AI Slop Is Everywhere

0:00 AI enabled, AI powered, AIdriven, AI

0:02 enhanced, AI slot. I'm getting tired of

0:05 this. This is what we're seeing with

0:06 every product we use since Chat GBT

0:08 broke everything in 2022. Now that

0:11 marketing's got a hold of it, you know,

0:12 is it AI? Is it glittery machine

0:14 learning? Thank you, Alan. Yes, every

0:17 industry, including networking, is

0:18 slapping on that AI sticker because

0:20 that's all investors care about right

0:21 now. Everyone's jumping on this AI hype

0:24 train, and I want to get off. I mean,

0:27 come on. Is adding AI to all of our

0:28 products actually solving real problems?

0:31 Is it making us better? And specifically

0:33 for IT and network engineers, are we

0:34 seeing real results? That's what I want

0:36 to figure out in this video. Is this all

0:38 just AI slop? Or are there real use

0:40 cases? Like, as a former network

0:41 engineer, I want to see this happen.

0:43 Maybe the network is up all the time.

0:45 Maybe I don't receive a call at 3:00

0:47 a.m. And while AI is kind of scary for

0:49 the job market, we're also kind of

0:50 begging for the things it's promising

0:52 us, like this Reddit post here. Where is

0:54 all this AI ops stuff that's supposed to

0:56 help us? But first, get your coffee

0:57 ready. This is a journey. And my journey

0:59 begins by attending the largest

1:01 networking conference in the world,

1:02 Cisco Live. With over 22,000 people, I

1:05 thought, surely I can find an answer.

1:07 But what's funny is the interesting

1:08 story didn't actually happen there.

1:10 About a mile and a half away, 10

1:12 minutes, another event was held by

1:14 another large networking vendor, Juniper

1:16 Networks. Now, Juniper heard I was going

1:18 to Cisco Live and they said, "Hey, shut

1:20 you know all this AI stuff people are

1:21 talking about, we're actually doing it

1:23 and we want to show it to you." They

1:24 also said something about making

1:25 networking sexy again. Now, it's no

1:27 wonder that networking has become sexy

1:30 again. So, I'm like, "All right, let's

1:32 do this." But seriously, thank you to

1:34 Juniper for inviting me to their event

1:35 and helping make this video possible. By

1:37 the way, this video was made prior to

1:39 the recent news about the HPE

1:41 acquisition of Juniper. You can check

1:42 out the links to the press releases, but

1:44 I just wanted to give you that bit of

1:45 context before we continue. All right,

1:47 let's take a quick coffee break.

1:50 And here we go. I arrive at the Juniper

1:53 event just after Cisco announced all

1:55 their major AI stuff. So AI was already

1:58 kind of top of mind and I'm like man

2:00 how's Juniper going to beat this? But

2:01 then Sudhir said this today across the

2:03 street you heard a lot of announcements

2:06 AI canvas AI this agentic AI all kinds

2:09 of stuff. We are not confused about what

2:12 we are trying to build and who we are

2:14 trying to be. We are trying to build the

2:17 industry's first self-driving network.

2:20 Period. end the story. So I was like,

2:22 okay, they're not worried. They're not

2:24 scared. But then I heard something

2:25 crazy. Rammy, their CEO, said, "We've

2:28 been leading the AI and network

2:30 convergence for more than 10 years now."

2:34 Uh, excuse me, sir. Chat GBT came out in

2:37 November 2022. There was no AI before

2:39 that, right? That is kind of how it

2:41 feels, but AI was very much a thing

2:43 before Chat GBT. Back then in the early

2:45 to mid2010s, it was not about generative

2:48 text or chat interfaces. It was about

2:50 machine learning and it was more about

2:52 being very focused at being good at one

2:54 thing so it could find patterns and data

2:56 make predictions and decisions and

2:57 that's exactly what the founders of Miss

2:59 Networks did. Miss Networks, who's that?

3:01 I'll make a connection. Don't worry,

3:02 it's coming. The founders, Sujay Hija,

3:04 Bob Friday, which is an awesome name,

3:06 sounds made up, and Brett Galloway, all

3:08 former Cisco employees by the way,

3:10 founded Miss Systems in 2014. Uh, we

3:12 were responsible for Cisco's wireless

3:14 business. We did the acquisition of

3:16 Moroi in December of 2012 and it

3:20 inspired uh the creation of Mist with

3:23 the goal of reinventing enterprise Wi-Fi

3:25 with cloud computing powered by

3:27 artificial intelligence AI. And this is

3:30 2014. Now the main difference with their

3:32 AI versus what we see now is that it was

3:34 trained very specifically on wireless

3:37 LAN data. What a good network looks

3:38 like, what a bad one looks like, how to

3:40 troubleshoot things. Very specialized.

