Fortinet AI-Powered Transformation

All right. Good morning everyone. Uh, my name is James Allen.

I'm a senior director of Products and Solutions with Fortinet. I have with me, uh, Alex B, who's in the back of the room, who's also a senior director of our product management, uh, uh, organization. So today I'm gonna skip forward here.

We're gonna kick off a, a brief discussion around ai. Now, I know we've presented to you guys before talking about ai, and one of the things we're always quick to point out is that Fortinet has a long history in in AI and AI operations. Specifically for us.

We started our AI journey with our security services. And in fact, a lot of our Forti guard services upwards of 15 years ago were utilizing AI and ML to be able to process big data and give us information as how we would, uh, protect our customers against ongoing threats. We began to introduce, uh, generative ai, uh, through our 40 AI product that integrates directly into our Ford manager.

I know there's a lot of 40 out there when we talk about our different management platforms that, that, uh, Sumina brought up for A manager is actually a manager of our os, our 40 OS operating system. That operating system resides on every FortiGate, regardless of form factor, whether it's a virtual machine, hardware, all of it sits out within, uh, within, for os for a manager manages that and all the operations that it controls, right? Which includes our wired, our wireless, our 5G, our sd-wan, all of that, and the security of co of course.

So when we talk about 40 ai, it really is there to provide us with, you know, that, that that experience for our customers be able to have native language and so on. But as we've progressed, we've moved into a Agentic ai. And for us, what that means is it's the ability for us to be able to integrate not only what many of the, some of our others who've been out before you, AI network operations.

We're also having AI network operations talk to our security ai, talk to our ai that in fact is protecting ai. And that is the three names that you see down there for the AI Protect, for the AI assist, and 40 ai, secure ai. And again, 40 AI Protect is what you would expect from Fortinet.

That is really what we're talking about when we're talking about our Forti Guard services, 40 AI assist. That is really what we're getting into with our management capabilities. And 40 ai, secure AI is really looking at securing AI itself, because if we are protecting the inputs into ai, then we're likely to get incorrect outcomes from the information that AI is processing.

So when we talk about these three pillars again, and just go into a little bit more detail, the 40 AI Protect, as I mentioned, is our security services. The differentiation for us here is that this is all part of an integrated security fabric. We're not talking about individual point products.

We're not talking about AI as a, as a, um, as a specific product. We're talking about AI across the network domain, the WAN and all of the security elements, including the AI itself. And that differentiates us with proven AI models that we've been using for over 15 years.

And with over 500 patents that are pending currently, some of them are pending. And of course, with the amount of sensors that we have just based on both our clients and our hardware, we're one of the most deployed firewalls in the world. We are obviously growing, uh, when it comes to network as well.

We have over 750 million sensors that are out there that are providing information that are feeding our AI engines across both the network and the security domain. James, uh, quick question, yes. Is Foret being enlisted by your customers to not only, okay, measure improve security outcomes, but using these products to measure improved AI outcomes?

Is that something that is, uh, cross-domain? So right now where we're at is having the ais be able to talk to each other and be able to communicate and share information from better outcomes. The measurement of that right now in, in looking at saying, okay, how are we measurably better, providing better outcomes, as you can imagine, are a little bit more of a challenge, right?

We can look at it from a more of a siloed approach and saying, we can pro provide better outcomes from a network perspective and show how we provided better outcomes. We can do it from a security, but then right now we're still early on in being able to give you a holistic view in saying, we improved this much in security, we improved this much in, in, in, uh, in network. And then of course, we've protected your AI information.

That's your information feed. That that's helpful. Yeah.

So we're, we're in the beginning of that journey. Where we're really getting at is we're pulling all of this information and having these uhp disparate groups be able to talk to each other, because at the end of the day, um, what you want is you want better outcomes across the entire spectrum. And, and to your point, you want to measure that.

Um, to give you an idea of what, when I talk about 40 AI assist, which is what Alex is gonna be covering off on in just a few moments, this is really here both from a security and a network perspective. So we're looking at the security operations and pulling in AI information, and that's where you start getting some of our forta guard services that are then feeding into Ford a manager and taking actions, right? Certain automated actions within to protect the network.

And we're also then having AI ops be able to provide predictive insights, be able to look at baselining, be able to have trend analysis, and then be able to offer insights that are then acted upon within Ford Manager. So what you have is, rather than looking at AI ops as saying, well, AI ops is making these changes on the network, and so on and so forth, what's actually happening is, is AIOps is communicating to Ford a manager to be able to make the changes, and again, using more of a native language aspect of things, I want to improve x, I want to improve y those type of actions being taken from a centralized unified management platform. Now, when we talk about what products are really in there, and again, I realize that these are all kind of 40 names, but the good thing about it, the little decoder ring is it really is 40, whatever it does.

