When Cisco says the “AI vs. human” debate is settled, it isn’t just making an observation about the industry. It’s setting up a sale.
The company recently published a guide outlining the top five skills engineers need for the age of AgenticOps, a term Cisco has coined for its vision of AI-driven IT operations where autonomous agents reason through problems and resolve them at machine speed rather than simply surfacing alerts for humans to act on. The framing is timely. Gartner predicts that by 2029, 70% of enterprises will deploy agentic AI as part of IT infrastructure operations, up from less than 5% in 2025. Agentic AI in IT isn’t hype; it’s a genuine operational shift. But the skills Cisco recommends map, with notable consistency, to Cisco’s own training portfolio, and that’s worth understanding before you build your career plan around them.
The five skills, network programmability and APIs, Python automation, wireless telemetry, secure networking policy, and AI literacy, are each real and valuable. The first two in particular represent a generational shift in what network engineering actually means. The article’s observation that SSHing into devices and manually adjusting configs is giving way to API-driven workflows is accurate and not overstated. As DJ Sampath, SVP of the AI software and platform group at Cisco, notes, AI agents with advanced models can help engineers configure networks, understand logs, and address issue root causes more efficiently. The human-in-the-loop engineer who can validate what agents are doing, catch a false positive, and interpret telemetry trends is a genuinely valuable role. That part of the argument stands regardless of vendor.
Where the article gets slippery is in its training resources. Each skill section ends with a set of recommended learning paths. Follow them and you’ll emerge competent in Cisco Catalyst Center, Cisco Meraki, Cisco Splunk, Cisco AI Canvas, and the newly rebranded CCNP Automation track, which was the DevNet certification until February 2026. Starting February 3, 2026, Cisco officially rebranded the DevNet certification track as CCNA, CCNP, and CCIE Automation, with anyone holding an active DevNet certification automatically recognised under the new naming. The timing of this guide isn’t coincidental. Cisco needs engineers to understand why the new Automation track matters, and an AgenticOps urgency argument is a more compelling frame than “we renamed some certifications.”
That doesn’t make the advice wrong. Roles requiring network automation skills command 15 to 25% higher average salaries than traditional CCNP positions in the US market. The business case for upskilling in automation is real. But there’s a difference between “automation skills are essential” and “Cisco’s automation skills path is the automation skills path,” and this article blurs that line. Network engineers working in cloud-native or multi-vendor environments would find the recommended resources partially applicable at best. The broader agentic operations conversation encompasses concerns like agent governance frameworks, Model Context Protocol adoption, multi-vendor orchestration, and handling hallucinations in production systems, none of which appear in Cisco’s five skills.
The piece also presents wireless telemetry and experience analytics as a top-five career skill for the AgenticOps era. It’s a legitimate area of expertise, and Cisco’s wireless portfolio is strong. But as a universal career priority ranking alongside Python and API fundamentals, it reads more like a product placement for Cisco’s wireless track than a dispassionate assessment of where engineering value is actually concentrating in 2026.
The broader point is that this kind of vendor-published career guidance has a structural bias that’s easy to miss when the underlying skills are genuine. Microsoft publishes similar content pointing to Azure certifications. AWS publishes it pointing to cloud practitioner credentials. The pattern, visible in how AWS and others approach skills programmes, is that the most credible vendor skilling initiatives are honest about where their recommendations end and the wider market begins. Cisco’s guide is less clear on that boundary.
For a South African IT professional deciding where to invest in upskilling, the honest read is this: the first two skills, API programmability and Python automation, are close to non-negotiable regardless of what infrastructure your environment runs on. AI literacy, in the sense of being the human who validates and contextualises what agents are doing, is similarly vendor-agnostic and genuinely important. The wireless telemetry skill and the Cisco-specific security policy framing are more situational. If you’re in a Cisco shop, the certification path is a coherent one. If you’re not, the skills are still worth having; just pursue them through resources that aren’t built around a single vendor’s product suite.
AgenticOps as an operating model is coming. Cisco’s guide to navigating it is useful. Just read it knowing who wrote it.