At the Cisco Live EMEA event going on in Amsterdam this week, Cisco unveiled a range of updates across its networking and security portfolio that are aimed at helping customers tap agentic AI technologies to more effectively run their enterprise infrastructure.
“Agentic AI orchestrates workflows, moves data, communicates with other agents, and makes decisions autonomously. This represents a fundamental shift in how technology operates within organizations,” wrote Jeetu Patel, Cisco’s president and chief product officer, in a blog post about the news. Agentic AI also changes what infrastructure must deliver, Patel stated. “When agents act independently and communicate machine-to-machine at scale, the network becomes more than connectivity. It becomes the foundation for trust, performance, and competitive advantage. We believe that the organizations that will thrive in this era will be those that treat infrastructure not as a cost to manage, but as a strategic asset to leverage.”
At a high level, Cisco’s AgenticOps is an agent-driven IT operating model that was first launched last year. It uses system‑wide awareness and cross‑domain telemetry gleaned from Cisco’s portfolio of networking and security products, including Security Cloud Control, Nexus One, Splunk and ThousandEyes, for autonomous action. New AgenticOps capabilities are spread across a number of product lines, including Cisco Defense and Cisco’s SASE portfolio.
“The AI age punishes complexity. When organizations cobble together separate tools for networking, security, and operations, they inherit friction, cost, and risk at exactly the moment they need speed and clarity,” Patel wrote. “Cisco is taking a different approach. We’re delivering a unified platform — networking, security, observability, and sovereignty working together as one system — so customers can scale AI without the usual trade-offs.”
Updates to Cisco’s AI Defense
Cisco unveiled what it says is the biggest expansion to AI Defense since it launched the security platform last year. Updates include:
- AI traffic optimization: Detects AI traffic patterns and applies techniques (like packet duplication) to keep agentic interactions reliable and low latency during spikes.
- AI Bill of Materials (BOM): Brings a centralized inventory and visibility into AI assets (models, agents, MCP servers, third-party tools) to strengthen supply-chain governance.
- MCP support: Discovers and inventories Model Context Protocol servers and public or private registries, helping assess risk across tool registries. These capabilities bring transparency and centralized governance for models, datasets, tools, third-party dependencies, and other critical AI resources.
- Advanced Algorithmic Red Teaming: Expanded adaptive testing — including multi-turn interaction scenarios and multiple languages — to uncover vulnerabilities in models, agents, and workflows. Cisco says it has a completely redesigned AI Validation feature that enables single and adaptive multi-turn testing for models and agents with broad multi-lingual support. It provides clear insights and immediate security recommendations based on findings from AI Validation assessments. It also weaves prominent AI security frameworks and standards throughout testing, including those from NIST, MITRE, and OWASP, as well as Cisco’s AI Security and Safety Framework.
- Intent-aware inspection: Combines rapid techniques with cloud analysis to understand why an agent is asking for something — catching malicious or risky intent that legacy tools miss.
Cisco SASE upgrades
For Cisco SASE, the company is adding:
- The ability for agents to detect and classify AI/agentic traffic patterns and apply optimization techniques so autonomous agents get predictable, low-latency performance even during traffic surges.
- Features such as in-path discovery, visibility, and logging for MCP communications. The idea is to help customers govern and enforce agentic actions and policies about how agents communicate with tools and data services.
- With SD-WAN intelligence powered by Network-Based Application Recognition(NBAR), Cisco SASE automatically identifies and classifies AI traffic across cloud, edge, and hybrid environments. It steers each type where it needs to go, keeps latency low for sensitive workloads, and it lets teams apply ready-to-use AI policy templates, so controls stay consistent without manual tuning, according to Cisco.
New AgenticOps capabilities in networking
Other core AgenticOPs enhancements include:
- Autonomous troubleshooting: The ability to deploy end-to-end agentic investigations across campus, branch, and industrial networks to handle connectivity and experience issues. Applies reasoning from telemetry to root cause, validating multiple hypotheses simultaneously and executing deterministic remediations with CCIE-grade precision, Cisco stated.
- Data center: For its Nexus One data center management platform, Cisco added early detection and intelligent event correlation with AgenticOps for data center networks. The package provides prescriptive recommendations to optimize performance. By providing actionable insights across traditional and AI workloads, the package drives proactive operations and significantly improves business outcomes. Controlled availability in June 2026, according to Cisco.
- Firewalls: Cisco said it added proactive analysis of firewall traffic, including the applications accessed and mode of access. This capability lets agentic policies identify and recommend more robust zero trust controls for sensitive applications. These capabilities can then propose actions tailored to each customer’s environment. Meanwhile, new AgenticOps troubleshooting and optimization capabilities detect issues such as elephant flows impacting firewall performance, perform full-context analysis, and propose remediation options.
- AI agent monitoring in Splunk Observability Cloud visualizes agent workflows and will soon integrate with Cisco AI Defense to mitigate risks that inhibit trust in AI models, such as bias, hallucinations, data leakage, and prompt injection. The feature is available now.
Analyst response
“We see that modern IT environments have become more fragmented and volatile, especially in the AI era, putting immense pressure on teams to maintain security and uptime,” wrote Ron Westfall, vice president and practice leader for infrastructure and networking at HyperFRAME Research, in a research note about Cisco’s news. “Cisco’s AgenticOps addresses these burgeoning requirements by providing a structured foundation that handles complex operational tasks. The model is designed to deliver intelligent, automated execution without sacrificing the governance, accuracy, or reliability that enterprise organizations depend on.”
“By embedding AgenticOps across the entire infrastructure, from air-gapped industrial sites to complex cloud environments, Cisco meets the urgent demand for intelligent execution that doesn’t sacrifice reliability. Enterprises need to scale operations without exponentially increasing human headcount or operational risk,” Westfall wrote. Customers are no longer looking for mere alerts; they require the closed-loop automation and prescriptive recommendations found in Nexus One and Security Cloud Control to manage the sheer volume of data generated in the AI era. This alignment with customer needs for faster troubleshooting and continuous compliance ensures that Cisco remains a vital partner in maintaining business continuity amidst growing complexity.”