Transcript
with an emphasis in our observability solutions. And I am Staff Technical Trainer for Academy, Cheryl Nomanson, who also specializes in our observability solutions. Today, we're here to address a very special topic to us and hopefully to you, which we've been hearing about kind of in the THWACK forums, we've been hearing about from you live in person and even maybe in office hours or in other places, right? We know that you're asking a lot of questions. We announced in October, the SolarWinds AI Agent. And a lot of people are curious if it's just going to be for SaaS or where we're working on this, if we're going to stop working on other things. Well, we're here to reassure you and to share with you a demo of it when it's going to be in the self-hosted solution. So what we're going to do today is talk a little bit about what is the AI Agent? What is the agentic AI doing? And then we're going to dive into a hands-on demo. We're going to show you what's in Tech Preview right now. And you can look forward to this coming to your products in 2026. So the agent that we're using is called an agentic agent. And really what it allows you to do is have some level of autonomous reaction by the agent for what's happening in the environment, but still have user input into those reactions. So it's not, you're not passing off the complete management of your network without hands into it. So you still have control of it. The nice thing about it is that any kind of question or anything that you ask is asked in plain English. And so you just type in what you're thinking. Tell me about this. Or I want to do this. How do I do it? And the agent's going to respond back to you. And it's going to interact with other APIs and other agents, essentially, to give you the information that you need. And we'll be seeing that in the demo. One of the really cool things is that it's going to be built on MCP, which is Model Contact Protocol. This means that in the future, at some point, we'll be able to interact with other LLMs, other resources, APIs, that kind of thing. So you can get more context from your SolarWinds agent from other places. Because we know we're not your only vendor. You've got information in other places. And maybe you need information that's connected somewhere else. So in the future, hopefully, that'll be something that you'll be able to do. We're really excited about the possibilities by using the MCP to build this AI agent. And we haven't started from nowhere. We have AI features already, like anomaly-based alerts. So those of you that have used the anomaly-based alerts, and those of you that may have been building alerts for a long time, know that you can build a really, really good alert. But it still wasn't able to dynamically react. And so using this kind of AI technology, we've been able to really optimize alerting to react to the dynamic changes in the environment and have the alert be able to react to those as well. So you've already been using this technology. What we're going to do is take it a big step further. So instead of talking about it, we're going to show you in the demo right now. So now what we're looking at is a demo environment that has the tech preview. As you can see, up here, we have a new little icon that you probably haven't seen yet. So this is our AI agent. And just like that, it just pops out to the side. So we got a standard home summary dashboard here. What would we like to ask it? I would like to ask it to show me how to use a particular feature. How do I trace traffic between my nodes? Now it's analyzing. It's thinking about it. Looking for relevant documentation. And this is linked up to the SolarWinds documentation. So you can be assured that it's going to be relevant information and is going to be accurate, because it's the information we put out there for you to use. So you can see here, it's answered your question. I didn't ask it about a specific feature. I asked how I can do tracing of traffic. And it's going to give me a lot of information here on how we can do that, including how to get there and what you need to do. So very quickly, we got that information. It gives us links to our sources so we can see what documentation it's pulling from. And this is going to be true pretty much any question you ask it, that you need to know how to do something. That's great. And then I can just refer to the documentation. It gives you other maybe features that you didn't know existed, like the QoE feature, but also results for using a NetPath feature. So again, you do not have to know what the feature is called and specifics about it, because if you knew that, then maybe you didn't need help with it. But you can just say, I need to do this. I need to track this. I need to see this. And the AI agent is smart enough to go say, oh, you're trying to use this feature or that feature. It finds that information and matches the feature to your query, and then says, this will help you do this, or this, or this, depending on exactly what you want. And then just gives you the reference document, which also would be very difficult to find in the knowledge base if you don't know what you're looking for. Finding a reference document for it is a little hard. Yeah, that's true. So that is a really good use for the AI agent. You don't even need to know what the feature is. You just say, I need to do this. What else do we think we could maybe ask it? Something not about documentation. Right, so looking at the Home Summary View, when I was an admin, I'm kind of putting that hat on. And if I was looking at this, some of the first things that would jump out at me is, I have 14 nodes in Warning. I have 16 nodes in Critical. So from the perspective of managing or monitoring my environment, I'd be like, what is happening with my 16 nodes that are in Critical? Environmental data provided. I like these little updates. I do, too. Because it tells you what it's doing. What I like about that, too, is I could click on that and then start going to click on all these nodes. Or I can just ask the AI agent to do the heavy lifting for me. I'm still going to get at the same information, but I'm going to get at it faster. And if I have entities in my environment that are critical, I want to get at that information as quick as