By Enterprises deploying AI agents set intentions clearly. The agents still produce the wrong outcomes. Agentic systems do not behave like human colleagues who interpret intent; they follow instruction paths, and any gap between what is written and what is meant becomes a failure mode at production scale. Without measurement loops and enforced guardrails, those failures compound silently.
In this interview on TFiR, Kamal Hathi, SVP and GM of the Splunk Business Unit at Cisco, covers why intention does not equal outcome in agentic AI deployments, what architectural decisions reduce that gap, and how Cisco combines Splunk observability with network-level enforcement to make remediation fast enough to matter.
Guest: Kamal Hathi, SVP and GM of the Splunk Business Unit at Cisco
Show: The Agentic Enterprise
LinkedIn: Kamal Hathi
Watch the discussion on YouTube
Here is what every platform engineer and AI architect needs to know.
Technical Deep Dive
Q: Why are agentic AI outcomes so difficult to predict in enterprise deployments?
Kamal Hathi, SVP and GM of the Splunk Business Unit at Cisco, states that the core problem is assuming intention equals outcome. Teams build agents with a mental model closer to a human colleague who can interpret intent, but agents require extreme explicitness to behave as expected. The gap between what a builder intends and what an agent actually does is wide by default, and it widens further without continuous measurement of real outputs against intended ones.
“Intention doesn’t equal to outcome. So when people go build AI solutions, they have certain intentions and they think like, oh, I have an agent, just a human being almost. I can tell them what I want to do and they will just do it. The fact of the matter is that this is not the case.”
Kamal Hathi, SVP and GM of the Splunk Business Unit, Cisco
Q: What architectural decisions should teams make before deploying or rearchitecting an agentic AI solution?
Hathi identifies two non-negotiable requirements: very clear, explicit instructions for every agent action, and rigorous measurement and evaluation of whether outcomes actually match those instructions. Guardrails must then enforce those evaluations in production, not just at build time. Skipping any of these three elements leaves the deployment dependent on luck rather than engineering.
“You have to measure everything. You have to measure the outcomes very, very carefully. And so the thing to make sure is when you go build the solutions, a, you have very clear instructions what you want to do, and B, you have got very clear measurements and evaluations of what that outcome is actually being met or not. And then you have guardrails that enforce those.”
Kamal Hathi, SVP and GM of the Splunk Business Unit, Cisco
Q: What does Cisco Splunk’s observability stack for AI actually monitor?
Hathi describes a stack that spans the full AI runtime: GPU utilization, vector databases, agent activity, and token consumption. This breadth matters because failures in agentic systems can originate at any layer, and a gap in visibility at one layer makes root cause analysis at another layer unreliable. Cisco positions this as the parachute that makes AI deployment confidence possible rather than accidental.
“We provide this parachute with our observability for AI, all the way from the GPU, its usage, to vector databases, to agents, to tokens, all of it. And that’s really important. Without that, it’s hoping for the best.”
Kamal Hathi, SVP and GM of the Splunk Business Unit, Cisco
Q: How does Cisco use network position to enforce AI security policy faster than traditional patching?
Hathi explains that the network is both the primary source of observability for lateral movement and the enforcement instrument for remediation. Cisco Hypershield uses intelligence from switches and network endpoints to deploy protection shields directly on endpoints, informed by anomaly signals from Splunk, before the underlying vulnerability is patched. This closes the window between detection and enforcement that traditional patch cycles leave open.
“We talked about this idea of hypershields, that’s something that Cisco has been delivering and the idea of a live protect, which means we’re able to go off and on the network itself using various switches, able to go off and come up with these protection shields that are available right on the endpoint, but they are informed by what you’re seeing in things like Splunk.”
Kamal Hathi, SVP and GM of the Splunk Business Unit, Cisco
Q: How does combining Splunk observability with Cisco network enforcement create a security capability that neither delivers alone?
