0:07 Good morning, good evening wherever you are, and welcome back to the Enterprise AI Show. 0:11 This is your host, Aaron, and today we have Brian and Brandon ask what happens to enterprise AI if the market cools off and the era of token maxing gives way to hard questions about ROI. 0:26 Brandon argues that the real divide is between people who have used AI to finish something that once felt impossible and everyone else. 0:36 He makes the bold claim that nobody really writes code anymore. 0:40 And Brian compares this moment to the early days of cloud when Netflix showed us what was possible and most companies couldn't imagine getting there. 0:52 And all of that is coming up right after this quick break. 1:00 Today's show is sponsored by Nord Security. 1:02 Your employees are using AI every day, but do you know which tools they're using, what data they're sharing, or where their AI agents are connecting? 1:09 You can't govern what you can't see. 1:11 Introducing NordLayer AI Scout, the visibility layer for enterprise AI. 1:16 AI Scout gives you visibility into AI activity across web and desktop apps, detects sensitive data exposure, helps control AI agent connections, and creates an audit trail for governance so you can protect your data without blindly blocking innovation. 1:30 See what your AI is doing, then decide what happens next. 1:34 Connect with Nord Security to learn more about NordLayer AI Scout. 1:38 Visit nordlayer.com/ai-scout/enterpriseai and get started today And we're back. 1:51 And Brandon, good to see you again, man. 1:53 How was your weekend? 1:54 It was fantastic. 1:55 Happy to be back 1:56 Good. 1:56 Good. 1:56 We are recording this on November 9th. 1:58 This will go out as the November 11th show. 2:01 Most of you are probably going, "Oh, I know what they're gonna talk about. 2:03 They're gonna talk about that thing." We are gonna talk about that thing but we're going to hold it for, The show for Sunday will be, we don't know whatever date that is, the 13th or 14th, or… That'll be the cloud news of the week, the, or the, AI news of the month, Patch Low issue. 2:15 We will talk about that thing, and Brandon and I were just talking about it offline. 2:18 That might be a two-hour thing that we talk about the thing. 2:20 But we're not gonna talk about the thing today. 2:22 The thing we're gonna talk about today is there is there is some perspective in the market right now that, we might see some sort of whether you wanna go extreme and call it bubble burst, whether we have some sort of correction in the AI market, whether we have some deflationary thing. 2:36 There's a lot of stuff, whether it's, people looking at the An- Anthropic S-1 or some of the data center stuff that came out of the election, which we'll cover next, the next show, whenever, that maybe there's some slowing down that might happen with AI. 2:47 And you and I have had this conversation for a long time 'cause you really just went all in on AI, and you went as far as to be like, "I'm gonna get certified on stuff. I'm curious about doing forward engineering." You're building projects and stuff. 2:59 And, th-this is… I think we're at the stage where people are gonna start having to, I don't wanna say rethink AI, but they're gonna have to figure out the difference between the last few years, which has had a lot of move fast and break things, the, the token maxing conversations, the AI first conversation, all this, and, what typical enterprises then have to go through as far as ROI. 3:22 I'm curious, you're looking at this, you're all in on AI, man. 3:25 How are you looking at people-- how are you thinking about the stuff that you're doing, where you're utilizing it all the time, versus people that you see aren't necessarily utilizing it all the time or aren't getting the full benefit of it? 3:37 Yeah. 3:37 The thing, the big disconnect I see is, there's people that really, let's call it the normal people, who've maybe only ever used ChatGPT and just a the chatbot, right? 3:46 And then there's people, a lot of us in the middle, who have maybe dabbled in maybe they made a small script, maybe they did one or two things. 3:54 And then there's let's say at the frontier model approach where it's like, no, the goal of AI is to basically do everything, right? 4:02 Where if you look at a, a presentation by an anthropic engineer, they're usually talking about "We're building systems to build stuff for us." So they're more, if you will, it's like we're creating the system to then create something that will build whatever it is that they ultimately want. 4:16 And I think this is where the huge disconnect in, is in the world of just AI and people talking about it, and why people maybe don't want data centers. 