3:42 You couldn't ask it how many Rs were in

3:44 strawberry. They didn't care about that.

3:45 All it cared about was Wi-Fi. They

3:47 coined the concepts AI for IT or AI is

3:50 in the air. Kind of cheesy, but I like

3:52 it. And tell me if this sounds familiar.

3:54 They wanted to deliver a self-driving

3:55 network. One that could proactively

3:57 detect and adapt to issues in real time,

3:59 saving it teams money and time. Kind of

4:01 sounds like what we're talking about

4:02 with AI now, except this was over 10

4:04 years ago. Now, were they successful?

4:07 Um, yeah. They were purchased by Juniper

4:09 in 2019 for $45 million. See the

4:13 connections made now. This became part

4:14 of Juniper and Juniper's like, "I love

4:16 your AI stuff. Let's make it part of all

4:18 that we're doing." Now, think about that

4:19 for a second. Juniper has had AI in

4:21 their portfolio, their stuff since 2019,

4:24 miss system since 2014. The thing I'm

4:27 wondering is, does that matter? Like,

4:28 for what we have now, does that 10-year

4:30 head start make a difference? That's

4:32 what I want to figure out. Now, one of

4:33 the co-founders, Bob Friday, again,

4:35 amazing name, said this about grapes or

4:38 AI. I make a barrel of wine. I always

4:40 tell people, you know, great wine starts

4:42 with great grapes. You know, AI has the

4:44 data. So, it's like getting getting your

4:46 arms around the data is usually the

4:47 first step on the journey to mastering

4:49 AI. Now, one thing I've learned about AI

4:51 so far is that it all depends on the

4:53 data you feed it, the context you

4:55 provide. This is true across the board

4:57 for anything. You can ask an AI like

4:58 ChatB. Hey, what should I have for

5:00 dinner tonight? And it might say

5:01 Hawaiian Sunrise Pizza with pineapples

5:03 and cashews. Perfectly fine answer. But

5:05 it may not have the context of you that

5:06 you might hate pineapples and you're

5:07 allergic to nuts. Context and data is so

5:10 important for AI. In the same way, we

5:12 can't ask an LLM about our network

5:13 unless it has the context of our

5:15 network. So, how are we giving it that

5:17 context? How is it learning about our

5:18 vast, massive, complex networks? Well,

5:21 what we're seeing right now is a lot of

5:22 bolt-on AI again slapping that sticker

5:25 on. We're feeding our LLMs a ton of data

5:27 about our network, telemetry data from

5:29 our routers and switches, packet

5:31 captures, and just everything. Here you

5:32 go, friend. Figure it out for me,

5:35 please. And that's kind of what we're

5:37 seeing with vendors like Cisco. Now, I'm

5:38 going to pick on Cisco first because

5:39 they are one of the largest networking

5:41 vendors. No hate, I love Cisco. But

5:43 right now, they have a lot of things to

5:44 pull data from. They've got thousand

5:46 eyes, which when you really think about

5:47 it is a scary, scary name for something.

5:50 Makes me think of one of those angels

5:51 from Revelation. You know what I'm

5:54 talking about. It's watching you. And

5:55 they use it for monitoring their network

5:57 stuff. So, they're getting telemetry

5:58 from that. They also have Splunk,

6:00 Catalyst Center, App Dynamics. Now, it's

6:02 not important for you to know what all

6:03 these are. I just want to point out that

6:04 they have a lot of things, a lot of data

6:06 pouring into an LLM. And while they do

6:08 have this really neat DPM or deep

6:10 network model, specially trained on all

6:12 their networking stuff, they are relying

6:14 on its ability to take all of that data

6:17 fed to it and just kind of figure it

6:20 out, which this might work, but it's

6:22 still pretty new and I've only seen a

6:23 demo. But if you've been using AI for

6:25 any amount of time, you know how tricky

6:27 it is to maintain a large context and

6:29 make sure the AI doesn't like

6:30 hallucinate. Now, this right here is how

6:32 a lot of vendors handle adding AI to

6:34 their products. Here's all the data and

6:37 just please figure it out. Juniper,

6:39 because it had a head start, doesn't do

6:41 it this way. And if that looks a little

6:43 bit gnarly and complex with a lot of

6:46 hardware and software and clouds that

6:50 you have to stitch together, maintain,

6:53 upgrade, say a prayer when you do it.

