Streamlining Cisco Catalyst SD-WAN: Harnessing AIOps and AI Assistant

I good morning, everyone. Thank you so much for joining our webinar for SD WAN AI ops. This session is called streamlining.00:01Cisco Catalyst SD WAN harnessing AI ops, and AI assistance. Before we start our session, I would like to share with you some housekeeping activities. The 1st is that this session is about 60 min with the final 10 min reserved for live Q and A, but at the same time we will, our panelists would answer.00:20To your question throughout the sessions. And so please if you have any question, use the Q and A panel for your questions and our panelists will help you with the answers if they can, and if any question remaining, we will answer it at the last 10 min. These sessions will be recorded and the presentation would be shared with you after.00:40Through the events, so all attendees would get a recording and the slides from these sessions. All participants are muted on the as entered the sessions. We will answer any questions you have on the chat and raise your hand if you have any questions, but there are Q and A panel is the main place that we will be getting your.01:00Is an answer, ok? And as I mentioned, we will revisit all the questions at the end of it. We really appreciate your feedback. Once the session is over, please share, stay online and answer those four questions. And also during the sessions, we will have a sliders and for our presenters that would like to hear your.01:20For Feedbacks and also as part of the presentation. My colleague Chris, would share the links to Slido and the code so throughout the session, please check the chat part for the links to the Slido and the questions in there. Deepak can move to the next session, please. Next slide. Thank you. So as I mentioned the logistics, you.01:40You can see on the screen on the right hand side are the the icons for the Q and A and also the questions. Please, if you have any questions, put them in the Q and A space, not the chat. The Q and A is where our panelists are reviewing the questions. With that said, Deepak, the floor is yours sir.02:00Yeah, thank you. Good morning, good afternoon, good evening folks. My name is Deepak Katari and I'm joined by my colleague Ashwin Murdley. We are both part of Cisco's routing business unit, and we are excited to talk to you about today, using or leveraging the power of AI, AI ops.02:18For optimizing your Cisco Catalyst infrastructure. So here is the agenda, we'll touch upon what we'll, we'll share with you. We'll begin with sharing with you what we are observing and what we are hearing in the market. We'll then share our vision followed by providing a brief overview of the.02:37AI ML capabilities we have on the platform. We'll also then give you a deep dive and in demonstration of some of these capabilities.02:57And in the end, we'll wrap it up with key takeaways, resources, and then Q and A So we'll begin with what we are hearing and seeing in the market. So organizations today are going through massive.03:04Digital transformations. And this is putting a lot of stress on the enterprise networks, especially wide area networks. And there is a fundamental shift in the way traffic is flowing over these networks. Now historically it was branches.03:21Connecting to data center hub locations, so it was pretty much a north south traffic, but but with the advant of cloud distributed workloads, distributed data centers, cloud based data centers, cloud based SaaS applications, distributed workforce, and the agent traffic flowing over this network.03:36There is a huge amount of traffic increase on these networks. That's number one. Number two also from the connectivity point of view, enterprises are now leveraging not only traditional links such as MPLS links, but they're also leveraging other form of link types like internet links or broadband links.03:54And then cellular links. So network is NO longer closed ecosystem. It is more like an open ecosystem, so overall attack surface has gone up drastically for customers. And what does this mean for itts? Now, there are more and more compliance requirement or mandatory regulatory repliance.04:13Requirements. So for IT teams, it is becoming more and more complex to manage this infrastructure. They're asked to deliver more while the team sizes may not be growing, they're asked to deliver more on these environments. So this is really, really creating a lot of operational challenges for IT teams.04:33So, and this is precisely what is driving innovation at Cisco. We are looking to help our customers leveraging by leveraging AI, AI level capabilities, how we help our customers operate their WAN environments. Now, when we talk about, AI, there are two ways to think about it.04:58One is optimizing when to carry AI traffic, and that could be about giving a preferential treatment, a quality of service treatment for certain business critical AI traffic or it could be the other way around. Giving less preferential treatment for shadow AI traffic that is flowing through your environment.05:14So that is about optimizing your WAN for carrying AI traffic. That's one part of it. 2nd part is leveraging AI for for optimizing your van. So effectively it is when for AI and then AI for WAN.05:33So in today's topic, today's session, we are going to focus more on AI for WAN or leveraging AI ML for optimizing your WAN infrastructure. Now, here is what we are also hearing from our customers. E.g., one of our customers.05:49Saying that AI is not only enhancing their operations, it is pretty much reshaping everything they are doing