, the practical steps are clear. Revisit campus and branch capacity plans, keeping east-west agent traffic in mind. Treat telemetry as a design requirement, not an add-on. Bring the network, security, and observability teams together now, because no single team can decide alone whether an agent’s behavior is a bug or a breach. Recognize that only 9% of organizations believe their networks are ready for AI. Getting yours ready may matter more than which model you pick.