4:24 Because unless you've done something with AI that you've touched the, quote-unquote, "superpower." And all I mean by that is if you have had something that you have wanted to do, and it can be anything, right? 4:36 It can be, like, helping with your taxes, building a, a Mac app, reading a bunch of documents, summarizing it, creating a report. 4:43 If there's something that you wanted to get done that used to take you a very long time or a task that was so hard that you never really got around to it, it was like, it's just I wish I had that thing, but it's just too long. 4:53 I don't have time to do it." Unless you've actually taken a task like that and actually completed it with AI, usually incredibly rapidly you don't have that sense of wow, this is a set of superpowers that once everyone gets it, it's going to, if you will, turbocharge all of us. 5:11 All of the things that we wanna do in our life could generally be done faster, better with AI or some AI assistance. 5:18 And I think that is really where the massive disconnect is, I think, between those that have actually experienced that and those that maybe are on the other side, just haven't really embraced that level. 5:28 So-- And I think token maxing is a good example of this. 5:31 I think if you are a CFO and your d- your day-to-day is "I'm in meetings, I'm doing budgets," but you're not really doing anything with AI, you just see these huge bills coming back, right? 5:43 And it's not immediately obvious to you like, what what's going on? 5:46 Like, why are we paying for this? 5:47 It's not-- But I bet you to the people that are doing that, most of the time I bet you those tokens were doing something that was, like, very impressive to that person. 5:55 And I think that's, like, where we're stuck, right? 5:57 It's like people who've it's… I don't know. 5:59 I don't-- It's, I sound almost religious here. 6:00 Coming off too much like I'm crazy. 6:02 But it's unless you've done it, unless you've really felt it, it just kinda seems like yet another technology, right? 6:08 It's yeah, whatever. 6:09 It's just something else. 6:09 But it's, it really, to me, it's much more than that. 6:12 Yeah. 6:12 Yeah, I think what you highlighted, it's going to be it's going to be the, the, the next… What was the thing that, that Gartner had? 6:20 It was like the haves and have-nots. 6:21 It was the, the, the people that move fast and the people that move slow kinda Gartner concept. 6:26 God, it's been 10 years. 6:26 I forget what they called it. 6:28 Like the early adopters and the visionaries and all that, 6:30 for it. 6:31 I'll- it'll come to me at some point as we're talking about this. 6:33 But I think, we've gone through some stages here, so we went through… It first came out, there was a lot of okay, there, there's people who went out and figured out, hey, w- how does ChatGPT work and that sort of stuff. 6:43 And then we went through the, the immediate sort of corporate, okay, we really kinda want this'cause it seems cool but we're not sure what anybody can do. 6:51 And so you had the, these companies making like AI policies, and a lot of their AI policies ultimately just became like, give a bunch of money to Microsoft and let Copilot, you can all have access to Copilot. 7:00 And then it became Copilot's not that great, or Gemini is not that great, but I could get those as corporate things. 7:06 And then, people were like, "Yeah, but I wanna use Cloud Code," or, "I wanna use Codex," or, "I want whatever the…" Like we've gone through two or three phases of, what would've been, what we called shadow IT or shadow cloud or whatever. 7:17 But those were more like access issues, right? 7:20 And then we had the second tier of stuff that was the sort of, the equivalent of the Facebook move fast and break things, right? 7:26 So we, we… I don't know if it was beginning of this year or the end of last year, like token maxing became a thing, and it was just like, look, you don't understand it, just give people as much as they, they can possibly have. 7:36 Don't ever put any things on it. 7:37 But again, there wasn't, there wasn't really like a like a goal for token maxing. 7:42 It was just, look, if you just give them unlimited, then unlimited things will come out of it, right? 