·
6 minute read
Summary
On this episode of Catching Up with, the Collab Collective's Craig Durr speaks with Jonathan Chang, Head of Product for Webex Customer Experience at Cisco, live from Cisco Live 2026. Jonathan shares how Cisco's AI Agent 360 strategy helps enterprises deploy AI agents with greater security, accountability, and visibility through AI Agent Studio, AI Defense, and advanced observability capabilities. They explore custom AI guardrails, policy-driven governance, AI and human agent management, workforce engagement, and the role of connected intelligence in delivering trusted customer experiences while maintaining control over increasingly autonomous AI systems.
Their discussion covers:
- AI Agent 360: How Cisco's framework combines security, manageability, and observability to support trusted enterprise AI deployments
- Custom Guardrails for Enterprise AI: How organizations can transform policies, procedures, and compliance requirements into AI-enforced controls using AI Defense
- AI Agent Studio: A look at Cisco's centralized platform for building, configuring, and managing AI agents through Collaboration Control Hub
- Observability and Performance Monitoring: How enterprises can track usage, sentiment, quality, security, and business impact across AI-powered customer interactions
- Human and AI Workforce Management: Why Cisco views human agents and AI agents as equal contributors to customer experience and how Workforce Engagement Management supports both
- Connected Intelligence and Future Innovation: How integrations with Splunk and broader Cisco technologies are expanding visibility, governance, and trust in AI-driven customer experiences
Transcript
Craig Durr: Everyone, this is Craig Durr, Chief Analyst and Founder of the Collab Collective. We are here live at Cisco Live 2026 and we are here on the expo floor. Right now, we are in the future-proof workplace solutions, and I've got Jonathan Chang, Head of Product for Webex Customer Experience. How are you doing?
Jonathan Chang: I'm doing well, Craig.
Craig Durr: Great to see you. Hey, so I had a chance just a few minutes ago to talk to Vinod. This is after he came off of his main stage, and he walked me through the four releases, the four-strategy pillar, and everything around this. But I want to take a moment to deep dive on one of them in particular. We want to talk about the AI Agent 360 point of view, because that's a lot of things in there, right?
Jonathan Chang: Absolutely, yes.
Craig Durr: So, why don't you talk to me about how an end user or customer should think about what Agent 360 means?
Jonathan Chang: I'm sure Vinod talked about, as part of our four-pillar strategy, trusted and secured by design. In the world of AI—agentic, autonomous—if you don't have the right level of accountability, it becomes a problem, right? So AI Agent 360 is really meant to provide enterprises with a full suite of capabilities that include security, manageability, and observability.
Craig Durr: And that's a great one because this is one of the unique things that Cisco is gonna bring together, especially with the security element and the observability from the beginning.
Jonathan Chang: Yeah, we're a full-time security company, a full-time observability company, right? And now we can bring that into our AI Agent Studio to drive AI agents for our customer experiences.
Craig Durr: Well, talk to me about that. Let's go and look at some of these different areas here. Why don't you go ahead and walk me through this?
Jonathan Chang: This is the AI agent studio. You can access this through the Collaboration Control Hub, all your Cisco Cloud Control. And within the AI Agent Studio, here's a list of all the agents that you already have, let's say created. You can actually click into one of them, and you can set up and configure your AI agents. We have AI memory—you can bring that in your organization, your knowledge bases, but there's a security tab here where you can actually bring in custom guardrails. So this is our integration with AI Defense to build very specific custom guardrails.
Craig Durr: Got it?
Jonathan Chang: These guardrails are beyond standard guardrails. We do have a full set of pre-built guardrails here below. There’s security, prompt injection, jail breaks. Even on the privacy side, you can do anything from blocking PII data, and PCI data. Those are your standard guardrails, very common. What we've done now with the custom guardrails is actually taking an enterprise's specific policies. So if you think about a question like, “What is the annual mortgage rate for a 30-year fixed?” You're using an AI agent to think of that answer. But when you have questions like, “Should I refinance now?” Then, it really depends on what kind of company you are. If you're a financial advisor company versus a bank.
Craig Durr: If you can do that or not, right?