in their networks. That's number one. Gartner is also observing that agent in AI will autonomously resolve 80 percentage of customer issues a couple of years down the line. Similarly.06:05Is observing that agent AI is now creating a virtual workforce that can autonomously plan and execute.06:26Many multi step workflows. So ai's the way people are operating their infrastructure. So before we go into this feature specifics and our vision, we have a question for you.06:33When you, when you operate your networks, where do you spend most of your time? Example could be you spend most of your time in troubleshooting an issue or it could be configuring a new site or configuring the entire network or making changes to this network.06:49So we certainly would like to hear your feedback in terms of where do you spend most of your time? There is a Slido QR code on the screen. Please leverage it for getting seeing this question and responding to it. Perfect. The question is up and ready and people are answering.07:06Yeah, we'll give a few seconds before we go to the next slides and maybe Chris or if you could just voice out what you are seeing on the responses, that would be good. Yes, thank you Chris.07:28So, so far most of what we're seeing is troubleshooting monitoring. There is some day zero, understanding templates, troubleshooting and configuration, enhancements and new solutions there, we're getting.07:42Like a wide variety. Okay, ok. Sounds good. Thank you folks. So now we'll touch upon what is Cisco's vision. So there are three essential areas that we are thinking about helping our customers.07:59Leveraging AI. The 1st one is troubleshooting. Now, historically, IT team spend hours and hours and maybe days and weeks of time in manually investigating problems. Now this is where we, we want to leverage AI or we are looking to leverage AI for helping our customers bubble up key insights.08:15That will speed up the root cause analysis process for our IT teams. And that would reduce the mean time to resolution or mean time to innocence.08:35For this IT teams. So that's number one. 2nd one is optimizing the entire experience of operating this network. So this is about AI helping to continuously improve your network performance and your application performance. The intention is to empower our IT teams to become more proactive than being more react.08:47So that's 2nd area. And the 3rd area is about gaining the operational excellence, and that is about leveraging AI to help with day to day operations of.09:07Maybe deploying changes or making changes in a more consistent and reliable manner. So effectively the idea there is to help IT teams deliver more with less effort and with greater confidence. Now let's look at some of the features we offer today in this space.09:17Now Cisco has been innovating in this area for some time now, so predictive path recommendation is the 1st feature we introduced in this area. The, the, the concept here is we monitor various application flows going over various network paths in your network.09:39For a long period of time, and we feed this information to AIML engine. Now, as it is obvious, if your network path quality is good, your experience for the applications or your end user is going to be good. So better the network path, better is going to be the application experience.09:57So what we do is we feed the application flow information to AIML engine. It then tracks and understand make does the assessment for this network parts identifies all possible alternate paths and compares the network path quality it could offer. So at the end of this analysis, if it discovers that if there are alternate better parts.10:15Ask for carrying your applications, it will make that recommendation proactively to you to IT teams, and you can then act upon this recommendation so as to improve your application experience. So that's our 1st use case. 2nd one is about bandwidth forecasting, and as the name implies.10:35It is about forecasting the future usage of your WAN links. So what we do is again leveraging AI, we track the historical growth pattern on WAN links, historical usage pattern, and accordingly we forecast what the future usage may look like on these WAN links.10:53Now IT teams can leverage this information to either increase the capacity if they anticipate or they see their forecasting model is forecasting that in a capacity or a bandwidth use on a particular link is going to exceed a certain threshold. They can work with the the providers to increase the capacity of those links.11:10The other way around is also true. If you observe that the capa the links are getting underutilized on certain site locations, you can then potentially reduce the capacity of those links or potentially decommission those links which are hardly getting utilized. So that's a use case. It's a very good tool for capacity planning.11:30The 3rd one is about detecting anomalies in the network using AI. So again what we do is we, we monitor network performance over a certain period of time, then we establish baseline using AI. So baseline could be in the form of loss, latency, jitter on various network paths.11:49As well as the usage on cellular links. So we establish this baseline once it is established, then using AI, we look for deviations in this baseline. If you observe major deviations, we flag it up to our users so that way.12:08The IT teams can act upon it, so effectively nipping the issue with the bird, meaning before the issue becomes a major critical issue in your environment, we basically provide you insight which becomes actionable for you to