In my first few months at Cisco, I’ve had a front-row seat to the business’s key areas of focus. Every large tech business claims to be changing the way the world does things, but the scale of transformative solutions I’m seeing at Cisco is different. This year’s Cisco Live event was confirmation –
The conference felt unique, not just because of the announcements, but because of what they signalled. In this case, Agentic AI, and the way its capabilities are reshaping the underlying infrastructure. Cisco’s focus is clearly on building an infrastructure capable of supporting a world empowered by autonomy.
The first bit of proof was the announcement of a groundbreaking platform that redefines network management. While we’re all familiar with the last two decades of user interfaces and how they were built around visual constructs like buttons, icons, and dashboards – what was revealed at Cisco Live is the rise of something fundamentally different – the conversational interface. Through the maturity of generative AI and natural language processing, instead of clicking your way through software, you talk to it instead.
The announcement of AI Canvas proved that managing complex networks is about to become radically more intuitive. AI Canvas is a unified, conversational interface that allows operators to interact with their networks using natural language. By enabling seamless collaboration between network operations (NetOps), security operations (SecOps), and development operations (DevOps), AI Canvas aligns with the trend of conversational UIs. It allows operators to ask questions like ‘How is my network performing?’ – and get clear, context-rich answers. That answer is backed by a Deep Network Model trained on over 30 years of Cisco’s networking expertise, courseware and CCIE materials – and gives you 20% more precision than most large language models.
Another theme that couldn’t be overlooked was how the infrastructure story is evolving to meet the real demands of AI. If you want to support AI at scale, infrastructure simply cannot be an afterthought. With Nexus Dashboard now serving as a unified console, there are plans to integrate NX-OS and ACI. The key announcement here was the introduction of 400G bidirectional (BiDi) optics, which allow data centres to transition to 400G networks without rewiring existing duplex multimode fibre (CRN). This is incredibly useful if you’re trying to scale AI fabrics that depend on speed, density, and low latency – because you can do it without overhauling everything underneath.
Cisco’s collaboration with NVIDIA is also accelerating, with deeper integration between our G200-based switches and NVIDIA’s Spectrum-X. On the compute side, Cisco has added support for the RTX Pro 6000 GPU in our UCS C845A M8 servers – which means twice the memory of the L40S. It’s becoming a key enabler for enterprise inference workloads.
The deepening integration with Splunk stood out as well. By bringing ThousandEyes, observability, and assurance together, users are alerted to a problem, its cause and what they should do next.
Of course, AI scale comes with AI risk, so it was no surprise to see that security was so core to the event’s announcements. From Cisco’s Secure AI Factory to the integration of the Deep Networking AI model across the portfolio, there is a significant focus on defending against AI threats with AI-native tools.
Cisco Live 2025 showed both clear direction and significant momentum. AI is driving the next wave of infrastructure design, and the way we interact with applications is shifting from dashboards and scripts to conversations, intent, and intelligent action. But while these intelligent systems will inevitably work on our behalf, none of this can happen without the right foundation.
Just a few early takeaways from the event. I’ll continue sharing more as time, and deadlines, permit.
#CiscoAI #AIReady #AgenticAI #SecuredAI #AIInfrastructure #CiscoLive2025
