Cisco: Agentic AI is going mainstream

Artificial intelligence (AI) has been the focus of discussions and product launches at the Cisco Live event as the vendor spoke of this being the year for agentic and physical offerings.

Last year was seen by many as the start of shift from chatbots to more capable agents entering the workforce, with Cisco expecting that shift to accelerate. In addition, physical AI robots are starting to emerge in specific use cases in some vertical markets.

Jeff Schultz, senior vice-president of portfolio strategy for Cisco’s product organisation, said this was going to be a year of change, with increased agentic AI adoption: “We are firmly moving from the era of chat bots, where humans engage with AI to ask questions and get information, to a world of agentic AI, where workflows are now getting automated and agents can perform tasks on behalf of human workers.

“At the same time, we’re seeing the emergence of physical AI, where robotics and other mechanical forms of AI are becoming prevalent in many use cases in the general workplace, as well as in lots of specialised industries. This is fundamentally changing the way all our customers must think about their infrastructure. We are massively optimistic about this change and think it’s going to create a huge amount of productivity globally,” he added.

Those shifts spell out an opportunity for the vendor’s partners, but the vendor warned there were constraints that were currently holding back AI adoption, with the event outlining the solutions it was rolling out to overcome those hurdles. The vendor indicated that 98% of its business across Europe is going via partners, underlining the importance partners play in delivering its technology to customers. Schultz said there were infrastructure constraints, which were exacerbated by the agents’ abilities to work for long periods of time, placing higher demands on compute power.

“We also have a trust deficit, and so we know our customers are pushing very hard to adopt AI, but at the same time, they know there are issues that they must think about. They need to think about safety and security,” he said.

“Are the models that are behind the AI agents behaving properly? Are they susceptible to attack by bad actors? And as we see agents enter the workforce, things like agent identity and trust and behaviour become issues that need to be very carefully governed by our customers.”

In response to these considerations, Schultz said Cisco had reimagined its portfolio to support AI-ready datacentres and to provide digital resilience and ensure there was secure global connectivity.

“One of the things that we are really focused on is building what we call a ‘platform advantage’ across all these products,” he said. “So, for our customers, as you start to use more of our Cisco products, you should see a compounding benefit of doing that. And that is something we’ve been very heavily focused on.”

Kevin Wollenweber. senior vice-president and general manager of datacentre and internet infrastructure, said that it was still early days building out AI, but the number of agent-driven workloads was rising.

“We are at the early stages of AI build outs, and AI build outs across a large set of our customer bases,” he said. “The bulk of the investment has been around building LLMs, training LLMs and really within the large hyperscalers, the deal clouds and these model builders themselves.

“What we’re starting to see are these agentic workloads taking hold inside of our datacentres, and people realising how to harness the models that are being built and the technologies for real, useful outcomes. And so, in 2026, you’re going to see a lot more real-world examples of how people are using agents,” he added.

Wollenweber said that the opportunities are around both infrastructure and security being able to “connect and protect” customers, and this would in turn accelerate the adoption of AI.