6:57 You're not alone. Now, Juniper, because

6:59 they've been doing this for a minute,

7:00 they've structured things a bit

7:02 differently, a lot differently. And the

7:03 question I want you to have in your mind

7:04 is, does that matter or does this

7:07 matter? What I'm about to show you. Now,

7:08 Juniper claims that they are AI native.

7:10 They've been building their platform

7:11 from the ground up with AI in mind. And

7:14 for me, I was like, okay, what does that

7:16 mean to be AI native? And again, it all

7:18 comes down to data. Now, real quick,

7:20 it's time for a coffee break, and I want

7:21 to tell you a scary story. Get ready. As

7:24 I was leaving for this trip to go to

7:26 Cisco Live and visit Juniper, I'm

7:27 walking out the door about to catch my

7:29 flight. Boom. Power outage. No big deal.

7:31 Power comes back on. But you know what

7:33 didn't come back on? My two Proxmox

7:35 servers. Both of them just would not

7:36 power on. Running on those servers were

7:38 my Twin Gate connectors. I was

7:40 devastated because Twate is what I use

7:42 for remote access for my entire team

7:44 while we're traveling and we were about

7:45 to do that very thing. Now, I was in a

7:47 pinch. Didn't have time to troubleshoot.

7:49 I can only think to do one thing. In

7:50 just a few moments, I was able to log

7:51 into another computer at the studio and

7:53 deploy another Twing connector. And just

7:55 like that, I had connectivity back to my

7:57 studio from wherever I go. And I was

7:59 back up and running. I could remote into

8:01 everything I have except for my Proxmox

8:03 servers, obviously.

8:05 But seriously, that's one of the reasons

8:06 I love Twin Gate is it's simple but

8:09 powerful. You can deploy it in moments,

8:11 anywhere in your home network. It's free

8:12 for home labers or as a business owner.

8:15 Man, they make it so easy, but also it's

8:18 zero trust. It's not your dad's VPN

8:20 where it's just a big tunnel that

8:21 everyone can access everything. No, I

8:24 can allow my editors to connect back to

8:25 my NAS, but not to my AI server. I can

8:28 restrict the ports they're able to

8:30 access on a server. I get detailed logs

8:32 of who's accessing what. Seriously, if

8:33 you're not using Twate for your remote

8:35 access, try it out. Just right now, try

8:37 it out. It's not going to hurt you. It

8:38 literally take you 5 minutes to deploy

8:39 it. And honestly, it's the best remote

8:41 access solution I've used. So, shout out

8:43 to Twing for helping make this video and

8:44 my trip to Cisco Live and Juniver

8:46 possible and for being one of the main

8:48 sponsors of my channel, helping me do

8:49 what I do. So, please show them some

8:51 love. Check it out. That's the end of my

8:52 scary story. Proxmox servers were still

8:54 down until I got home, but that's a

8:56 story for another time. Back to learning

8:58 about AI stuff. Now, keeping in mind,

9:00 data is the most important thing with

9:02 AI. Juniper's data is stored inside the

9:05 Mist AI cloud. The entire context of

9:07 your network in one spot. wireless APs,

9:10 switches, routers, clients, the Zoom and

9:12 Teams APIs, the config, the state, their

9:15 performance, the quality, all that

9:16 telemetry is being consistently fed to

9:18 the Mist AI cloud. And since the

9:20 beginning, they've had this

9:21 microservices architecture. So, think

9:23 Kubernetes that will ingest all that

9:25 information, figure out what it is, what

9:27 it means, how it all relates to each

9:28 other, so it can do a better job at

9:30 identifying problems, predicting

9:31 patterns, and zeroing in on the root

9:33 problems so you can fix them faster. But

9:35 hold up, what about the data center? I

9:37 didn't forget about him. Juniper also

9:38 throw all their data center stuff up

9:40 into the Miss AI cloud. But the way they

9:42 handle that data is interesting. They

9:44 don't just throw it up unstructured

9:45 hoping the AI will figure it out. They

9:47 have an intentbased networking

9:48 technology called Abstra. And it's built

9:50 around this, their contextual graph

9:52 database. Inside this database is

9:54 everything you need to know about the

9:55 current data center, how it's connected,

9:57 all that's in there, how all the network

9:58 elements and data points are related to

10:00 each other. And it's that database that

10:02 is thrown up into the Mist AI cloud.