potentially take action against those anomalies observed in your environment.12:25And finally, the AI assistant. Now, this one is very interesting. It is fundamentally changing the way we envision people operating their environments going forward. So if you, if you go maybe 2030 years ago.12:45All, all of us who are managing our, our network on a device by device basis using command line interface. You go to one device, make some changes, do the configuration and then go to the next device. That's how people were operating the networks.13:00Subsequently came the time when we got a very nice graphical interface, a very nice console to manage multiple devices at a single time. So that is what we have been doing for now several years now. Now, now comes a new era where.13:16We are now potentially leveraging AI to speak to the network. Now, you know, again, what if you are able to ask questions to your network itself in a natural language and network then responds back to you with responses? And that could be a troubleshooting tips. It could be questions about your features or traffic flow.13:31Because any of that, whatever you think of it from the operations point of view, what if a network is able to respond to you with some responses? And that is precisely what.13:51AI assistant or a chatbot window does for you. It is a window into your network to ask, to let you ask questions and that's how it will change the way you operate your networks. So overall, AI is shifting the way customers or IT teams would be operating their environment.14:01So with that, I'm going to invite Ashwin, my colleague to go into details of these features. Ashwin over to you. Thank you Deepak. Let me go ahead and share my screen. All right, yeah, let's get into some more details of these features and we also have.14:20Have some very cool demos lined up for you.14:40So let's take a deeper look at predictive path recommendations. So traditionally, right, SD WAN has is very effective at reacting to network conditions. So it detects degradation and dynamically shifts traffic after you know the performance has been is impacted.14:45But as applications and user expectations continue to grow, reacting after the fact is, isn't always good enough. Cisco SD WAN predictive path recommendation takes a proactive approach here.15:05This capability monitors their network path conditions and performance over a historical time period and it applies predictive data modeling to it. Based on this, it is able to forecast likely issues in the future and accordingly makes recommendations.15:21For using alternate network path parts for your application. These recommendations can be reviewed and applied with a single click eliminating manual analysis and reducing your operational risk.15:39Well, what this means is it results in a more consistent application experience for your users, a fewer unexpected outages and a shift from a traditional reactive troubleshooting to a more proactive and intelligence.15:57One network operations.16:16With that, let's take a look at a demo for this feature.16:18So predictive path recommendations analyzes a large set of telemetry data using your advanced models to predict which network paths have a lesser probability of having SLA violations.16:38And based on this, it recommends the most optimal path. Ok, sorry about that. So on your SD WAN manager, you can see the notification.16:52That is generated by predictive path recommendations. And in this case, in this example, we picked the San Jose site.17:12And when you click on it, you get to see the recommendations for San Jose site. Here you see that there are multiple applications with recommendations, but if we let's focus on Webex, where we see that the the current path quality is 85 % and if you apply the recommendation.17:22It would give an increase of around 15 % in your.17:43Path quality in your quality for Webex application. And here you also see a button where you can click to apply these policy changes. Of course, you do have a opportunity to review all these changes, make sure it's.17:48You know it's aligning with your business needs and with this with a few clicks, you can go ahead and apply these changes.18:07You click on activate here you get to review the, the conflict. You can see the conflict preview as well as the config diff. And if you are satisfied with the changes, we are comfortable with it, you can go ahead and activate the.18:22Or implement these changes. So here as you can see, right, this simplifies applying the recommendations and at the same time, it also ensures most optimal experience for your applications.18:40Let's move on, the, the next feature I want to talk about is anomaly detection. Anomaly detection is essentially designed to help operation teams catch deviations from normal network behavior very.19:02Early before they turn into more widespread performance issues that impact your users or applications. So we have two flavors of anomaly detection available today. The 1st capability focuses on.19:22Or network KPIs such as your loss latency, and jitter. It's essentially designed to help operation teams catch deviations in your tunnel performance which are indicated by your loss latency jitter before they manifest into larger issues.19:39Using, AIML, the system continuously learns what normal looks like in your SD WAN environment. Essentially we we compute a baseline for it. And then based on that, it it tracks the deviation from that baseline for those network KPIs.19:59Oh, rather than instead rather than relying on static thresholds, it continuously adapts to changing network