Tim: 0:12

Hello and welcome back to another episode of the Cables to Clouds podcast. I'm your co-host this week, Tim. I'll be running the show. With me as always is Chris Miles, my co-host. See, we're co-hosts, not host and co-host, but co-hosts. We're we're equal. We like equality. Very equal. And uh with us is, and you might have you might guarantee you might recognize just from the laugh, but if not, uh with us is uh Chris Williams, who has been on the show before, and uh we had we don't have him back often enough, but uh he's he's also a very busy person, so it's probably good that we don't.

Chris Williams: 0:51

So everybody's glad to be here. Wait, is is my laugh like super unique in the annoying way? Or like you can recognize that brain donkey laugh in the background.

Tim: 1:06

If it's like that, then mine is right there with you, probably. No, no, man. Maybe I just uh maybe I've just heard it uh lots of times and I've go I've grown used to it where I can recognize it.

Chris Williams: 1:18

Is that Fran Dresher? Gee.

Tim: 1:21

Mr. Sheffield.

Chris: 1:24

Watch first unveil a new laugh uh on the next episode of eBrow.

Tim: 1:29

That's all right. It's like, man, it's gonna get uh concerned and start changing it up. Uh anyway. So Chris, uh obviously we've had you on before, and God knows you've got your own podcast and everything. But for the you know, five percent of the people that might be new to this podcast and not and not have been there, please intro yourself.

Chris Williams: 1:48

Hey everybody. Uh my name is Chris Williams. I'm the DevRel manager over at HashiCorp. Uh I also am the co-host of the podcast V Brownbag, a technical learning podcast uh that's been around for a little bit.

Tim: 2:00

And one of these days we might actually get on that podcast. We've been planning it for like a thousand years, but we've never had it line up. So we'll get there eventually.

Chris Williams: 2:09

Right as soon as we get off the off the as soon as we cut, we're booking you.

Tim: 2:12

All right, fair enough. Fair enough. Um, okay, so uh this episode, honestly, this episode's coming out a little past reInvent. We we usually try to time it if at all possible so we can get the episode uh about reInvent out fairly shortly after the show. Um but I I actually don't remember where it falls in the calendar, Chris. I don't think it's immediately. But anyway, so Chris and I unfortunately could not go close enough. Close enough is the answer. Chris and I could not go this year, which was sad. We both have started different jobs, and I got in too late to uh to join this time. Um and yeah, so but that's okay because Chris Williams in AWS Hero has been many times, including this one. And so he's gonna help. Yes, he took notes. He's gonna help us recap and go over some of the just the really cool stuff and maybe some of the not cool stuff, uh, whatever came out of reInvent. So I we saw all the news, but we won it straight from the uh the the horse's mouth, so to speak. So Chris, lead us off, man.

Chris Williams: 3:14

It is a donkey laugh. I okay, I'm I'm gonna go. I did not say donkey.

Tim: 3:18

I did not say donkey. I thought about it.

Chris Williams: 3:23

There's there's a there's a trend here, Tim, and I don't like it. All right. Um that's fair. AWS reInvent 2025. What what did I learn from there? Other than a massive head cold, which I'm currently rocking, uh, the the day quill is keeping me afloat right now. Um, so I actually got in a day early. I got in on Saturday this year because I wanted to, I was, I was helping to live stream the Road to Reinvent, which was they got rid of midnight madness. They decided that stuffing your face with a record-breaking amount of chicken wings at midnight, the day before the show, the day before the conference starts, isn't necessarily a good idea. Maybe somebody got hospitalized. I don't know what could have happened. So instead, they are now having what they call midday madness. So they do the Tatanka, the chicken wing challenge in the middle of the day. And around that, they also have something that is really cool. They started it this year. Um, Brooke Jameson, I believe, was the one that started got this going. So you know it was going to be good. They took five buses of people to run a hackathon in the bus on the road from LA to Vegas. Um, they wired up all the buses. They actually pre-ran the route to make sure that they had internet access the entire way. So it was it was fairly clever how they did this. And they piled all of these people, these these 10 teams, four, four people per team, into these five buses, staggered them out throughout the bus so they could all work on their own hackathon projects. And they did um a six-hour-long hackathon on the road as as they were driving to reInvent, towards the road to reInvent. They were then given 60 seconds to deliver their sales pitch for their what was supposed to be useless product. So that so the the the rule was you got six hours to build a useless product that leverages AWS services and AI, because everything has to have AI in it now. Of course. Um they got there about five o'clock. At six o'clock, they gave the pitch to Swami and Jeff Barr. So Jeff Barr and Swami were the judges on stage, and each team came up, gave their 60-second pitch. There was a big gong. So, like they had the gong show where once the 60 seconds were up, they were gonged off and they had to they had to get off the stage and and get the next people up. And they, the winners, won $100,000. Oh my gosh. Holy crap, man. So the the winning team actually had five people in it. Each person got $20k cash. Wow, that's it.