7:47 And then we s- you know, we saw companies where there was distinctly a top-down thing where it was like, we need to go tell the market that we have an AI policy, and we think that's gonna make our stock price go up. 7:56 And we saw we saw little bumps with that, and so forth. 7:58 But I think the, the thing that we're gonna quickly go, get into, especially if the market has any sort of like downturn or correction or right-sizing or whatever it might be, is I think we're gonna, we're gonna almost get back to where we were when cloud was first starting to take off, and folks like Adrian Cockcroft were giving their talks about,"Hey, this is what we do at Netflix, and we're able to deploy 10 times a day, and we're able to do chaos stuff." And people were just like, there was a group of people that were like, "That is the way forward. 8:26 It's, it's-- The technology now exists. 8:28 That's the way forward. 8:29 That's the way we're gonna build software applications better." And then there was a much bigger group of people who were like, 'I can't fathom that because my world doesn't look anything like that. 8:37 And to get to that, you are gonna have to change so many things." And that re- that requires policy changes, and it requires that old guy who doesn't wanna give up his job or doesn't wanna give up his control over some, simple process to do these things. 8:51 I think we're gonna go through that with AI, in which, there's going to be… I think there's gonna be three things that happen. 8:57 There's gonna be the companies who all of a sudden are like,"No, we're pulling back. 9:00 We're not doing that stuff. 9:01 It's too expensive. 9:01 I don't trust those companies," whatever it might be. 9:03 The classic you know- slow and concerned companies. 9:07 You're gonna have a lot in the middle that are trying to figure out some balance of things, and then you're gonna have, some that just go, "Oh my God, the, the market is…" Like the, it-- I, I feel like there's going to be a certain number of people, and again, this is the, the 10X engineers, the whatever, that are gonna almost have the mindset that Andy Jassy used to have, where he'd go, he would say things like, "I can't believe the competition is still not even responding to us in the cloud, and they're giving us all this room to just go run." Like he used to say "We're seven years ahead of our competition, and I can't believe they're not responding to stuff." I think it's gonna be really interesting to see how companies surface the folks that, that have experience, like even just the level of experience that you have, which is, like a year of kind of living in this sort of thing and being like, "How do we double down on that?" Because like you've got like tangible things where 9:07 you're like, "I was able to do this. 9:56 I use Claude to plan my stuff. 9:58 I don't just l- use it as an afterthought. 10:00 I don't like half-ass it and whatever." And I think that's gonna be really an interesting sort of second act to this, this thing, especially from an enterprise perspective. 10:10 And I think we're gonna see, instead of people just being like, "Yes, we gave Microsoft a whole bunch of money, and I don't really know what we did with it, but like we're AI ready or whatever," to where we're gonna start to see those, those stories and those use cases that we've been asking for the last couple of years. 10:25 And I think people are gonna be like,"Man, why did it take four years?" And you go like, you're like that's as long as it took cloud to start figuring…" I'm like, Netflix wasn't there, on day one in 2006 and 2007. 10:36 Invent in like 2010 and being like,"Yeah, we did some stuff." So I feel like it's starting to play itself out. 10:43 We just got ourselves so compressed that things could go a million miles an hour with AI, especially in the enterprise, that it's working itself out. 10:50 But I think we're gonna go through a really bumpy period here for the next year and a half just because the market dynamics are gonna get weird, and it's gonna be really interesting to see how much more bottom-up lead stuff versus leadership sees this as an opportunity and goes and finds certain people and all that kind of stuff. 11:07 Yeah I think it probably, it's one of these things that probably take longer than that. 11:09 And I think if we think about it's just human beings, like how we, adopt new ideas and how complicated it is. 11:14 And really, if we're being more honest about it, it's like, being skeptical of a bunch of technologists saying"Hey, here's the new technology. 11:21 It's great. 11:21 You should adopt it." There's plenty of evidence that this has not gone well, right? 