conditions over time and highlights, you know, anomalies that mean meaningfully deviate from your, the expected behavior.20:18In addition to that, it also provides a high level view of sites and site pairs with the highest anomaly levels. That way you know where to focus your attention on, and it helps operation teams quickly pinpoint where the abnormal behavior is concentrated.20:38So that they can prioritize investigation and expedite troubleshooting. The the real, you know, the real value or the benefit for you is that it helps surface issues earlier and reduces manual analysis and.20:54Great troubleshooting. And as you can see in this slide.21:14We do have a graphical representations of these anomalies and along with that, we also show other supporting information such as, you know, usage during that time so that way you get a holistic view into what could have caused those anomalies and.21:21What you could do to kind of prevent them from happening in future.21:39Moving on, the other flavor of anomaly detection we have focuses on cellular links and helps identify unusual usage patterns across your SD WAN network. This is valuable because abnormal cellular usage is often a symptom of something else happening in the network. E.g..21:48Failover or you know your your your your your main circuit or the preferred circuit is down or there is a sudden surge of traffic or applications. So all those things can be troubleshooted or inferred.22:08Using this anomaly detection for the cellular link usage. Once again the so as you can see here in the graph, we do show the anomaly percentage over time and then you can.22:27Narrow down to a specific time period where you, you're seeing lots of anomalies.22:46And if you click on that, you get to see more details. Let me show, let, let us jump into a demo for, for this feature. Okay, so here what you see.22:53See is the the anomaly detection in cellular usage.23:13And the 1st graph on the top, is, is a anomalous percentage of anomalous circuits over time. And you also get to see the, the usage which is in in in the, the gray bar. So.23:18When you click on a specific bar or a point, you get to see the table below updates to show information for that particular time period so.23:36In this case you see that at 10aM. On eh, 18 August.23:56There were, the total usage was four GB and there were about 5 % of anomalous circuits which boils down to around three circuits where anomalies in usage were detected.24:04During this time period. And when you click on this, the table updates and here you get to see more information about the different cellular circuits. And then in the graph below, you get to see the actual deviation. So in this case the baseline.24:20Is two GB, but we see that the actual usage is four GB indicating that there is a anomaly in the, the usage on this particular cellular link. And below that you get to see supporting information. E.g., you get to see.24:39At that specific site, what are the other circuits that are available? So e.g., in this case, we see that the LTE circuit is there and we also have a public internet and we see that majority of the traffic is going on the LTE and not on the public.24:57Internet. So either it could be, you know, policy configuration issue or maybe for whatever reason there maybe some issues in the public internet link, which, requires further investigation.25:17In addition to that, you can also see what were the applications that went over that cellular link during this time period. And here we see that Google Services was the most used application.25:33During that time period that went over cellular link. And we can also see who are the client IPs associated with the traffic that travels the cellular link during this time period. So as you can see, we not only bubble up the anomalies related to cellular link usage, but.25:50We are also providing a lot of supporting information to help you narrow down the issue and quickly troubleshoot as to why your traffic might have taken the cellular link.26:09With that, let's come to the, so we're at 2nd slider question.26:31So this question, you know, we would like to know how you, you would like to use AI in your WAN network. And there are six options here, troubleshoot issues faster, optimize network and application performance.26:36Use AI to configure new sites or make policy changes or maybe perform things like security audits or certificate management. So I'll give a couple of seconds for you to respond to this slider question.26:54Chris, do we have any responses? We absolutely do. So far troubleshoot issues faster is the most often selected. Optimized network and application performance is next followed by perform security audits.27:32Those are the top three by far, so cool. Yeah, thank you for providing your feedback. Let's continue with the presentation. The next feature I want to talk about is the bandwidth forecast.27:49So one of the challenges that we have heard from a lot of customers is that obviously without adequate bandwidth, your business operations can be impacted very quickly. You know, your application performance degrades, the end user experience suffers.28:08And critical, your critical workflows may slow down at exactly the, you know, the wrong time. The problem is that with many teams, is that many teams only recognize that a bandwidth issue recognize a bandwidth issue after condition has already started impacting your users. So again, you're.28:25You're reacting rather than proactively trying to address the issue before it happens.28:45Bandwidth forecasting addresses that by tracking historical