Chris: 5:50

I don't know if it was cash, but they got, you know, and yeah, or it was not in compute credits.

Chris Williams: 5:57

I assumed, I assumed that it was gonna be in compute credits. It was not, they gave it to them like in coal hard cash. I don't know what like tax implications were as a result of that, or if they were just gonna uh shave it down. But yeah, they walked out of there with with money, so they paid for the bar tab for the rest of the uh week. Oh, that's awesome. Each oh, so each team also had an AWS hero. Uh, I believe they also I I believe a couple of them had multiple AWS heroes. They also had multiple community builders and user group leaders in there as well. So they came up with some very, very clever, um, useless hackathon products. The the uh winner was was actually really, really clever. It so you know the um the AI enabled IDE that that AWS built called Kiro. K-I-R-O. Okay. The winning team built Oric. O-R-I-K, Kiro's evil brother. Now, what you do with Auric is you put in a short script or something, like a three-line, four-line script, and it turns it into this massive multi-repo, multi-folder monstrosity with tons of libraries, and and it just totally obfuscates everything and it turns it into this behemoth. Um, and of course, uh AWS would like that, so they won. Oh my gosh. You mean it consumes how many resources?

Tim: 7:13

Sold. This is the winner, right here. The the social commentary implications of the whole thing of taking the you know, of taking something simple and then turning it into spaghetti, you know, and using, of course, leveraging AI to do it is not lost upon me.

Chris: 7:30

Yep. That is a surprising uh I guess manifestation of what a useless product is. Uh I would I was expecting something much on much smaller scale, like uh an app that tells you how to tie your shoes every morning or something like that. I don't know. That's that's pretty much very clever things.

Chris Williams: 7:49

What uh another team called the Lou Finder, Michael Walmsley was one of the heroes on that team. It took a look at your face, it did facial recognition to figure out how urgently you had to use the bathroom. And then it found in real time, depending upon how urgent you had to go, further and further away bathrooms from you. Uh so the more you had to go, the further away the bathroom was. But you could pay a service fee to get the bathroom to maybe come closer.

Chris: 8:20

Oh, okay. That's okay.

Chris Williams: 8:22

That's that's pretty good too, actually. Diabolical. Um Mike Fiedler's team built a infrastructure building app based upon uh glyphs. So like like Harry Potter wand waves or or uh kanji script. You could you could scribble something into a pattern and then uh it used AI recognition and stuff to like figure out okay, this is this is the service that I would build with it. And then it would build it would actually build out the environment with multiple services. Um there were they they made several very clever, useless things. That's that's just a slash slash evil who's who's set aside, right? Diabolical, all the things. Yeah, so I mean the vibe from reInvent was uh as always like super energetic. I mean, the the hallways were always crowded. All of these sessions, as soon as they came on the catalog, they were immediately sold out. Um it was it was impossible to get pre-booked for a room, but you could invariably always like walk up and get in on the on the walk-ins because people would book out everything but then not show up for it. So you could get in via the walk-ins. Parties were were phenomenal as they always are. Um, I spent the vast majority of my time either doing recordings um or talking to customers or spending time on the expo hall floor catching up with friends and doing some like little mini recordings for V Brownbag, catching up with my co-hosts and all of the people that I've worked with in the industry that you know this is our big friends reunion kind of hoo-haw get together. And um, yeah. So there was a lot.

Tim: 9:58

Did you did you uh did you get to meet uh Hiroko Nishimura? I that was somebody who I never got to I never oh, brilliant.

Chris Williams: 10:06

So so Hiroko pinged me before we got to reInvent and she was like, Hey, would you be my reInvent buddy? Um, she had only been one other time and she didn't actually like go and do anything. So she wanted to like, you know, have have the experience of somebody that had been seasoned and and could could walk her around and everything like that. So like every morning we we got together and made sure that she got into like the the proper buyouts and got her badge proper, you know, just just all of the things. Uh she she confided in me that you know, she's one of those people like the first time she has to do something, she's nervous about it and a little bit trepidatious. So now she's a seasoned pro. She was she was wandering all around uh Vegas.

Tim: 10:41

Yeah, she's great. She asked me if I was going, and I was like, uh, I'm sorry, I'm not gonna make it this time. Someday, someday we'll uh she's such a gentleman. Something we'll meet up lover.

Chris Williams: 10:48

And I actually introduced her to Keith Townsend and she did a um a lightboard piece with Keith along with uh Calvin Hendricks Parker.

Tim: 10:54

Oh, very cool, very cool.

Chris Williams: 10:56

Yeah, and Keith lightboard there. Keith and Melissa. Um, so so they have the oh he's gonna shoot me. It's called the Advisorium. He's he's uh he's divested himself from the CTO advisor uh brand. Uh and now he's working with Kelsey Hightower on a new effort, and he was he was there in the in in the auspices of the new effort. It was him and Melissa, and they brought their light boards out and uh, you know, got the suite up in the top of the Venetian like he always does, and invited people up to to do different sessions. So I did a session with him, Calvin did one, and Hiroko did one with him um on that on that particular morning. But he'd been recording like a madman, like he always does.