Hathi frames Splunk as the correlation and anomaly detection layer: it finds what is happening and generates insight. Cisco’s network layer is the enforcement layer: it controls endpoints and applies policy. Neither is sufficient independently. Splunk without enforcement produces alerts that require slow manual response; network enforcement without Splunk’s correlation acts without the intelligence needed to target the right threat at the right moment.
“Splunk can tell you what is happening, get the correlative insights, find anomalies. Cisco can control the endpoint as well as enforce policy in a very effective way.”
Kamal Hathi, SVP and GM of the Splunk Business Unit, Cisco
Resources & Documentation
- Splunk by Cisco, enterprise observability and security platform covering AI workloads, GPU, agents, vector databases, and tokens
- Cisco Hypershield, network-native security enforcement that deploys protection shields on endpoints before patches are applied
***
👇 Click to Read Full Raw Transcript
Swapnil Bhartiya: I always say the Cisco sits the network, nothing goes without that. So you sit in a position that everybody would want to be in. And now when you put Splunk in, that mixes the whole observability. So let’s look at also how does this combination position you folks in helping some of these, Whether it’s token related problem, whether the whole harnessing problem, whether it’s also taming the agents problem. How do you look at this combined story?
Kamal Hathi: So the network is an interesting thing. Network is a source of observability, of visibility and also is the instrument of enforcement. So when you have security exploits, this is lateral movement, it’s across the network. Usually you know, when something jumps from here to there, the network is where it’s happening. And so we are able to go and get data from the network at a very deep level, but also by making sure that we are taking advantage of the smarts in the network that we have to filter the data, make sure the right information comes through. And then we are able to go and use network segmentation, for example, to enforce any kind of remediation that we want to do. So this combination is really important. So the role that Cisco plays is very interesting and somewhat unique. We combine the visibility, the observability, the scale of Splunk with the ability to connect across the network to get the level of detail that is nearly impossible otherwise, along with enforcement points that are on the network itself. So I don’t know if you heard a little bit today we talked about this idea of hypershields, that’s something that Cisco has been delivering and the idea of a live protect, which means we’re able to go off and on the network itself using various switches, able to go off and come up with these protection shields that are available right on the endpoint, but they are informed by what you’re seeing in things like Splunk that are telling you, hey, this is the kind of odd public exploit. And we immediately, very rapidly can go and deliver those even before the network is patched. So this combination becomes very powerful. Splunk can tell you what is happening, get the correlative insights, find anomalies. Cisco can control the endpoint as well as enforce policy in a very effective way.
Swapnil Bhartiya: Excellent, thank you. Now of course almost every company is on the AI journey, but a lot of them have started rethinking how they’re using it. So if somebody asks you are there the specialty car rails or architectural decision that they should make if either they are embarking on their AI journey or they are rethinking their rearchitecting their AI journey. What would that be?
Kamal Hathi: I think the one thing that everybody’s learning very fast is that intention doesn’t equal to outcome. So when people go build AI solutions, they have certain intentions and they think like, oh, I have an agent, just a human being almost. I can tell them what I want to do and they will just do it. The fact of the matter is that this is not the case. The outcomes of agentic solutions actually are extremely difficult to predict unless you’re very, very explicit about what you want. But then, second thing, you have to measure everything. You have to measure the outcomes very, very carefully. And so the thing to make sure is when you go build the solutions, a, you have very clear instructions what you want to do, and B, you have got very clear measurements and evaluations of what that outcome is actually being met or not. And then you have guardrails that enforce those. Without that, it’s like jumping off a plane with a parachute. It might turn out that, I don’t know, you get lucky and you want to float and land in some nice snow bank and survive. Unlikely. But with a parachute, you’re confident it’s going to work out, and so you need the parachute. And we provide this parachute with our observability for AI, all the way from the gpu, its usage, to vector databases, to agents, to tokens, all of it. And that’s really important. Without that, it’s hoping for the best.