usage on your individual WAN links and using AI ML to project or forecast future usage several weeks in advance.28:51So this gives your operations teams early visibility into where capacity demand is heading and and help identify links that become more constrained and take appropriate action. Again, as Deepak.29:06Mentioned earlier, it's not, not only about increased capacity but also it will help you determine if you have over provisioned certain circuits or you may want to right size them and that's exactly where bandwidth forecast gives you that.29:26Insight which you can leverage to ensure you have the right capacity for your networks. And, and with that, you know, you can all proactively work with your ISP to make better capacity decisions and avoid situations where.29:45Business is impacted simply because, you know, demand growth was not visible, you know, early enough. So as you can see in this slide, we, we, The the blue line shows the actual usage of traffic.30:02And the dotted line is indicating the forecasted usage. And you see that the forecasted usage goes beyond the, the, the current or the past usage and not only do we show the average forecasted usage, but we also show a band, so that kind of helps you understand what.30:17What is the, the, the maximum or and the minimum.30:37For a capacity that you can expect in the future. With that let's move on to the next one. Yeah, so I'm really excited to talk about this feature. In fact, I feel this is one of the most exciting capabilities that.30:42That is available in the SD is catalyst SD WAN portfolio.31:01Typically right, for troubleshooting or monitoring, you, you probably have to navigate multiple dashboards or search through, you know, the same time search through documentation to kind of get to the information that you're looking for. With AI assistant, we offer a conversational intuitive interface.31:07Where you can simply ask questions and the assistant responds to your query.31:27It can surface relevant documentation information or provide guidance during troubleshooting monitoring workflows or, and even help with tasks like managing support cases, your Cisco TAC support cases.31:34By bringing this knowledge and assistance directly into the operational workflow, the asset and helps make information easier to access for your, for your teams. It improves overall your team productivity and reduces the time it takes to find answers or move forward on issues.31:51Here are some use cases that we offer today through assistant. The 1st one again, as I mentioned is troubleshooting, where you could ask questions such as how is my network performing, what are the top used applications? How are my circuits doing? What are the.32:14Most utilized circuits and various types of questions. The key thing, the key thing to note that it is completely non prescriptive, so you can ask the question in any which way you want and the assistant is intelligent enough to interpret your question and respond to it.32:352nd use case is more around documentation where you could ask questions such as, you know, what is how how do I configure a specific feature or what does a feature do? You know, what is new in this specific release of SD WAN manager? So anything related to documentation.32:55And for this we are leveraging the public the Cisco public documentation for SD WAN to respond to your queries. The 3rd use case is case management where you can actually create or open a tag case through the assistant. You can get a.33:13Summary of all your TAC cases that you have opened, you can sort them by priority, and you can also close a case through A assistant. You can also get.33:33More detailed information about a tag case. So all your operations can be performed through the assistant without you having to navigate through to multiple dashboards, either within the SD WAN manager product or to CX portal from where you would typically open tag cases.33:45With that, let's jump into a demo for EA assistant.34:07Right, so this is the interface, the assistant interface.34:25And as you can see I've already opened it and at the very top is you see an icon here so you can click on it and you can, and the assistant UX or the window would open.34:31For for your convenience, we have also added some starter prompts. E.g., network overview, help me troubleshoot, help me configure and what's new in this release. So you could either leverage those or you can ask any free flowing question.34:50In the assistant prompt window. So let's for this 1st demo, I want to showcase how you can leverage assistant to monitor your application health and also perform some network planning.35:05So the question I'm asking is, to begin with is, for the AS system to provide a summary of my network and application experience. So what the AS system does is the moment you ask your query, it 1st analyzes your query.35:21And then it creates a strategic plan on how it wants to address this query of yours. So in this case, it is, as you can see, it is looking, it is going to be looking into various metrics such as your app sitewide health, your application.35:39To help circuit, device, et cetera.35:59And note that even though I have not provided any specific time period in the query, it considers the default time period of 24 h. Of course, you can customize it based on your needs. Since this has multiple elements to to.36:03Wait, the query has multiple elements to it.36:23The assistant is now executing some of these queries in parallel. So it's executing five independent queries in parallel to get you the information very quickly. So here's