Tim: 11:33

Oh, that's crazy.

Chris Williams: 11:34

Gotta respect the hustle. Yeah, totally.

Chris: 11:37

Um, yeah. All right, go ahead, Ted.

Tim: 11:39

So did uh did they um I remember last year I was so annoyed because all of the restaurants at all of the venues were like staked out by all of the big vendors. It's worse.

Chris Williams: 11:51

It's worse now. Every restaurant has a buyout. The entire Venetian palazzo walkway, um, all of the things down near the waterfalls, they're all bought out now. You can't you can't you can't you can go to the food court and you can get a bowl of ramen or you can get a bowl of noodles or something in one of the open air areas. Yeah. Uh but yeah, if if you want, if you want to go, if you remember going to the sugar cane back in the day and getting those uh shrimp toasts, that's that's not happening anymore.

Tim: 12:19

Well that last year uh Chris and I we found this uh Chinese place called Asian Noodle, which is like the most basic like, but that place is awesome. That place is legit as hell.

Chris Williams: 12:29

It's legit awesome. It's right right past the uh the black tap on the right hand side, yes.

Tim: 12:33

That's right, yeah. And speaking of someone who's been to China a few times, like their food is legit. Like it's the real deal. So but for whatever reason, and thank God for it, they never sell, they never you know agree to stake out or sell out or whatever.

Chris Williams: 12:46

I think it's because it's too open air, like it's it's too it's too open to the area. Yeah, it's like on the floor. Close it up. Right, exactly.

Tim: 12:54

Good call. Yeah, we I told I told you we took uh Andrew Brown there last year and had the uh the Don Don noodles, and they did not he didn't they didn't apparently didn't didn't work well for him, but yeah, it was a good time. Poor Andrew.

Chris Williams: 13:08

He's got a very delicate northern Canadian stomach.

Chris: 13:11

He was sensitive to a specific ingredient, and I can't remember what it was, but it yeah, probably the mala.

Tim: 13:16

It was probably the mala. The the spice that uh numbs your numbs your tongue or whatever. Like probably I think it was that. But anyway, no, I was cute. I was just curious because again, we didn't get to go and I I was just I remember just thinking how uh frustrating and annoying it must have been for everyone.

Chris Williams: 13:32

Yeah.

Tim: 13:32

That like every restaurant is completely booked out. Um, so that's I'm sad to hear that that's the case.

Chris Williams: 13:40

I I do I do live in the rarefied like hero place where I have like a badge to get into half of those places. So for that's fair. It's for for me, I mean it wasn't it wasn't a problem, but I absolutely empathize with everybody else that's walking around going, Hey, how come I can't go into Buddy V's and get a slice of pizza? What's up with that? Oh, is that they I'm sorry, go ahead. What I what I try to do is I try to like not sneak people in, but I can plus one people into into venues. So like I make sure that you know any any friends that I'm with are all fed uh uh appropriately as I go. It's the Greek in me. I have to feed people.

Tim: 14:14

The community builder team, uh, is that where it was? The the Buddy V's uh again? Yep, yeah.

Chris Williams: 14:18

Yeah, buddy Buddy V's was bought out, sugarcane was bought out, Chico was bought out, the cut was bought out, um uh yardboard was open. I so you you could go in you could go into yardboard and uh actually get a meal. But um gotcha. Every everything along that one palazzo strip uh right where the Venetian connects with the Palazzo, instead of banging right when you go straight, all of that was bought out. It was like like everything was gone.

Tim: 14:41

Yeah, no worries. So what about what what was your what did you see there that was like not AWS on the floor that was interesting? I'm very curious because we walked around last year and there were so many vendors like showing off different interesting products. I mean, not obviously other than Hashi Corp, of course, but like just anything that struck your struck your fancy that on the Expo Hall floor that struck my fancy that was not AWS affiliated. It's okay if not. I was just curious because we there were so many last year, just random like products and just things everywhere.

Chris Williams: 15:13

There was a lot of cool stuff, I I will say. Um, but the one sticker that I got that kind of like encapsulated every most of the people was this right here.

Tim: 15:22

Yeah, slap that on every single thing in AWS.

Chris Williams: 15:27

Their product with plus AI, even when it absolutely did not need it or was not necessary. Um so so these act these stickers actually came from Corey, uh Corey Quinn. Uh he had a give me some of those.

Chris: 15:42

But for the audio listeners, he just held up a sticker that just says now with AI.

