Cisco shifts security strategy to ‘built-in resilience’

Cisco believes adding more security tools can ultimately make organisations less effective at detecting and responding to threats. The strategy, therefore, is increasingly centred on platformisation–bringing multiple capabilities together and reducing the number of consoles and tools security teams need to manage.

“You (now have) have multi cloud, hybrid cloud, SaaS (software-as-a-service) products, and AI adoption. All of this has created much complexity. Most people we talk to tell us don’t just give us more tools. Nearly 84% of the organisations say that managing more than 10-point solutions actually slows down their ability to respond to and detect (threats),” Agarwal said.

Built-In Resilience

A key element of this approach is shifting from “bolt-in security” to “built-in resilience”. Rather than treating security as a separate layer, Cisco is looking to embed it across the infrastructure where data, applications and workloads reside. The broader objective is not simply to prevent breaches, but to ensure businesses can continue operating when disruptions occur.

Cisco is also positioning a unified data fabric that brings together security, network, application and IT telemetry, providing teams with a common view of their environments.

“One of the things we have always believed is (that) a lot of complexity results from having a security stack which operates in a silo,” Agarwal said.

Enterprise IT architectures typically have a stack where data, applications, networks and workloads reside. This sits separately from the security stack, which includes firewalls and related software, he said.

Instead of routing traffic through a separate security stack, Cisco is working with clients to embed security directly into the infrastructure where data and applications reside. “So, from bolt-in security, we want to move organisations to built-in resilience,” Agarwal said.

AI agents can then use this context to carry out certain tasks, while humans continue to supervise them and retain control over critical decisions.

Securing Enterprise AI Deployments

The same philosophy applies to securing enterprise AI deployments. Cisco’s approach involves first establishing visibility into an organisation’s AI inventory—including models, applications and datasets—then identifying vulnerabilities through testing and red-teaming, and finally applying security, privacy and safety guardrails.

Continuous observability is central to this model. As AI systems evolve and their behaviour changes, security cannot be treated as a one-time exercise, Agarwal said.