And when we talk about the analyzer for the analyzer, this is really our data lake. This is where we're having a correlation of not only information that's happening across the products and the security fabric, but also we're getting information from third party vendors as well. And we're using that as well as our event management and our SOAR to be able to provide automated triage, threat hunting, root cost tracing, all of the things you would expect on the security side, on the network side, or sorry, on the management side, this is where, as I was mentioning before, we're able to take action upon those threats and action upon the alerts that we're getting.

Right? So the AIOps piece, which Alex will be going into more in depth here in just a few moments, that's giving us our baselining, our trend, our forecasting, but again, it's helping us with regards to troubleshooting. It's helping us with regards to optimization.

And again, those actions are then being executed through our for manager, which is managing 40 os. So with that, I'm gonna go ahead and pass the ball over to Alex. Thank you.

Thank you, Alex. So, uh, Sumina and James, were talking about, uh, the approach of Fortinet around ai, and also the fact that we all bundle everything through the management, uh, piece of for manager, uh, before that, the FortiGate is our key product where we have it acting as well as controller for you guys. But it's not only that, as we said over the time, and today, it's a switch controller, it's a firewall I tend to jog.

It's also a firewall. Uh, but also what we, uh, what we use as similar was saying is, uh, has one. And what we aim to get is we have a lot of customers where they have one a quarter on multiple branches through the country globe or whatever.

We aim to facilitate with this example, what we are aiming to get is helping them to come up and say, uh, this is my network. These are the sites I want to interconnect. Could be from natural language, or it even, it could be un crafted network.

Ja. So even you or your kid could draw a drag Ja and then come up with, I want to interconnect that. And then what we would do from the AI backend is we will correlate all that and make the changes on the, uh, firewall policies on the, uh, SD one IPSec to make the communication happening.

And, uh, easy. What we're also doing is, okay, you did the initial setup, but what if tomorrow you bring another site or two sites, or 10 sites? So one aspect is we can also onboard a new device and make it part of that quite easily from there, which is helping also a set, uh, around that.

But it's not only that, if you think that you have, uh, already defined, uh, sites configured, and you want them to say, well, I, I want to come up with an idea to get better and to easily onboard them or easily get a new configuration deployed to sites. So what we can do is come up with an analysis from the AI across multiple branches and come up with a kind of golden template where you would have your variables that can then be used to deploy any new site or any new feature, uh, across your network. So really here to improve and to help assist, uh, on the day-to-day, uh, deployment of the devices.

But you will tell me, Alex, okay, but we are today at Mobility Field Day, so it can also hack, uh, on Wallace. So we can even say create me an SSID, uh, create me a QS profile, V profile, whatever you want on the 48, and it would analyze and take the variables that you have configured. Don't come next to my house because this ID is walking as of today.

I'm kidding. But, uh, this is really here to help for people that are not knowing, uh, or don't want to spend time into, uh, configuring that or going through the deck of, uh, of doing that. So really here to facilitate the life.

And then, so you have deployed your network, you have configured your wireless, and what we can, uh, can also do is we can assist, so assist from a NetOps point of view. So what are the issues on this specific ap? What are the issues on this specific station, uh, across time?

Uh, and then the, uh, AI agent will take the data that we already have in the product and come up with a summarized, digested information for the admin to come up and look into and, and go further with the troubleshooting. So it could be from more, less switching as you one, as I said, uh, around, uh, an ap, uh, a switch station, but it could, could also be, uh, uh, on the SLA check. So for example, on the SD one side, because you all know that a lot of people come to you and say, wall sucks, but most of the time it's not the wall.

It could be the one side. It could be even the application on the remote side that is, uh, that is failing. So we come up with, uh, an holistic way to give you insights not only on the, on the land side, but also on the one side to, to be able to, uh, to give insights into that so that the AI part of, um, uh, uh, of the NetOps and then has, you know, like Fortinet is a security company, so we'll never do without security.

Uh, and GEMS was talking about it earlier. We have, uh, another aspect of that where you can, from the same chat come up with what are the, uh, spreads, what are the malwares going through a network, uh, can I quarantine a specific station based on that? Also, what you can do is if you are browsing through the log is, uh, on the fly, create some with the AI agent, some donuts, historical bar, Ja, representing all the events, all the security events that did flow, uh, through your network.

So really like here, giving you, uh, I would say, uh, a full vision of the network from, uh, the day-to-day to the troubleshooting to the security aspect. So James was talking, so more specifically, what helps is doing at Fortinet, uh, we have in-depth monitoring, uh, with historical trends, network visualization, like others, we have a lot of troubleshooting tools on the day-to-day for admins to go and, uh, and do some elements. But also we have the, uh, a IP, so we saw the, uh, gen AI part, uh, but we also have some forecasting on the land.