Chris Williams: 15:46

Um with the with the cartoon pizzazz uh look around it. Um it's some of the cut some of the companies have a product where it absolutely makes sense to add AI to things. Other companies, my my own included, my own company included, um, we had folks coming up to the vault to um the Hashikor booth, and they were like, hey, so um what are you doing with Vault with AI? I was like, excuse me? The secrets manager? And and so I I I kind of I mean, it's kind of like the de facto question now, like what, okay, you got your thing. Well now how are you adding AI to it? It's kind of like the new reaction because it's being beat over our heads with it. And and I invariably would make them take a step back and go, what do you want a very confident hallucination machine to do with your secrets exactly?

Tim: 16:39

Yeah, it's like you have to walk them through the Socratic method of like have the introspection to ask yourself what we're really asking for.

Chris: 16:49

Well, I wonder, I wonder if people have gotten to the point where they're asking it in the up like kind of in the opposite direction. Like it, it's it's a trick question in that when they say, like, oh, what is what are you doing with AI with this product, they want you to say fucking nothing. There's no there's no there's no AI in it. It's already perfect.

Chris Williams: 17:08

Yeah. Well, I'm I mean, in in respects of Vault, there are a couple of things that you could use from like a query perspective, like, you know, how many how how many things are out of rotation, or just give me a brief synopsis. Don't don't enumerate them, just give me like a wide swag at something. Or yeah, you could you could do it for that. But if you're putting in a key and you want a value back in in encryption, you don't want something just making stuff up on the fly.

Tim: 17:33

Um or keeping it, keeping that data somewhere else. Yeah, exactly.

Chris Williams: 17:37

Incorporating it into a vector metal basis. Um dangerous. So so there was a little there was a lot of that back and forth. I mean, in the in the keynotes, um or do we want to get into the keynotes yet or do we want to get into the end of the phone? Yeah, yeah, I should. I should, yeah. Okay. Um, day one keynote, Matt Garmin said, Hey, I'm gonna take the keynote and I'm going to uh download it, put it into Adobe Premiere. Ironically, I'm going to use AI to clip out everything except all of the times that he says AI. I'm gonna glom them all together and I wanna know how long that lasts because it's gotta be 10, 15 minutes of just him saying the words AI. It was a two-hour talk. It la the first hour and 50 minutes of it was AI. The last 10 minutes, he basically speed ran everything else, all of the other services, the 90 the what, 90% of the humanity and services and people and effort and time that goes into AWS that isn't AI. He speed ran that in 10 minutes. And if if I'm a member of those teams, I'm I'm feeling a little bit deflated that yeah. Just don't care, right? And here's the rest of this crap. Yeah, here's DNS, which you know might have been an issue recently. Here's some improvements that we've made to it. Gosh, thanks, US East One. Yeah, um, yeah, so there were a lot of cool, cool um announcements. Um, they they announced that Graviton, uh, they they've got new flavors of Graviton that are coming out that are going to save five times as much energy as as the previous ones, you know, for like different types of queries and things like that, which is very, very cool. I mean, we're all concerned with you know giant data centers powering AI, consuming all of the energy of you know small towns in America. If if they can sip energy instead of guzzle energy, that that is a good thing. Invariably, uh, the sticker kind of raised its head a couple of other times in the keynote. Um, they have they have AI outposts now. So so remember the outposts, the the the racks that you could rent and you could like they were AI, they were AWS outpost services. Well, now that now they have those specifically for agentic workloads.

Tim: 19:43

It's like okay, so like GPU as a service or something, basically. Yeah.

Chris Williams: 19:47

AI outposts. Um, I I can't remember what they were calling them. I uh AI factories or something like that. No, they're basically uh they're trying to they're trying to cover all of their bases from the actual silicone itself all the way up to layer. Eight consumption uh and and everything in between. So everything up all the way up to agents. Excuse me. Yeah. Yeah. And then then they said AI a lot.

Tim: 20:10

Yeah. I mean, I don't I don't know where the this factory term comes from, because I've heard it elsewhere as well. Like Cisco has a a partnership with NVIDIA. They're they have compute, essentially compute, Cisco compute, like UCS, with NVIDIA GPUs, and it's called an AI factory. So I don't know where this whole factory terminology is uh coming from, but we have multiple vendors that are rolling out these this concept of factory, which is really compute plus GPU either as a service or some kind of siloed thing.

Chris Williams: 20:37

Which feels which feels very reminiscent to the old um uh triple company stacks of email firmware store. Yeah, yeah, the the the V Blocks. Yeah. I I I it's been so long I forgot the name, but do it's it's interesting that what's old is new again. Um I was just talking about this with uh DiscoPosse and Rob Hirschfield over in KubeCon. How recently they're solving the NUMA architecture problem with uh with cross-RAM, cross-cpu memory access for containers, the problem that we solved 12, 15 years ago for VMs. Right, for virtual machines. And and yeah, so I'm I'm watch, I mean, I've been around, we've we've all been in this industry long enough now to like watch the cycles of, you know, oh, here's the new silver bullet and it's gonna solve everything. Oh, it didn't solve everything. Oh, here's the next new silver bullet, and it's gonna solve everything. Oh, it didn't solve everything. So I am taking my notes. I am uh thinking about this judiciously. I'm using the Socratic method to ask the the right questions about how these things are gonna be properly implemented. And then and then I'm stepping, you know, I I still get excited about it. Oh, um I was in the I was in the Matt Garmin keynote uh on v Brownbag at the in about the middle of the year, we did this heroes on Kiro episode of V Brown Bag where I had six or seven AWS heroes on, and we all built something using Kiro and then kind of showed show and telled it on the show. AWS found out about that, and so they interviewed all of us for a little blurb for Matt Garmin's Kiro Kiro segment. Um, and they played the video in the background. It was hilarious because like they interviewed me for like an hour and a half, two hours, and literally maybe one and a half, two seconds of of uh video made it onto the onto the keynote stage.