11:25 Social media, and we won't go into all the things. 11:27 But it-- So I think you just have a, a, a world that is very primed to be like there's been a lot of promises before, and there've been a lot of unintended side effects and some broken promises, and we're not necessarily happy with what's going on. 11:42 And I think this is a large reason why data center is just a tangible thing that everyone can understand. 11:46 It's like, it's a building, it sucks power, it's in my neighborhood, I don't want it, right? 11:50 This is a very easy way to be like, "I don't want that thing." And I understand that. 11:54 I… Like, when I get it, I get where people would have some natural reservations. 11:59 At the same time, I don't know, at the risk of just sounding all Pollyanna, it's like it really feels different this time. 12:03 And as we've talked about on this show, it's like AI can be a single-player game, right? 12:08 It's like I understand changing a whole organization, trust me, I understand its fates, how imp- how at times it feels impossible. 12:15 But that doesn't mean that right now everyone listening to this can't be like,"I'm gonna automate my weekly reports. 12:20 I'm gonna make my spreadsheets better. 12:22 I'm gonna make them faster. 12:24 I'm gonna create my slides faster. 12:25 I'm gonna find ways to automate many of the things that I am doing," right? 12:30 Over and over with, one of the popular AI tools, right? 12:34 And I think it's just the willingness to be like, "I'm just gonna go in there and start trying this and doing it," because that's what's going to… back to your question about, like, where the market goes or differentiate it, just people that are willing to get in there and try it. 12:48 And I know it kinda lends itself more to the tinkerer personality type, so there's people that are just like computers, if you will. 12:54 But I just can't tell… I don't know. 12:56 I s- I feel like I… even people that I know that are, in technology some of them are even real slow, to wanna do some of this stuff, right? 13:03 They just they just wanna be like I'm going to, it's for example, someone would be like, "I'm gonna have it look at my WordPress and give me some ideas on, like, how to customize it." And I'm like no. 13:12 You're gonna just let AI build the entire new website. 13:15 Have it do the whole thing." And they're ha- and then they really push back. 13:18 They're just like my website is really good, and I want a new theme," and all this other stuff. I'm like, "No. 13:23 You're like, you're totally missing it. 13:25 Just have AI go do the whole entire thing. 13:28 And if you don't like it, you can just, you can move away from it." But it's a real… it takes a real, change in mindset around "No, that's how I get stuff done, is I just tell AI the goal, and we just work through it and get it done," versus "I put AI in this tiny little box, and it does one little thing for me, and it's maybe not that impressive." And I kinda walk away like, "Ah, that's not that impressive," right? 13:49 So Gonna be hard, right? 13:51 I guess that's my thought. 13:52 And then, one of the things you have on here, and I almost think this phrase vibe coding has really gone the wrong way. 13:57 It's like I know what people say"I'm gonna vibe code something in 30 minutes, and it'll do something." But coding has completely changed. 14:03 There is no coding anymore. 14:05 No one actually sits down, even in production settings, and writes code. 14:09 It is completely done. 14:11 And I think I think, sometimes people don't really internalize no, it's over. 14:15 It is absolutely over. 14:16 I'm not saying someone doesn't occasionally open something up, but no one is really as good at coding as the AI stuff. 14:21 And it's so the idea that even vibe coding is like no. 14:24 Y- what people almost kinda throw away this vibe coding is the best software developer you've ever worked with, I think you have to really internalize that to be like, "Wow, there's some real superpowers in here." 14:33 Yeah. 14:33 Yeah. 14:34 And I think that's it-- the, the vibe coding one is an interesting one 'cause you're right. 14:38 You're we're gonna quickly realize, the, vibe coding is going to become the equivalent of yeah, do you know somebody in your organization that's like really good at pivot tables in Excel? 