the response where you get to see an overview of your site 1st where.36:26Where it gives you a high level information about your sites, the availability, latency, and other important information related to your sites.36:45Now that is followed by information about your application quality where it lists the, some of the critical issues in your network related to applications. And here we see that Office 3605 has a fairly low quality of experience score.36:54And it also provides the associated loss and latency. Furthermore, it gives you information about your circuit health, the total circuits active and the some of the key performing indicators for your circuits as well.37:14And finally, it also gives you some additional information about critical anomalies that were detected in your network during the last 24 h. It also gives you some recommendations which you can use to further investigate the issues.37:35To kind of narrow down to what could have caused some of the issues reported by assistant. Now let's switch to the network planning use case. So here I'm simply asking assistant.37:56To analyze my flows in the last month and based on that, should I be configuring a feature such as TCP optimization? Of course you could ask this for other features too, but this kind of helps you with the you know network planning to know what features are best.38:13For your network and applications. Once again, the assistant analyzes your query, comes up with a plan, and note that in for this specific question, not only is it leveraging the, the, the flow data.38:33To, to provide recommendation, but it is also searching the documentation. So it's in parallel, it's it's looking at different data points, so operating in a truly agentic manner.38:52To get you the the right information. And here you see as part of the response, we again see that Office 3605 is has exhibiting a significant TCP performance degradation and which is also what we saw in the previous query about network.39:08That.39:28And finally it recommends that you should configure TCP optimization based on your 30 day traffic pattern, and it provides some additional key findings. It also lists why TCP opt actually would help in your case. Now you.39:29You could either investigate further to see if you really need this feature or if you're satisfied with with this response, you can also ask the assistant.39:49How you should, what are the steps needed to configure TCP optimization feature? So in this case, it once again goes through, looks at the documentation available and comes up with a response.40:03Giving you the steps required to configure TCP optimization feature. So this was a quick demo where you got to see how AS system is really helpful with in when it comes to basic monitoring your network and application health monitoring, but it also is helpful.40:19In some advanced use cases where you with network planning where you need to analyze your flows over the last month, which is not a trivial task. So, so that's where the true power of shines through, where it can do a lot of that analysis and.40:38Gives you and give you the recommendation.40:58Which you know you you can choose to implement. So with that, let's, I think we are, we have come to our next slider question. So the question is, how comfortable would you be allowing.41:00AI to take automated actions in your network.41:20The 1st option, AI provides insights and recommendation, but the user makes the changes manually. 2nd option is AI provides insights and recommendation. User approves it and the system applies changes automatically.41:24The 3rd is AI automatically applies low risk changes without approval, but broader changes require human approval. And finally, AI fully automates network remediation and operational challenges.41:42So let me pause for a few seconds and give you time to respond to this. And Ashwin, just to do a quick time check, we've got about 15 min less than the left in the webinar.41:59Yeah, so far the main selection has been AI provides insights and recommendations with user approving and applying automatically. Sorry, is that which option is that? It's B You're B Okay, got it, ok.42:16Cool. Alright, yeah. Let's move on in the interest of time. So I have another EA assistant demo for you. I'll quickly walk you through that.42:35This is more related, this is related to troubleshooting use case. So once again, you know, you can use the AS system even for troubleshooting use cases. So in this specific scenario I'm asking if there are any circuits whose utilization was more than.42:5170 % in the last 24 h.43:11Once again the assistant does its magic where it's interprets your query, analyzes it, and then gets the information that you are looking for. So as here we see the response where it has identified this specific circuit.43:13In my San Jose site.43:33Whose peak utilization was 74 % and it also provides some additional information what the actual usage was and what the configured bandwidth on that circuit was as well. Now, I could ask a follow up question where I want to know what are the.43:35You know, applications that were contributing to this.43:55Increase in utilization. So again, assistant does its magic and then it lists the applications that were most that contributed to this utilization spike, and in this case, since it's my lab setup I was.44:01Running a lot of IPOV traffic, and that has contributed to about 95 % of the total traffic during that time period. So you could further ask questions such as who are the clients that contributed to this flow or you could, you know, you know, if in this case