We have some forecasting on the SE one side to predict how the SLA is going to be a over time, uh, give some insights on the issues, give some insights on the, um, on the root cause and the remediation to really help, uh, the admins, uh, to troubleshoot issues in the network. And if you remember last year for the ones that were there, we did present you the, uh, uh, FortiGate as a service of the wireless controller as a service that Fortinet was offering. Uh, we did progress with that, where we still have this offering, but we wanted to also give the management piece as a service means that, uh, bringing fertilizer, bringing 40 manager and by the end of the year bring also 40 AIOps in all data centers, meaning that we would give the choice to customers to use 40 points, uh, to use points and to rent the hardware for the time they want, uh, rent, uh, based on the data center, wherever the D devices will be deployed.

So we started, uh, by example, AOPs in, um, in Chicago. And then we will grow over time based on what customers want and based what on what customers has meaning that you offload everything from the customer is having the hardware, not having via deck to have a huge giant vm. Because when we talk about AOPs, we talk about terabyte of data, gigabyte of ram.

So it start to be quite challenging from customers to start and spinning like a half a terabytes of ram, uh, with a single VM by example. That's for the biggest one. So what we have is we have the hardware for them to consume.

We do the RMA, we do everything, so they just have to pay with points and they can get it for the time they want. Are your SaaS offerings direct to customer then, or are you still through the channel at That point? Uh, I would say generally that is going through, uh, channel, Through channel still, yes.

Okay. Yes. Are there any partnerships, uh, like AWS marketplace where they can, uh, look so Well, this, um, this like hardware has a service is more in our data center, so I'm going to cover that, uh, a bit further.

We have quite, uh, a lot of data centers, uh, but uh, this product can be also a VM and, uh, is proper is listed into G-C-P-A-W-S, uh, and things like that. So it's up to customers to decide either a VM on-prem, uh, either, uh, public cloud or hardware or hardware has a service. So we give the flexibility to customers.

Uh, I was two weeks ago, I, I, I was in, uh, in a, b, c with a customer that said that it didn't want cloud, uh, it wanted to have everything on-prem. So it depends on what customers want it through. The trend is people want to go more to cloud, but some prefer to host everything, uh, in their Premises.

Yeah, no doubt. Multi-cloud, I have to ask. Uh, Chicago Q3.

Yes. Uh, what's that? Uh, Uh, it's aop.

So, uh, as of today the Wallace controller, uh, as a service. So the FortiGate that we did present were here last year, if I remember right on. So they are, uh, they are already in multiple location across the globe.

Uh, we have the manager on the legs office that are also in different DC and we are starting with AOPs just in Chi Chicago, but then we'll expand to a lot more, uh, data centers and also in Europe, because I'm French, so I need to represent myself. So Thanks Alex. I I do have one question about that.

So when, when you're talking about the, for day AI ops kind of offered as a service, yes. Is this gonna be like a single landing portal experience for users, or is it there gonna basically have cloud-based access to hosted FortiGate for manager for analyzer as separate? So It, it, it all depends.

So bear in mind that, uh, our wireless controller or lack of FortiGate can be hosted, uh, as hardware or even the vm. Uh, so by example, what we do see is we have a lot of customers having on-prem devices, but they don't want to have the deck to host all the management portion. So either they go to AWS or here, it's a way to just have the management piece as a service in our data centers while they still have their own, uh, devices, uh, on premises.

Okay. So it, so you, there would still be just separate instances or, or logins for each of those products that are SaaS based? Well, that will like consolidate all the 40 gates across the globe for these customer.

So it will be on the per customer basis. It's not like a, a pure SaaS offering. So it's not like what you would deck tech, sorry, with others where they offer a SaaS, uh, platform and, uh, multi-tenant and like each customer is having its own view.

Uh, fortunate approach is you own your data and your own your equipment regardless of what you do, uh, ever in the cloud or, um, on premises. But, but you'd still have a for analyzer as a service login and afforded manager as a service login. So you're still gonna have separate portals to hit those.

So yeah, so okay. So that's what we are working towards. So we have, as, uh, James was saying, we have a fabric, so we are building the fabric to interconnect those products and to get data.

So that's coming, uh, by September where we would have data from, uh, AOPs, uh, fed directly into 40 manager and more to come. So we are, uh, towards getting a unified vision for our customers. If it was the question, sorry, that was, thank you.

We didn't get it. So, but yeah, we do have a connector that is planned between 40 merger and AOPs by September. Uh, and a lot more to come down the road.

So yes. Okay.