Jonathan Chang: Exactly
Craig Durr: And this is the same thing that you might have, like a standard operating procedure for humans, right? But what we're trying to do here is apply those to an agent acting on my behalf.
Jonathan Chang: Exactly, and I can let me show you, actually. So over here you can actually go ahead and create a custom guardrail, and you can actually give the guardrail a name and some direction. Do you want it to apply? We actually do bi-directional guardrails, so the prompts going into an LLM as well as the output coming out, and you can actually apply it one way or both.
Let's say we do both the prompts and responses, then you'll actually launch it to our AI Policy Studio, which is from AI Defense. You can actually describe the policy you want, so let's say you don't have a pre-trained AI, you can just use natural language, or you can go ahead and upload an existing policy that you may already have.
Craig Durr: And this might, like I said, be an SOP. This might just be a policy manual, or it could be a lot of different things that you're using to train humans.
Jonathan Chang: Exactly, and once you apply that, we will actually use AI, and it will go to the policy and pull out what are the specific block behaviors, as well as allowed behaviors, based on that policy. And each one of these is now a guardrail for your AI agent.
Craig Durr: And so we've just taken this, and we've really customized this to this use case, to my business, even for this use case, right?
Jonathan Chang: Customized down to specific departments, specific groups, all for specific agents—high level detail.
Craig Durr: That's fantastic. Now, all that is under Policy Studio. It's part of the AI Defense story, which is across Cisco, which is one of the powerful stories. Now, talk to me about the observability piece as well.
Jonathan Chang: Also, within our AI Agent Studio, you'll see Observability. This is actually a really exciting area for us because this is just the starting point. I’ll talk more about where we're going, but what we'll have is a comprehensive set of metrics and KPIs in here, everything from how much usage happened, both on voice or digital, all the way to different actions that were performed by our AI agent—quality, your security, business impact, and you can actually go take a look at customer experience sentiment. What does that look like? You can actually click in, and you get a holistic view of how that sentiment has been across all your AI agents. We also have this in a way where you can actually manage all your AI agents together, so this is actually the rolled up view. You can filter for individual agents, or even come down here and look at all these specific agents that are all working within this enterprise's company's customer experience. And you can actually go and click into each one down to specific views on what's happening for that agent.
Craig Durr: Wow, that's pretty incredible. I mean, the level of details that you can get out of this, and that goes back to maybe managing the guardrails or other things as well. And we're not even getting into some of the WEM features right now. But now, all of a sudden, what you're talking about is managing agents and humans side by side.
Jonathan Chang: Yes, and I think that's really important. We really believe that in the world of AI, AI agents and human agents are both first parties and delivering great customer experiences. You have to rethink how you actually manage that; how you do quality, how you manage demand, and how you schedule human agents, and what does that mean for AI Agents. So we built a native WEM portfolio that really starts with: what does it look like when you have a hybrid work? That's one of the unique differentiators that we provide right with our Workforce Engagement Management (WEM) solution.
Craig Durr: You know what makes me really impressed is that the term connected intelligence isn't just a marketing term here. You guys are showing use cases, workflows between humans and humans, agents to agents and humans to agents, and that's a perfect example of that.
Jonathan Chang: Yeah, absolutely. I'm very excited about where we're going, too. Even on observability, we have an integration with Splunk coming soon. And a couple of the recent acquisitions that they've made to provide even more detailed observability capabilities to rightly have full control and trust of area agents. So you can sleep better at night, knowing that you're providing the best experience for your customers.
Craig Durr: Well, this is a great update. Now, if people aren't here, we were just talking that you might be back here for CCW, back in Las Vegas again.
Jonathan Chang: Yes, we will be back here at CCW, sharing a lot of these great innovations, as well as a few new things.
Craig Durr: So we got some announcements coming, potentially. All right, we're gonna hold him accountable for that one.
Jonathan Chang, this has been a fantastic update. I appreciate you bringing this to life, actually seeing the strategy and seeing how it comes together.
Jonathan Chang: Absolutely, thank you so much Craig.
Craig Durr: Everyone, this is Craig Durr, Chief Analyst at the Collab Collective. We are here live at Cisco Live 2026. We'll bring you some more updates. Take care.