Chris: 22:23

Um I was like, your signature laugh make it into the video?

Chris Williams: 22:27

It it didn't, it didn't my signature donkey braid did not make it into the segment. Um effectively, everybody was just clipped long enough to say, Kiro good, yay, and then move on to the next person. Yeah, nice.

Tim: 22:41

That's well, if you'd stopped saying the F word so much, they probably would have given you a longer clip. You know what?

Chris Williams: 22:47

I was I was dropping F bombs a lot. Maybe it was something like you can go from zero to POC and ten times as fast or something like that. I I I was I was saying something uh true, something that I felt was true. I wasn't I wasn't trying to like you know gloss it up or shine it up or anything, but it was hilarious because they were like, okay, you've said zero to POC a lot, Chris. What let's talk about production. Uh and I was like, I would never use Kiro for production. Are you kidding me? I'm like, okay, let's concentrate on POC a little bit more. All right, let's do that.

Chris: 23:16

Well, that's I guess I'm assuming that this probably happened. Look, if I guess for those that are listening that don't know, Kiro is um, keep me honest here, Chris, is kind of AWS's developed um kind of a native app with AI integration uh IDE, um kind of similar to this cursor or something like that.

Chris Williams: 23:35

So cur cursor's a fork, uh VS Code, Kiro is a fork of VS Code. The the new IBM Bob is a fork of VS Code.

Tim: 23:43

Yeah.

Chris Williams: 23:43

Um Kiro's uh claim to fame is that you can uh basically do oh they they have they have a word for it. Uh hold on, I'll just all fired up real fast. Specify. Yes, thank thank you. Spec spec driven development. You start you start with the spec, you have a conversation with Kiro, you it asks you the questions, it writes the spec out, you then modify the spec. Once the spec is done, it then builds things out based upon the spec, which is is how you know project managers and product managers and architects do all the things. It's like a PRD. It's like a PRD, basically. Yeah. Exactly, exactly. So it it works, it works great. It works as great as any other agent, a AI platform that hallucinates and makes things up and and can be can be frequently wrong. Um I use it, I use it for testing, I use it for writing really good README's, I use it for, you know, doing scaffolding to like, you know, get rid of some of the the undifferentiated heavy lifting of building out an application. Um so yeah. Well it's funny, it's funny that you say.

Chris: 24:53

Um was the these the these three kind of fleet of frontier agents that they launched with the um Kiro Autonomous Agent. I don't know how directly integrated that is with Kiro or if it's just kind of kind of keeping the namesake, um, and then the security agent as well as the DevOps agent. So um the Kiro agent, I think, is kind of meant to be that kind of autonomous, you know, working 24 hour a day type developer that's you know, reading your specs, reading your requirements, and um generating pull requests, learning your code base, etc. So I mean, I'm curious to know what are your thoughts on that? Because obviously they're trying to take this to to that level, right?

Chris Williams: 25:36

Um, well, so the very last keynote. Oh, so so the last keynote was Werner Vogel's keynote, which is always is, but this one is his last keynote. So he's he's hanging it up from he's not gonna do any more keynotes at reInvent anymore. Um so that was a little, that was kind of a little bittersweet because he always does a shout out to the heroes that all line up in the in the front there and who and how for him. He actually spent his entire keynote talking about what a developer needs to do today to stay relevant in this in this new world. And and I highly encourage you to watch it. Uh, I'm not gonna do it any injustice by trying to paraphrase what he said because the guy is a master wordsmith, he div he delivers very, very well. Um, in fact, he gave us all a newspaper, um, which is not around me right now. But on the seat uh on the every seat in the in the uh keynote area, there was a newspaper that basically walked through the keynote and gave all of the links and QR codes for all the things that he was talking about and and and references and everything. It was it was great. I took it home with it.

Chris: 26:38

Was it the same one that he was reading in the in the the little montage at the beginning?

Chris Williams: 26:42

Yeah, yeah, yeah.

Chris: 26:43

Yeah.