if it's IPOV traffic.44:20You know, you want to block it, you can ask a documentation question on how to go about doing that and you know if you after trying your various things, if you're still seeing any issues, that's where the case management use case, comes into play where you can create.44:40Tag cases right from the assistant. So as you can see, assistant is very powerful. Of course, you know, we have started off with some core use cases and we are continuing to expand on it. But I think this is really going to redefine network operations.44:57And kind of, make things make knowledge more accessible and at the same time improve your total overall operational efficiency. With that, let's go back to the.45:15Slide. All right, so we're almost to the end of the webinar. So as we wrap up, right, so here are some key takeaways. So the 1st is again van operations are becoming more far more dynamic and complex. You have applications that are now distributed.45:34Across data centers, SaaS, cloud, and increasing, and we are seeing increase in AI workloads.45:53So all this is kind of making the net you know network the operations more complex and dynamic. What AI enables is a fundamentally different way to operate the network. Instead of relying only on manual analysis and fragmented tools.45:59Your IT teams can now leverage predictive insights, intelligent recommendations, and AI assisted workflows directly within your SD WAN catalyst SD WAN solution. And we saw some of the features such as predictive path recommendations, anomaly detection, bandwidth forecast, and.46:18The assistant, we saw how they can help teams detect issues earlier, resolve problems faster.46:38And simplify many of your day to day operational tasks. And the result is a network that's not just easier to manage, but one that's consistently delivering the best user application.46:46For your end users. And that's really the opportunity that Cisco Catalyst SD WANs, AI powered capabilities offer, transforming, you know, how your network operations teams work while ensuring the network keeps up, keeps space with the demands of modern applications.47:02So the last slider question for today, what additional tasks would you like Cisco to assist you or help you with? And it could be anything from, you know, more troubleshooting support, configuration or, you know, any other area that you think Cisco should be.47:25We are focusing on.47:44And let me give you a couple of seconds.47:46Yeah Chris, any.48:07Really indicators for this question. Looks like configuration and policy optimization are coming in mostly. There's a lot of troubleshooting as well. Okay.48:11Okay, thank you once again. And here's a slide on resources. Of course, I think you will be will be sharing the slide deck with you so you can take a deeper look at these. And with that Chris or Costal, I'll hand it over back.48:30Perfect. Thank you so much, really appreciate it. I believe that there was quite a few interesting questions where on the chat and where we try to answer as many questions as we could and so if anybody has any questions guys panelists please there was any question.48:53That caught your eyes and you want to answer them live.49:13Okay? One question that I saw was that interesting was that if a single site is having a problem, but template is attached to multiple sites, how does AI decides to make the config change or not? Maybe I can take that one. So what we do is if you define a.49:16Site list which comprises of all these sites. Let's say one or two of the sites are where the recommendation is, we actually through AI automatically create a, we modify that site list, create a separate site list for those sites where the policy change has to be applied, and that's how we we manage that.49:36Excellent. And the one recurring question constantly was that, do I need thousand eyes licenses or and what's the lowest version of the software that I can use for this analytic in. One is for AI assistant, one is for predictive path.49:54Yeah, maybe I I can take that. So the AS system is available starting your SD WAN manager twenty.fifteen version. The case management use case that I spoke about, that is available starting twenty.eighteen version of your SD WAN manager.50:10To your other question related to thousand eyes license, NO, you do not require any additional thousand eyes license to use predictive path recommendations feature. And what's the lowest version for the predictive path and analytic?50:30Yeah, so predictive path is available the closed loop capability where you can apply the configuration through a simple click is available starting twenty.twelve release of SD WAN manager.50:46Yes. Yeah, maybe one thing I would add is, see broadly speaking many of the features we build through our cloud engine, so our our intention is to offer this functionality dynamically to you in an agile fashion, and and of course at times that could be version dependencies as well.51:02But broadly speaking many of the features become available to customers as soon as we deliver them through cloud. Perfect. Excellent. Thank you so much and to everyone, we will share recording and the slides and also some of you that we got your email that we would also include in that email some of the top questions asked and the answers.51:19So that is going to be where what we will share with everyone attendees on this call. Okay, with that said, Ashwin, thank you so much and thank you to all our panelists for the help and also thank you for everyone who joined this session today. With that, appreciate it. Chris can end the session.51:39