Cisco Live 2025: Networking in the Age of AI–A New Era of Intelligent Infrastructure

As Chief Technologist for Networking and Connectivity at Softcat, I’ve had the privilege of witnessing firsthand how Cisco is redefining the future of networking. At Cisco Live 2025 in San Diego, the message was clear: we’re entering the age of AI-native networking — and it’s not just about speed or scale anymore. It’s about intelligence, adaptability, and efficiency.

Here are the standout innovations that are reshaping how we design, operate, and secure networks in an AI-driven world.

Model Context Protocol: The Foundation of Agentic Networking

Let’s begin with the foundation of Cisco’s AI-native networking strategy, the Model Context Protocol (MCP). A groundbreaking new architectural layer that enables AI agents to interact bidirectionally with the network in a structured, secure, and context-aware manner. Think of it as a universal translator between AI models and network infrastructure. This protocol powers “agentic networking,” where AI agents can observe, reason, and act on networks in real time.

Use cases range from automated root cause analysis to dynamic policy enforcement. Imagine a university campus where an AI agent detects a spike in latency during a lecture and automatically adjusts QoS policies to prioritise video traffic. This is the kind of real-time intelligence that transforms networks into strategic assets.

AgenticOps: From Dashboards to Digital Co-Pilots

Cisco’s AgenticOps vision is a leap forward in network operations. In industry, we hear plenty about AI in networking, so what is the secret sauce that Cisco have applied to differentiate from the market? In short, accuracy. Cisco’s agentic AI is built on their Deep Network Model, an AI model trained on decades of telemetry from Cisco TAC, knowledge bases and is aiming to be the equivalent of a virtual CCIE.

AgenticOps enables proactive, predictive, and even autonomous operations. Instead of reacting to alerts, AI agents continuously monitor the network, learn from patterns, and take action. It enables IT Teams to speak to their networks, ask questions, instruct tasks and request bespoke insights in natural language. For example, “Why is this application slow in London?” and receive contextual, actionable insights. Integrated with ThousandEyes, Meraki, and Splunk, AgenticOps delivers end-to-end visibility and dramatically reduces mean time to resolution (MTTR). This capability is announced to be released in October 2025 and soon after, Cisco plan to expand integrations to cover Catalyst Centre and Intersight amongst other Cisco platform solutions.

According to Gartner, by 2026, 30% of enterprises will automate more than half of their network activities, up from under 10% in mid-2023. This shift is driven by the need for operational resilience and the ability to process vast telemetry data through intelligent automation. Cisco’s AgenticOps aligns with this trend by enabling proactive and predictive operations.

There are several agentic agents Cisco has announced:

• AI assistant is a conversational tool embedded into Meraki, TE & Webex that helps IT teams troubleshoot, automate tasks, and real-time insights.
• Adoption Agent enables you to quickly adopt new technologies by guiding you through setup, usage, and best practices in a simple, conversational way.
• TAC Assistant can detect a misconfiguration before it causes an outage and guide the admin through a fix.

These agents don’t just respond to tickets, they prevent them. You may be thinking, “what is the role of the human here?”. Well actually, the human is the key. Every network is different and to no one wants to completely let go of their keys to the kingdom. Human oversight is essential to configuring, prompting, reviewing and approving agentic AI actions.

AI Canvas: The Command Centre for AI-Driven Networking

The AI Canvas is Cisco’s unified interface for interacting with AI agents, visualising network health, and taking action. What sets it apart is its multi-player collaboration model where NetOps, DevOps, SecOps and even business stakeholders can co-author generative insights, dashboards and actionable recommendations in real-time.

With natural language prompts, users can ask, “Show me all devices with outdated firmware in our retail stores,” and receive a visual map with remediation options. This democratizes access to network intelligence and accelerates decision-making across IT and the wider organisation.

Security Embedded in the Network

Security is no longer a separate layer to the network and must be embedded into the network itself. This is enabled by Cisco in 3 new ways:

• ·Smart Switches – powered by Cisco’s Silicon One and leveraging built-in Data Processing Units (DPUs), you can now enable security controls at the switch layer powered by AI via eBPF.
• Live Protect – patching is always a challenge in networking, especially when downtime is not an option short term. Live Protect automatically blocks attacks and protects network devices that are unpatched by applying specific rules on the switch based of known vulnerabilities.
• Hypershield – brings macro and micro segmentation to switching. Starting in the DCN Nexus portfolio and coming into the Catalyst LAN switching and wireless portfolios, Hypershield enables you to segment in real-time with AI. Uniquely it enables you to actively test new policy and software updates on a live copy of the network to assure unexpected issues don’t occur.

Combined together, these technologies enable distributed network-driven security for hybrid environments.

One Network. One Dashboard

Perhaps one of the most anticipated announcements was the unification of Meraki and Catalyst. Over 12 years ago, Cisco and Meraki started their relationship and historically, Meraki was known for simplicity whereas Cisco was known for more complex, enterprise networking capabilities.

Over time, the clear differences between the solutions has become more and more blurred as they became more integrated and shared capabilities. For years, organisations had to choose between Meraki’s simplicity and Catalyst’s power.

Now, they can get both with unified hardware, management, licencing in a “pick how you want to deploy your hardware” model. Cloud managed, on-prem managed or a mix of both, you choose without compromising functionality, simplifying operations and accelerating time to value.

Why it matters

We’re at a pivotal moment. AI is no longer just a tool, it’s a digital co-pilot for networks that enables operational efficiency gains, proactive ambient NetOps and new security controls powered by intelligent networks. Cisco is delivering that network today, with programmable silicon, unified platforms, and agentic AI.

At Softcat, we’re excited to help our customers harness these capabilities, embrace innovation, adopt a new way of working and empower IT teams to enable their organisations to thrive.

The future of networking is here, and it’s AI-native.