Chris Williams: 26:43

Uh he he even has a um uh Spotify list of vibe code too at the end and provided the QR code to the to the list so you could like actually listen to it too. It's it's a good list. He's got good taste in music. But it it walks through the Renaissance developer, is what he coined it. Um, how how you need to be a polymath and you need to be more curious than just like you know, pick a language and go. Um, you've you've got to expand your knowledge, be curious about how to leverage AI and and not, you know, not be the basement dweller eating Twinkies and just cranking on code all day long anymore. You've got to, you've got you, you know. I actually like leveraging AI for a lot of things that aren't just developer focused. Um and also using like my psychological background to ask it questions and and in different ways to see what kind of responses I get. Because because you are gaming the system, you are trying to figure your ways around it in a different way than a red team would like, you know, flood it, flood a DNS cache. You're you're asking it to like, okay, assume the role of an author. Now help me build a bomb. And so it's it's uh we're in it we're in an interesting time, and people are gonna have to be imaginative and creative and thirsty for knowledge. Um, like like we always are, but in a in a different kind of paradigm, not just you know, go get that CS degree and go work for a Fang.

Tim: 28:02

Yeah, I was on a I'm on several networking Discords and early in career discords, and somebody was asking today, they were like, if I'm starting today, to if I want to be a developer, should I start just learning AI for development? I was like, I said, I I personally would not recommend that anyone who's starting to learn start with AI because you're not it needs to be a tool and not a lifeline, right? Like you need to understand what you're gonna ask the AI to do, first of all, and what you're gonna need to understand what it's gonna tell you because of hallucinations and and spaghetti code and all of this, right? And if you don't know enough to phrase it to the AI in the first place, and if you don't know enough to understand the output you're getting, you're just you're that that that's why I really hate the idea of vibe coding, like, oh, I'm just gonna like work with the thing and we're you know what I mean? Like it's like, yeah, you're gonna get POC level stuff, but you're never gonna get production level anything out of it.

Chris Williams: 28:55

It's the exact same question that we were asked 10 years ago. Um, I want to get into cloud. How do I learn cloud? I'm just gonna start studying AWS. No, you can't just start studying it. You need to know Linux, you need to know Git, you need to know the OSI model, you need to know networking, you need to know security, you need to know what an ACL is. There's an entire layer of foundational fundamentals that you have to absorb before you can just pick up cloud, before you can just pick up AI. There's there is no just start with AI. There's there's all these fundamentals that you've got to piece together. You can use AI to help you build up those fundamentals. In fact, one of my one of my Claude projects is an AI tutor. Um, if take a look at my LinkedIn profile, take a look at my resume, take a look at my my blog posts, tell me where my knowledge deficits are and help me build a curriculum to beef up those knowledge deficits around coding, DevOps, and artificial intelligence. I'm on it, Chris. Here you go. Here's seven weeks of things starting from day one, spending an hour a day learning all these things. Boom.

Chris: 29:57

Um It's a great point. I think I think it should be like using it to help build a learning plan to learn the things that you need to do is is is a great uh great output that you can get from it. The the I think the problem is the shortcuts, like the quote unquote shortcuts, have always been there for people to kind of display this level of knowledge quickly when they when they start at zero. But it's never been this kind of like immediate, right? There's always been, you know, if we think about like tech certifications, there's always been dumps and things like that for you to like, you know, find it out and and then take the test and have this, you know, kind of badge to display, um, even though you don't have the knowledge backing it. But now it's just like within a few minutes, like I can put something together that looks like I know what I'm doing vaguely at least.

Tim: 30:44

Just just passes the sniff tests, but that's about it, right?

Chris: 30:47

Yeah, it's hard. Like the the the wealth of knowledge that you need at the foundation layer is growing. And you know, I know it's probably hard to not take the shortcuts. I I don't know where where I'm going with this. All I'm saying is that I sympathize with people getting into this career field now because it's it seems like the the pool of what you need at the base layer is is probably twice as big as the when we started.

Tim: 31:11

Oh, and you got the grifters telling everybody that the only way to learn and they're gonna be replaced, and this is AI is everything, right? Anyway, 100%.

Chris Williams: 31:20

And I mean, this isn't easy. This is hard work. Uh the the the the things that we've gone through in our careers, the blood, sweat, and tears, you know, we've bled into sharp raid cages. Uh those those old uh I know I have the the the Packard Bell raid boxes with the sharp tooled edges. Yes, we've literally bled on motherboards for this for our craft. Um cage nuts. Those those are the kinds of scars that you know you you can't you can't just you know do an AI prompt and figure that stuff out. You've got you've got to have you've had to have been there. And humans like the easy way. I'm I'm just like every other human. If if there is an easier path to do something, I will invariably try that first and then, okay, well, that didn't work, let me go the hard way now. Um so I I can't I can't fault anybody that takes a look at an AI agent, puts, puts in a prompt, gets back the answer to a lead code problem in three seconds, and then tries to, as long as they try to like reverse engineer that in their heads and understand the fundamentals of why that was why it was, and not just regurgitate it or memorize a brain dump or something like that, then that is that is a new mechanism for learning.