14:46 'Cause it's just gonna be like, yeah, they just use the tool and they're good at using the tool. 14:49 I think the vibe coding thing, the other thing about the vibe coding thing that's gonna be interesting is I think we have this perspective that like if somebody builds an application and we're gonna put it in production, that it would've gone through like all of the, all of the things that the typical software development would do, which, the, the hard part about that is not the, "I built the thing." It's going to be like, did you build it for all the corner cases? 15:11 Did you figure out all the negative stuff and all the corner cases and the other, the last 10%, as they always say? 15:16 And then the, the real interesting part is going to be like, who's, who, who's gonna deal with sort of the security pieces of it? 15:24 Because, if you just did it and you put out a, you put out a, an MVP of it and you're like, "Yeah, it works fine for whatever we did," it still sits there as this, yeah, it, they might have done the PHP thing and then the PHP big bug comes along and it's like, "Oh shit, we didn't…" They didn't know the, what, what corporate thing we were supposed to use. 15:42 That does feel like that, but I feel like every enterprise already has that problem of there's 30, 40, 50% of stuff that you have no idea what's running in your organization, and 15:50 But even something like that, like I feel like that's another good example. 15:53 It's like, hey, like you build something, right? 15:54 Like how do… like we should talk for a second, like at least for I- how I see stuff actually happens. 15:59 Okay, so you build something and you start to get it like what you said before, the happy path done, right? 16:03 You get your application starting to do it, right? 16:05 Then maybe you show it to people like you would anyone else, beta, and then people start to give you problems. 16:09 And it's like literally I've done this, right? 16:11 People email in enhancements, right? 16:13 I'm personally doing this right now. 16:15 I'm running s- I won't bore people with this, but like I'm running a beta program right now. 16:18 People email me with feedback. 16:20 I take the email, I cut and paste the email into the agent, and it fixes the problem. 16:26 Within… I've done this in 20, 30 minutes, I'll cut a new build. 16:30 I'll go back to them and be like, "Can you try this and tell me if it works?" And s- most, a lot of times it works, right? 16:35 And then on the security side, it's so I, this is how I've tried to do stuff like that. 16:39 It's like, "You, Mr. Organization, give me all the security requirements you have. Give me, tell me everything I'm supposed to do." I, sometimes I'll fire up a separate agent from a different provider, right? 16:48 Sometimes I'll use Claude, sometimes I'll… Sometimes I'll be like, "Okay, read this whole thing. Go through, do an entire audit of all the issues," right? 16:56 Take all, and then I take that, and then I feed that back into the coding agent. 16:59 Fix all this. 17:00 And I'm like, is it perfect? 17:01 Yeah, but I'm like, this is like professional level development. 17:05 I would have never gotten a security audit like this on the software I was building, right? 17:10 And even in a large organization. 17:11 So I think it's just this idea of yeah, you can start with a nugget of a tiny idea and you… Can you… How many times can you go back to the… Every time you say to me what about this thing?" I'll be like can someone just tell me what needs to be done?" 'Cause I'm gonna now fire up an agent that's gonna go do all of that. 17:25 And then, after I've spent a couple of weeks doing that, I'll give it to whatever group you want me to, right? 17:30 And be like, "Go ahead, penetrate pen- do pen testing, do whatever you want, acceptance testing." So it's out there. 17:36 I just think it's like you just gotta keep pushing forward. 17:39 And I think that's the point where put aside all the craziness. 17:43 I have no idea if the stock market's gonna go up or down. 17:45 I wish I knew. 17:46 I would do… If I did, I would invest, right? 17:48 But I do know these tools are going to continue to make us all a lot more productive. 17:53 And the people that grab them, the, they're, the sky's the limit with what you can do 17:56 Oh, yeah. 17:57 No I think my, my biggest takeaway from this is I think we're going to, just like we saw the be-beginning when cloud took off, there were the people who jumped all in, and they were all in on it, right? 18:08 They were all in on it. 18:09 They distanced themselves significantly from the… they became the, the thing that people were like, "Oh, man, I wish we, we, I wish we could do that," right? 