possible. Yeah, and you'll see here, whenever it spits out the response, it's going to give us some context around that as well. So we don't have to go individually check each of those nodes. We can get information on why it might be in a critical status from the AI agent. So here we've got our Critical Node Status Analysis. So it tells us about the 16 nodes. We've got a list of what they are. We've got root cause categories, which I really like this list. Because now it's breaking it down for us. Well, it's grouping them together, right? We've got hardware health issues. We've got connectivity issues. We've got storage issues. Whatever they are. And then now you've got a quick look at where those issues are, what we can do about them. It'd be easy if all those 16 nodes had the same problem, but you and I both know they never have the same problems, right? So it gives me a list of the different things that are going on with those 16 nodes. Yeah, well, and it gives you more context even. So this one has SNMPv3 username not recognized. That's even deeper in. So it's not just that it's a connectivity issue. It's specifically a username issue, which means that probably that username and password are bad. So you should probably go update it. Right, so it's an authentication failure. That would be critical because it's not able to pull anything. If I was looking at a critical node, I would probably get to looking at the SNMP credentials at one point, but it may not be at the top of my list. So again, what we're doing is we're getting you to the potential root cause faster. Troubleshooting, a lot of times, is throwing something against the wall to see what sticks. I mean, let's all admit that is sometimes how you do troubleshooting. And this one's saying, no, no, don't throw the memory balloon against the wall. You need to go over here to authentication issues. Yeah, giving you a place to start at the very least. And then you have this immediate action required. These are very serious issues, and it's telling you why they're serious issues. And as you mentioned earlier, recommended next steps. So this will get you started down the road. And then we don't have to go dig into each individual device and kind of figure out all of those things. You don't have to necessarily have alerts set up for all of those individual things. You can just ask about it and then go solve problems. And what I like as well under the immediate action, again, it reiterates which problems are on what nodes. I'm like, oh, I have an authentication username issue. I'm sorry, what node was that? It tells you which ones they are. I think to follow up on this, I'm looking at the potential issues. And it says, there are eight nodes that have interfaces that have critical status or have problems. I'd want to know what interfaces those are. Those nodes could have numerous interfaces, and I could go look at the node and try to find the interfaces. Or I can ask the agent to tell me which interfaces are in the critical status. From these nodes, which interfaces are in critical status? Quick, easy question. Because in part of troubleshooting, I'm always going to want to ask a follow-up based on what I'm seeing in my data. I don't know what the follow-up question is going to be until I start troubleshooting. That's right. You can see that this has gone quite quickly, right? We're waiting for it to return. But think about how often you have to do a database query or you're trying to build a report. How long does it take for that report data to return? So we're still doing that. We're querying the environmental data. So it's querying the back end to find that information for you and get you more information. And as we can see, we followed up our question. Now we've got the interface information for these critical status interfaces across those 11 nodes. So we have a lot of information here, the specific node, then the interfaces that are in that status. We've got some individual status information here as well. So when they're administratively up, it's giving us a little bit more context on the interfaces on those devices. And you think about it, too, is that it had to look at 16 different nodes and all the interfaces on it because we asked it to tell us which ones were critical. And I see here nodes with multiple critical interfaces. That's actually really good as well because I know that I can prioritize these. So if I have a node that has one critical interface, maybe it's like an edge node or something, it's a little bit lower priority than maybe I have a central or core router that's got four or five interfaces that are in a critical status. That has more impact in my environment. So I'm going to want to troubleshoot that one first. And once again, it's provided us with some recommended actions. And one that caught my eye down here is there's a priority focus. So there's pay special attention to a specific uplink interface because it's marked as critical infrastructure. So it's pulling that context in to tell you, hey, this is critical infrastructure. Probably you knew it was down because somebody was yelling about it. But maybe you didn't. Maybe it's not down. Maybe it's just slow and no one's complained yet. Now you have more information, you have more context, and you have a place to prioritize and focus. So that's been a great look at some of the capabilities that the AI agent has in Tech Preview right now. This is what we've started out with. This is not finished. We're not going to be done here. There's going to be more things coming. So make sure that you stay tuned to the product spaces on THWACK. They'll be updating as new features come out. They'll be giving more information. You can also ask further questions. Put feature requests in, like always. So if you have more use cases after this comes out, and you're like, I really want the AI agent to do this, keep using those feature requests. They're great. But also, stay tuned to SolarWinds Academy because we're going to be putting training out to help put some guardrails and guidelines around how you use this tool and how to best use it, how to use it in an optimal manner. So please stay tuned to the SolarWinds Academy for our training videos that will come out for this feature. We hope that you're as excited as we are.