Tim: 32:30

Okay. Not to not to not to turn this uh episode into a you know AI uh Shakespeare stuff.

Chris Williams: 32:39

Exactly.

Tim: 32:40

Although the the parallels, we we talk about this several times about the parallels between the hype cycle and the the Gartner cycle, if you will, which is funny because the Gartner makes the Gartner cycle. Anyway, the the whole the whole hype train and then the Trophavage dissolution and all of that. Gartner fuels that whole thing. So it's funny that they have the exactly that they actually have it all mapped out. Uh but this was what we did with the cloud, you know, 10 years ago. This was the same hype cycle. Everything put all your stuff in the cloud, you're gonna save millions, like you know, like the same type of message.

Chris Williams: 33:12

That's how Gartner makes their money. They they perpetuate the cycle so they get more money so that people can pay to be in the magic quadrant. That's they they've got that nailed. It is a 100% a racket. That's why I never pay attention to the Gardner reports.

Tim: 33:25

I was t I was telling somebody uh just today, actually, I was like, uh, Gardner doesn't sell advice or uh what's the word, analysts, uh whatever, analysis. Gardner sells uh CYA to the C-suite.

Chris Williams: 33:38

Yeah. Access and CYA.

Tim: 33:41

So anyway, not to not to take this thing entirely off the rails.

Chris Williams: 33:45

Yeah, probably.

Tim: 33:46

This is gonna be one of our most controversial episodes. We're gonna get sued, it's gonna be taken down. Uh that's okay. That's why you're bringing on Chris. Oh man, this is ugly too. Oh none of us are gonna get our job offers from Gardner, I'm afraid, but uh oh well. Unfortunately. Oh okay. So um I'm trying to think. Any other any other insights? Like what was the like is there anything that you would wrap it, you know, like crunch it up into and throw it as like a paperball at somebody if they if they were like, what is what was your takeaway from reInvent?

Chris Williams: 34:23

Um AWS is still um doing very well for itself. They're they're full steam ahead. Um they they've made an a couple of missteps, uh, in my estimation, around missing the message. Uh but they they recover as as fast as a giant behemoth CSP can recover uh when when it you know when the chips are down. Um I will be at reInvent next year. I will be eagerly waiting to see what what they're gonna say next. So obviously they're not burning any bridges or pissing anybody off to the extent that you know we're all gonna stop using AWS tomorrow. So yeah, it it there wasn't anything that was like, holy ma, you know. I was actually talking about somebody about this. Six, seven, eight years ago when you went to a reInvent. Oh jeez. Oh no, I said it. 12-year-olds are giggling.

Tim: 35:12

My 12-year-old, my 12-year-old would be on the floor right now if she had heard that.

Chris Williams: 35:18

Oh, I apologize to anybody listening on the podcast with a child in thereby. Um, eight or nine years ago, when when reInvent was, you know, still young, we would go to reInvent and we would be jazzed. We would come home and immediately fire up the console and start playing with the thing that was announced or released, and we would be like, ooh, and on. And we would spend a week like completely energized and and wanting to research the things and and look into the stuff. I don't feel that anymore. I get home and I, you know, I do my recaps and we we do things like this and we and we dissect the things that were said and the things that were not said, but the the uh sense of urgency and the sense of like magic newness isn't there anymore. It's uh now it's like, oh yeah, you you better have X, Y, and Z. That's that that's literally table stakes.

Tim: 36:06

Table stakes, yeah.

Chris Williams: 36:08

Um I I think they're trying to use AI as the new sparkle. Uh I certainly know like, you know, that the the AI companies are desperate to make keep that sweet VC money aflowing. Um, but it doesn't feel as sparkly as you know when when you could first swipe that credit card and just build a data center without having to plug in a single cat five cable.

Tim: 36:28

Yeah, that's fair. It's a different type of hype for the AI versus the cloud. But yeah, I totally am following you there. Um so the one thing I forgot to mention I forgot to mention earlier is there was a ton, and we'll have a lot of this. We're gonna put this in the show notes, some links and stuff for the specifically for the networking stuff. What do you just real quick, what's your take on the uh expansion? Uh rather I should say finally the CSP is lining up and and investing in multi-cloud connectivity, like from the CSP rather than having somebody as a third party and and then kind of pretending the other clouds don't exist.

Chris Williams: 37:04

Um well, my heart goes out to the companies that have made their fortune on building out things like that because for the longest time the company the CSPs pretended that nobody existed. There were there was there was a long time you couldn't even say the word multi-cloud on an AWS page, um, which was absolutely silly. I mean, they they it's it's silly to just pretend something doesn't exist. Um that being said, it's about damn time. I mean, why why has it taken this long? Every every large company is an amalgamation of several smaller companies that one started on EWS, one started on Azure, one started on GCP, they bought they bought into each other and now they gotta figure out how to make all this stuff work. Okay, well hit large CSP, since you are so customer focused and customer obsessed, make this easier for me. Oh, we're not gonna do that because Azure doesn't exist. Oh well, okay, fix that then.