10:04 This is everything about your network,

10:06 the current state of it. I'm talking RF

10:08 stats from your APS, cable metrics,

10:10 optics, what's the current state of

10:12 spanning tree, what about BGP and OPF,

10:14 how's the jitter on Zoom calls? And they

10:16 have these things called Marvis minis.

10:18 They are AI native networking digital

10:20 experience twins. These little guys are

10:22 like virtual clients. They're spun up

10:24 automatically by the Marvis AI. And they

10:26 learn your network via unsupervised

10:29 machine learning. And like legit, they

10:31 are clients. They will authenticate.

10:33 They will get IP addresses. They will

10:35 hit your DNS servers. They'll get your

10:37 SAS applications and part of their job

10:39 is to map client journeys. Essentially,

10:41 they're always trying to simulate what

10:43 your clients are doing on the network

10:45 and they will try to find anomalies or

10:47 issues before your users do. This

10:49 happens all the time and especially when

10:51 you make configuration changes. These

10:53 little guys go in and like make sure

10:55 that you didn't misconfigure a VLAN or

10:57 the DHCP pools messed up. And you may

10:58 have heard of the concept of an SLA, a

11:00 service level agreement where someone

11:02 like your ISP might guarantee you a

11:04 certain amount of uptime. Juniper has a

11:05 thing called SLES or service level

11:08 expectations, the expected performance

11:10 of an application or experience. And

11:12 it's these little guys that help kind of

11:14 monitor that pretending to be clients.

11:16 Now, sitting on top of all this is

11:17 Marvis AI. And this would be more what

11:19 you would think machine learning is.

11:21 It's specialized. All it cares about is

11:22 networking. Eat, sleep, breathes

11:24 wireless data centers. This is what was

11:26 running the Marvis minis and doing all

11:29 the unsupervised learning. This sucker

11:31 knows your network. And on top of that,

11:32 we have the Marvis AI assistant. Just

11:34 keep in mind the assistant at the top is

11:36 not trying to figure out the network.

11:38 That's already been done. It just knows

11:39 how to ask Marvis AI good questions.

11:41 Translating your human speak. Now, the

11:43 star here is still the data and how it's

11:45 handled in the Miss Cloud. You might be

11:47 thinking, "Okay, Chuck, so what? One

11:49 place. Why does that matter?" Well,

11:50 let's do a scenario real quick. Let's

11:51 say the CEO of Hackwell Industries,

11:53 Bernard Hackwell, calls in. He said on

11:55 Friday at 3 p.m. his Zoom call was

11:58 terrible. He says, "Tell me why and fix

12:00 it." He's a CEO. You got to do it. Now,

12:02 real quick, think about before AI how

12:03 you would answer that question. For me,

12:05 I'd be like, "Oh, um, tell you what,

12:08 CEO, let me know when it happens again.

12:10 Give me a call and I'll come like see

12:12 what's going on. I'll do some packet

12:14 captures or something. Try to help you

12:15 try to correlate some stuff with my

12:17 human brain." Now, with a solution like

12:18 Cisco where you have a lot of different

12:20 sources you might have to pull data

12:21 from, you would just tell the deep

12:23 network model your problem. And I'm sure

12:24 it's going to do a great job in

12:25 understanding what you're describing,

12:27 but it's going to have to go out to all

12:28 these different sources, pull telemetry,

12:30 to pull data to figure out the context

12:31 of what's happening with your network

12:33 and what happened in that moment. And it

12:35 might actually find a really good

12:36 answer. Contrast that with Juniper and

12:38 how they handle data. And at this point,

12:40 I'm not saying if one or the other is

12:42 better. I'm just contrasting how they

12:44 access data and the context of your

12:46 network. You would tell Marvis the same

12:48 thing. Here's my problem. But it

12:50 wouldn't have to go out to a number of

12:52 sources and try to figure it out. No, it

12:54 would simply go boom to Marvis AI and

12:56 Marvis AI looks at the current state of

12:58 the network. It would already have the

13:00 context that Bernard Hackwell's laptop

13:02 was connected to AP03 which is connected

13:05 to switch one and it's a router for I'm

13:07 just making stuff up. Just trying to

13:09 move fast here going out ISP2. This

13:11 would already be the context. Accessing

13:13 that context would be like this. And

13:15 there might already be a node created.

13:17 I'm not going to go into the

13:18 architecture of the good graph database.

