itstargetnews
Presented by Cisco
AI agents are disrupting traditional IT operating models, increasing complexity, data silos, and workflow fragmentation. DJ Sampath, SVP of AI Software and Platform at Cisco, believes AgenticOps is the solution: a new operational paradigm where humans and AI work together in real time to create efficiency, improve security, and enable new technology applications.
In a recent conversation with VentureBeat, Sampath outlined why today’s enterprise IT management is fundamentally broken and what makes AgenticOps not only useful, but necessary for IT operations going forward.
The break point of traditional IT operations
The main problem plaguing enterprise IT today is fragmentation, says Sampath.
"Many times within these businesses, data sits in many different silos," he explained. "For an operator to come in and start troubleshooting something, they have to go through a lot of different dashboards, a lot of different products, and that results in more and more time spent trying to figure out where they are before they get to the real cause of an issue."
This challenge is about to intensify. As AI agents become ubiquitous within businesses, the complexity will increase exponentially.
"Every single person has at least 10 or more agents working for them to do different kinds of things," Sampath said. "This problem becomes ten times, if not a hundred times worse when you start to think about what is really going on with the inclusion of agents."
Three core principles of AgenticOps
To address these challenges, Cisco developed AgenticOps capabilities around three fundamental design principles that Sampath believes must hold true for this new operating model to succeed.
First, unified access to data across silos. The platform must integrate different data sources: network data, security data, application data, and infrastructure data.
"Bringing all those things together becomes more important so that the agents you deploy to work for you can seamlessly connect the dots across the board," Sampath said.
Second, multiplayer-first design. AgenticOps must be fundamentally collaborative from the ground up, enabling IT operations, security operations, network operations teams – and agents – to work together seamlessly.
"If you bring the IT ops person, the SecOps person, the NetOps person together, you can solve and debug issues faster than if you work in silos and copy paste things back and forth," he explained. "These are people and agents working together in a synchronous environment."
Third, purpose-built AI models. While general-purpose AI models excel at a wide range of tasks, specialized operations require models trained for specific domains.
"When you start going to the specialists, it becomes very important for these models to understand more specific things like network configuration or thread models that you care about and need to be able to reason about that," he said.
How Cisco operates with AgenticOps across the enterprise stack
Cisco’s approach combines telemetry, intelligence, and collaboration on a unified platform. Cisco AI Canvas is an operations workspace that replaces multiple dashboards with a generative UI and a unified collaboration experience. Within AI Canvas, operators can use natural language to delegate actions to agents – pulling telemetry, correcting signals, testing hypotheses, and implementing changes – while maintaining human-in-the-loop control.
Reasoning capabilities from Cisco’s Deep Network Model, trained on over 40 years of operational data including CCIE expertise, production telemetry, Cisco’s Technical Assistance Center (TAC), and Customer Experience (CX) insights. This purpose-built model delivers domain-specific intelligence that general-purpose models cannot match.
Cisco’s platform spans campus, branch, cloud, and edge environments, allowing agents to use telemetry across the entire ecosystem at machine speed, including Meraki, ThousandEyes, and Splunk. With MCP servers implemented in Cisco products, agents get standardized access to tools and data without traditional integration work.
How reporting data fragmentation undermines IT troubleshooting
The traditional approach to IT troubleshooting involves raising tickets and sharing information across multiple systems.
"People are taking screenshots. Sometimes it’s on Post-it notes," Sampath said. "All of this information remains in completely different channels so it can be really difficult for someone to start collecting them together."
Cisco AI Canvas addresses this by giving teams a shared, real-time workspace for the work at hand – so that context isn’t scattered across chats, tickets and screen shares. Teams can collaborate live, scale quickly, and contribute context (such as screenshots and notes) along with agent-generated charts and graphs. But the real power emerges when AI agents participate in these collaboration sessions.
"Machines are constantly learning from these human-machine interactions," Sampath explained. "If you see the same problem happen again, you will be faster to respond because the machines will help you."
This creates a virtuous cycle of continuous improvement, where the agent asks if you want to continue using the same method as last time, for example, and you can delegate more work to the agent. And time spent debugging is compressed as the system learns and accelerates future responses.
Security as an AI accelerator
Historically security has been considered a barrier to adoption and even innovation. But with the right guardrails, organizations can confidently deploy AI at scale, and even accelerate it.
Employees are already experiencing the productivity gains of tools like ChatGPT and want the same capabilities within their business environments. If organizations can detect personally identifiable information, prevent injection attacks, and maintain proper data management, they can unlock and unleash the adoption of AI within the enterprise in a different way.
The identity layer is required for cross-domain AgenticOps
Cross-domain data access presents one of the most complex challenges in implementing AgenticOps. Cisco’s strategic acquisitions, particularly Splunk, have positioned the company to address this, integrating data from traditionally disconnected systems. But integrating data is only half the battle, as who has access to what data is becoming increasingly important.
Cisco is evolving its Duo platform beyond multi-factor authentication to serve as a comprehensive identity provider, with strong identity and access management baked into the platform from the start, not bolted on as an afterthought.
"We invest in identity as a core pillar of how these agents are able to retrieve data from different data sources with the right permissions in mind,” Sampath explained. “Should this agent have access to this type of data? Do you need to correlate these types of data to solve a problem?"
People in the loop, but at a higher level
As AI agents become more autonomous, the role of humans will grow rather than disappear.
"We always have people in the loop," Sampath said. "What you will see is the complexity of the work being done as it becomes more involved."
Take coding as an example, which can now be fully agentized. The human role has shifted from manual coding, or even tab completion, to asking an agent to generate code wholesale, and then verifying that it meets the requirements before merging it into the codebase. This pattern will repeat itself in IT operations, with people focusing on higher-level decision-making while agents are in charge of execution. Importantly, rollback capabilities ensure that even autonomous actions can be reversed if necessary.
Why waiting for AI to ‘settle’ is the wrong move
For CIOs and CTOs, the message is clear: don’t wait.
"Many people are in this pattern of waiting and watching," Sampath said. "They are waiting for the AI to settle before they make some of their decisions. And I think that’s the wrong way to think about it. Partnering with the right groups of people, with the right set of vendors, will help you go faster, as opposed to just trying to stay on the fence, trying to figure out what’s right and what’s wrong."
Sponsored articles are content produced by a company that pays to post or has a business relationship with VentureBeat, and is always clearly labeled. For more information, contact [email protected].
