Cisco agentic AI efforts target networking, cybersecurity unification
Cisco agentic AI efforts target networking, cybersecurity unification
Publish Date: 2026-06-02 09:13:00
Source Domain: www.sdxcentral.com
Using an unordered list, summarize the following article with between 4 and 8 key points. Cisco’s agentic AI focus gained a new cloud-based centralized control point designed to help customers manage and secure their infrastructure estates across the networking giant’s growing set of products and services, a need that Cisco executive Jeff Schultz said is important as “we do not believe that security and networking in the future should be separate practices.”“These things need to be fused to be able to handle security at the rate at which agent AI is operating,” Schultz said during a press briefing ahead of its Live event.Cisco’s solution toward this goal is through the aptly named Cisco Cloud Control, which is positioned as a single login and management plane for products across networking, security, compute, observability, and collaboration. This unification is based on top of Cisco’s Data Fabric, which itself is run by the Splunk Platform acting as its unified data layer.DJ Sampath, SVP and GM for Cisco’s AI Software and Platform group, explained during a press briefing that this unification presents “your entire estate … together in one place, all of the assets easy to see, easy to understand, and a single source of truth.”“We’ve also included the topological view of this, so you know exactly how each one of these things are connected, and the context is shared across teams, and the operations are starting to align around your outcomes, as opposed to just trying to copy paste things from one place to the other,” Sampath added. “And … with trusted agents you can seamlessly move from all the visibility that you get to action, and that’s really where the breakthrough happens.”The platform is set to support more than 40 “ecosystem tooling connectors” and “customizable with the ability to create custom applications and agents using natural language directly within the platform.”This customization and collaboration will take place in the Cloud Control Studio, which includes two components. The first is Agent Builder, where AI agents for Cloud Control are created and can be connected to third-party tools using native connectors or the model context protocol (MCP) from the likes of Amazon Web Services (AWS), Google, Microsoft, and ServiceNow. The second is the App Builder, where Cloud Control custom apps and workflows are created using natural-language prompts embedded with agentic coding assistants, notably OpenAI Codex.Both of these components are set for availability later this year.Applications built using Cloud Control Studio will sit alongside others contributed by dozens of ecosystem partners within the Cloud Control Marketplace.Cloud Control will also house AI Canvas, which Cisco unveiled last year, and is a space developers can interact with AI agents on processes and solutions. Cisco notes developers will be able to see the “same live evidence that agents do,” with context persisting “across shifts and escalations, so nothing is lost and nothing is repeated.”Agentic AI for the networkCisco Cloud Control will also house an Agentic Actions for networking suite that acts as a root cause analysis and closed-loop remediation center for a deployed network. It includes a handful of agentic AI-fueled services like experience metrics, deep reasoning, and a digital twin that can parse out and solve networking issues.Anurag Dhingra, SVP and GM for Cisco’s Enterprise Connectivity and Collaboration Group, described it as “the mission control for how network operators can supervise the work of the agents in the system.”“So every time an agent raises an alarm, detects an incident, has a recommendation for optimizing the network, it all shows up here, and a network operator can go in here and inspect that work, look at the reasoning and the evidence that the agent used, all the telemetry that the agent used to come up with that incident or recommendation,” Dhingra explained, adding that “the human operator can also then look at what is being suggested and the AI agents come up with a confidence score and a risk score so that the human can assess the confidence in that remediation or optimization change, as well as the potential risk of making that change, and then if they feel comfortable, they can apply the change, or they can direct the agent to run some more tests.”There is also a new Cisco Multicloud Fabric that can help unchoke network traffic tied to different cloud environments. It connects deployed infrastructure sites and clouds through virtual points of presence (PoPs) embedded in hyperscale architectures.Cisco also has new devices, including the C9550 Core and expanded C9350 smart switches; 8000-series routers; a “premium” outdoor Wi-Fi 7 access point (AP); and its IR1000 rugged compact router.More security toward a quantum futureCisco’s security work also gains new backing with its Resilient Infrastructure Services that uses AI-derived insights and zero-trust principles from Cisco’s IQ to help organizations counter frontier-model threats using a process of assessing exposure, modernizing infrastructure, and building defense resilience. Cisco is also expanding its Live Protect across its platform, initially for its Nexus line.Tom Gillis, SVP and GM of infrastructure and security at Cisco, explained that this work opens the ability to initiate specific updates without disrupting operations, as it “allows us to identify vulnerabilities in that infrastructure and apply a compensated control with pinpoint accuracy on a running switch without rebooting that switch.”“Let me say that again,” Gillis added for emphasis. “We can apply compensated controls to a switch without rebooting the switch, or router, or Cisco infrastructure.”Gillis later added that this precision can be “almost pinpoint accurate controls that are applied into the network operating system to shield against the known vulnerability, and they can be as precise as this particular software process cannot access this particular file.”The vendor has also bolstered its DefenseClaw offering to now be “enterprise-grade” and ready for inclusion in Cisco Secure Client, where it can instill security and governance to local agents like OpenClaw, OpenAI Codex, and Claude Code. Cisco’s Zero Trust Access can also now restrict agents to predefined actions, and the vendor’s Agentic SOC can now use agents to analyze event alerts and anomalies.Cisco is also angling this security work toward a quantum future with updates to its Nexus, Firewall, SD-WAN, Webex Suite, and the publication of its “quantum-resilience” product roadmap targeted at the vendor’s core networking, firewall, VPN, observability, and collaboration tools.Cisco’s quantum preparedness includes a new assessment program through its IQ platform that can note areas potentially susceptible to quantum-based “harvest now, decrypt later” cybersecurity attacks.Sampath further explained that Cisco’s legacy Security Cloud Control platform now “evolves to security in Cloud Control,” which Dhingra took as a jumping-off point to highlight the broader focus of the new Cloud Control launch.“The one way to think about this is we have domain-specific agents that deeply understand that domain, and then they are purpose-built for that domain, but when you bring everything together in Cloud Control you have cross-domain, cross-agent operations possible now, and that is where these things, all the agents that we’re talking about, they actually collaborate with each other under … Cisco Cloud Control … and so they collectively deliver an end-to-end experience that makes it easy to operate your environment,” Dhingra said.