13:19 We don't have time for that. But this

13:20 node might be a quality score known as

13:23 MOS for that Zoom call of like 2.4,

13:25 which is bad. And that was already

13:27 correlated with a switch port on switch

13:29 one that had a very high increase of CRC

13:32 errors on port 1024. And so instead of

13:34 the AI coming up and saying, "Oh, we

13:36 found a bunch of errors on all these

13:37 things." It goes, "No, no, no. We found

13:39 the uh problem. We think it's going to

13:41 be a bad patch cable between AP03 and

13:44 switch one port 24. The recommended

13:46 action is to replace that cable." And

13:48 what I want to hammer home here is that

13:49 it didn't have to go out to any other

13:50 data source to find this information. It

13:52 went to one spot, the Mist AI cloud, and

13:54 it already had that client journey

13:56 mapped out. And let's say they do go out

13:58 and replace the patch cable and they

13:59 mark that ticket as done. Then a little

14:01 Marvis Mini might be spun up to test out

14:04 that client journey to hit DHCP to hit

14:07 DNS to simulate that Zoom call and then

14:09 come back with a good MOS score and go,

14:12 you know what, it's good now. Now I was

14:13 talking with Kyle from Juniper and he

14:14 was telling me about how Marvis will be

14:16 able to predict cable failures or optic

14:19 failures before they become a problem.

14:21 So we have enriched data that we talked

14:23 about from the graph database that has

14:24 all the relationship and the context

14:26 information of it. So let's take a

14:28 example of a common data center issue an

14:30 optic failure and we can start looking

14:32 at trend analysis and that's where AI

14:35 and and machine learning models come

14:37 into play to predict with a level of

14:39 confidence that this optic is going to

14:41 fail in a couple weeks. You need to

14:43 order a part now before it goes bad and

14:46 impacts your applications that you are

14:48 providing as a service. Now, at this

14:50 point, it's like, okay, all this sounds

14:52 really cool if it works, right? Like,

14:56 we've heard promises. It looks shiny.

14:58 I'm excited. I'm also like grain of

15:01 salt, you know? But Juniper was pretty

15:02 bold in saying, you know what? Ask our

15:04 customers. We already have this stuff

15:06 deployed in our campus and wireless

15:08 stuff. Anybody wearing a green lanyard

15:11 today is an existing Juniper Mist

15:13 customer. If you call on

15:15 anything I'm saying, they're right here.

15:18 they they will not lie to you. They are

15:20 the same as you. They will they will

15:22 carry the story for you. Speak to our

15:24 customers. And I'm like, "Okay, let me

15:26 talk to one of them." So I talked to

15:27 Allen. And while Allan is not a customer

15:28 directly, he does work for a partner,

15:30 Nexum, and they deploy Juniper Stuff to

15:32 customers. So I work for an integrator.

15:35 And so I work with a lot of different

15:36 customers and their their level of

15:39 technical ability is varied. And it's

15:42 been a huge help for them cuz they can

15:43 go in and say, "Hey, Marvis, who's

15:44 having a bad day?" and it says, "Hey,

15:47 these people are probably having a bad

15:48 day. Here's what you need to look at."

15:50 And with clicks, you can investigate and

15:52 they can drill down and say, "Oh, hey,

15:53 this is the problem they're having. It's

15:55 because they need to install this patch

15:56 on their laptop." Turns out Allan is

15:58 actually from my Discord. I didn't

15:59 realize this. So cool to get to meet

16:00 you. And I asked him some questions

16:02 about his experience with Juniper's AI

16:04 or Mist AI. So, it's able to look back

16:07 and for all the clients, it's keeping

16:09 client state um you know, hundreds of

16:11 pieces of client data every minute. And

16:13 so I can go back and when the CEO calls

16:15 and says, "Hey, last Thursday at 3 PM I

16:18 was in this building and my wireless

16:19 dropped for a second. What happened?" I

16:21 can back up to that, I can look at it

16:23 and I can say, "Okay, here's what

16:24 happened." What that's allowed me to do

16:26 is now offer those services above and

16:27 beyond. So I've gotten a lot more into

16:30 programming, a lot more Python

16:31 scripting, that kind of stuff to add

16:32 value. Um, which is where the AI stuff

16:35 comes into play. So I gave the example

16:36 here, very demo-ish example of the CEO

16:38 call. If you want to ask why the CEO had

16:41 a bad Zoom call last Tuesday, you can

16:44 actually do that. I asked Allan like,

16:48 Can this really be done right now with

16:46 Marvis AI? And then are we at the point

16:50 now to where it can actually fix things?