Tim: 37:57

Oh well, uh right. And all CSPs have always operated as we are the one-stop shop and you don't need to go anywhere else, right? Which ignores, of course, things like mergers and acquisition, all the things you just talked about, right? So it's I'm curious, I'm very interested to see if the CSPs are gonna follow through with it. I think a lot of customer pressure is is is why this is happening. So I it's good. Like you said, it's about damn time. And I'm also curious uh what happens to the companies that you know that's their product is is building this connectivity.

Chris: 38:28

So yeah, I've been talking to a few other individuals as well that were there um talking about this, and it sounds like this is much more than just kind of dipping a toe in it as well. Uh, this is uh a bit more serious in that capacity, and there's probably more to come. Um, probably other cloud providers are gonna get announced at some point. Um uh no one no one's promised anything, but it sounds like things could be headed that way. Um, and if the pricing can be competitive, then I think it really is. Yeah, it's gonna be uh it's gonna be a tough, tough uh few years for, you know, if we think about certain companies that, you know, the company Tim and I used to work for, uh, even the the colo and data center providers like Megaport, Equinex, things like that that make a lot of money on intercloud connectivity. Um, we could see quite a shift um from that where there's you know ultimately no physical infrastructure for for some of these companies. So yeah, it could be quite uh quite the sea change.

Chris Williams: 39:27

On the on the other side of that, there's also how much ink is going to be bled onto paper by these large CSPs trying to figure out who's gonna get their 20 pieces of silver from rare from from like you know, get getting the m like who's who's gonna get the the money when the pi the the packet is here? Okay, now who's gonna get paid when the packet is here? Okay, now who's gonna get paid when the packet there's gonna be so many lawyers up so many other people's keisters on on this one.

Tim: 39:53

Yeah, uh I I don't envy them on that one for sure. Especially when you talk about okay, well now we're because all this is gonna happen and Kind of meet me like a digital realty or wherever the circuits are there's gonna be a uh an LOA and a cross connect and a a cable at one point, you know, one to the other, and it's gonna be that it's probably you're right, you're probably gonna get to that point where they're like, where do I start taking over? Where do you start? And all of that hammered out.

Chris Williams: 40:17

Maybe Megaport and Aviatrix have a little breathing room before uh before it gets really dicey.

Chris: 40:22

I'm just I'm just thinking, think about managing SLAs when the two parties involved are both CSPs. Like, oh my god, that sounds awful.

Chris Williams: 40:32

I mean, I mean, even now when they when they shit themselves, they don't pay out anything to the customer when they talk to SDS one. So, you know.

Chris: 40:41

Well, yeah, maybe if you're on the customer side, if you're consuming AWS, yes, but if if two CSPs have noticed lost data in between, they're gonna get their money from somebody. Yeah, who's on the who's on the bridge? Who's on the bridge is on that one? Um, because they're that's true enough, they're gonna get their money back for sure. Um whoever has the most lawyers wins. Yep.

Tim: 41:03

It's probably yeah, it's probably true. All right, we'll go ahead and uh wrap that. Uh thanks.

Chris Williams: 41:09

Thanks for a cheerful episode of Tables to Class.

Tim: 41:14

Hey, I mean a little realism never hurt anyone.

Chris: 41:17

Yeah. Well, as as Tim said, there were there were a few other very interesting networking adjacent announcements, and and if you want to read those, we'll we'll populate a list in the show notes. There's some interesting stuff around ZPC encryption, uh, some interesting stuff with uh Route 53 now being usable over, which is which is interesting. Um so yeah, we'll put all that in the show notes if you want to kind of look at a technical deep dive into some of those things. Um yeah, I guess Chris, we should probably also give you an uh an opportunity to plug yourself. Where should people come and find you? No, but yeah, you gotta do it now.

Chris Williams: 41:53

I gotta do it now. Okay. Uh if you Google my handle, Mistwire, M-I-S-T-W-I-R-E. Uh, I am the first six or seven pages of Google Hits. Uh I'm the co-host of V Brownbag, um, and you can find all of that if you if you pop in Mistwire, you'll see you'll see all of the the blog, the the podcast, the show, all the things.

Tim: 42:14

All right, man. Sounds good. And thanks for joining us today. I appreciate it. Oh no, it's my my pleasure. No, no, I'm gonna go take a night call. Yeah, yeah, yeah. That's fine, man. Get yourself a hot tidy and a blanket and uh one of those big comically large hot water balloons for your I've got I've got my giant EKS took just in case I need to uh there we go.

Chris Williams: 42:33

There you go. Just in case my head gets a little cold.

Tim: 42:37

Nice, nice, nice. All right, so yeah, so this is Caleb's clouds. Uh if you enjoyed what you heard and or saw, or both, if you're really weird, um and uh share us with a friend. Uh, you know, uh just get us out there. All right, everybody. We'll see you next time.