16:52 Like fix it before we even realize it's

16:54 a problem. Marvis will never send you in

16:56 on wild goose chases. We've gotten to

16:58 that point in maturity where where we

16:59 can start to trust it to make more

17:01 decisions. So, okay, here we are.

17:03 Juniper is saying and their customers

17:05 are saying the same thing that AI is

17:07 here in their platform and they're using

17:08 it and it's awesome. The way they've

17:10 integrated their AI has been more AI

17:12 native. They're not new to the space and

17:14 the way they deliver the context or the

17:16 data of the networks of their customers

17:18 is the unique proposition here. We're

17:20 already at a point with Juniper's

17:21 portfolio with their enterprise campus

17:23 networks, wireless networks where

17:26 self-driving autohealing networks are

17:28 here on the data center side that is

17:30 coming soon according to Kyle. So where

17:33 our ultimate vision is going is we want

17:36 to get to self-driving you know I think

17:39 we're we're you know close to it um and

17:42 we'll get there but uh but that that is

17:44 the direction we want to get to and I

17:46 think we're we're near that and that

17:47 that possibilities are very close on

17:50 what we can do um and obviously we'll

17:52 start with the right use cases um and

17:55 and be able to provide that that

17:56 confidence and and still that that you

17:59 know it's going to work all the

18:00 detection and troubleshooting stuff is

18:02 built right in. And if you look at other

18:03 vendors right now, let's pick on Cisco

18:05 once more. Right now, they're slapping

18:07 on the AI sticker and they're trying to

18:08 get all their legacy components to work

18:10 together with AI, dumping a lot of data

18:13 into an LLM that I'm sure is amazing and

18:15 it probably does a great job at

18:16 correlating all that data. But it's

18:18 still a lot a lot of moving parts. And

18:21 when if you've been in it for any amount

18:23 of time, you know, the more moving

18:24 parts, the more complex, the more often

18:26 it breaks and the harder it is to

18:27 troubleshoot. And let's be honest, we

18:29 know LLMs, even the greatest ones right

18:31 now, get very confused with too much

18:33 data. So, I find Juniper's approach very

18:35 interesting. Now, I have not played with

18:37 it or used it myself, experienced an

18:39 outage or anything like that, but their

18:41 customers have. And conceptually, it

18:43 sounds cool. That one single place, the

18:45 single pane of glass that we always talk

18:47 about, the unicorn in it, seems to exist

18:50 in Juniper's world. And it's not just

18:52 Juniper stuff either. They do have

18:53 support for third-party routers and

18:55 switches, including every other major

18:57 vendor. they can interface with their

18:58 APIs or even legacy monitoring

19:00 protocols. So, what do you think? Do you

19:02 think we're at a point now to where AI

19:04 is finally delivering on the promises?

19:06 Do you think it's all marketing glitter

19:07 shiny tools or is it real? Maybe you are

19:10 a Juniper customer and you can tell us

19:12 in the comments below like what your

19:13 experience is like or maybe you're using

19:15 other vendors and they've implemented AI

19:16 in an amazing way and it's saving you so

19:18 much time. I would love to hear that

19:20 story. Also, I would like to know if

19:21 this concerns you because let's be

19:24 honest, this is starting to automate and

19:26 do parts of a network engineer's job,

19:28 which at first sounds scary, but then

19:30 you think about the job of a network

19:31 engineer. I don't do the job now, but

19:33 back in the day, I remember my main

19:34 stressors were the network goes down. I

19:36 have to figure it out fast. And that

19:39 stress is insane. Or you get that 3:00

19:42 a.m. call. You're just groggy and you

19:44 need coffee, but you don't have it. and

19:46 you're trying to figure out a very very

19:47 complex issue at your worst state. If we

19:50 can avoid those situations and bring an

19:52 AI to bring better uptimes, that's going

19:54 to be phenomenal. Anyways, thanks to

19:55 Juniper again for inviting me out to

19:57 their event. But whatever the case, my

19:58 goal for this video was just to show you

20:00 what the landscape is right now. It was

20:01 kind of a deeper dive into Juniper

20:03 because I got a chance to go dig into

20:05 what they have. I only got a brief

20:06 keynote from Cisco. And right now, Cisco

20:08 stuff is still pending. We should see it

20:10 later this year, but Juniper stuff has

20:12 been out for a minute. And honestly,

20:13 this stuff is kind of exciting. Anyways,

20:15 that's all I got. I'll catch you guys

20:16 next time.