18:16 They were the ones who like, digital transformation was like no, we were just digital native. 18:20 We didn't have to go through all of it." And then, yeah, there were so many, there were so many variants of like but, like it's, yeah but the, you know, and… But yeah, I think you're, I think you're right. 18:29 This is as big an opportunity as ever and probably even more so than if you started this thing three or four years ago, and you were one of those people who were like, "Oh, I just vibe coded this thing over the weekend with my whatever." Like this is a, a huge opportunity that if you're already starting down this path, especially if you're like concerned about, oh is my company gonna, gonna not lay off people? 18:49 They're gonna, do that. 18:50 Th-th-there's been no better opportunity to distance yourself or showcase yourself or highlight yourself or whatever. 18:57 And if you can figure out a way to do it that's like immediately obvious to folks, you're just gonna have to, you're gonna start seeing leaders go, "You, I need you. 19:04 I'm pulling you out. 19:05 I'm, I need you to work on this. 19:06 And you, okay, you, 'cause you, you get what's possible, and you can showcase what's possible." But I think it, I think there is just… I think it's so hard for people to not have the yeah but if, but, okay, I gotta throw a pers- I gotta throw a, an old process in the loop. 19:20 I don't even wanna say throw a human in the loop. 19:22 I think it's more like I don't, how do I not automatically throw an old process in the loop? 19:27 And I think that's gonna be, it's gonna be hard. 19:30 It's gonna take some, it's gonna take some work, but I think it's something that, you know, the, the people that, that figure out ways to do that. 19:36 Even, and again, like this is, this goes back to the, the flip side of you know, AI, everything's AI first. 19:42 I don't think you have to do the sort of everything's AI first. 19:44 But I think you do look at the things, like you said, where you go,"We are just gonna apply AI to this thing, and we're just gonna let it r- do its thing, and we're gonna take feedback quickly, and we're…" Put it in places that it can succeed. 19:56 Don't try and apply it to everything where you're like, man, that's nothing but like rocks and salt and gears grinding and like people that are gonna get in the way. 20:04 Like fine, avoid that stuff. 20:06 But I, yeah, I think you're right. 20:07 And I think this just the what's going on in the market, what might happen in the market, and I don't know the timing of it, and you, like you said, you know the timing of it. 20:14 But there, there's going to be some reactions because people react to whatever. 20:18 I think this is gonna be a bigger time than, than we've ever seen to 20:20 gonna say right before we wrap up here, I think maybe we'll just build on your statement there and say "Hey, what if you replace yeah, but with what if?" What if is a great prompt, right? 20:28 For an agent. 20:28 It's like, what if I ask the agent to do X, Y, Z? 20:32 And just go put that in. 20:33 Go see what it says, right? 20:34 Start there. 20:35 Before you do the thing, what if I ask the agent to do the report? 20:39 What would it do? 20:39 Ask it if it could do it, see what it would do. 20:41 And I think that's the kind of just that switch, the switch in your mind of you may be surprised what it, what things you'll see. 20:48 So I, that would be my advice to everyone. 20:50 All right. 20:51 Let's wrap it up there. 20:51 I know we're a little tight on time. 20:52 This one's a little shorter than normal, but trust me, folks we've got some stuff to talk about coming up this this Sunday, and then we've got an Aaron-- Brandon and I just talked about a few on the prep show for this thing, that we're gonna have some, some long ones. 21:02 So the good news for that is the long ones are always useful when we've got holidays coming up, 'cause we know people like… They go on car rides with their family. 21:09 They wanna get away from the kids. 21:10 They want-- They need a two-hour one that's not just the end of year wrap up. 21:13 Oh, we've got a few of those coming for you. 21:15 A little shorter this week, but Brandon, thanks for everything. 21:17 Thanks for the insight. 21:18 Again, I think we're gonna, we're gonna see some some shifting here happen in the next few months, few quarters or whatever, so be prepared for stuff. 21:25 So folks, with that, we're gonna wrap it up. 21:26 We'll talk to you next week. 21:29 Thanks for listening. 21:30 Check us out at theenterpriseaishow.com for past shows, newsletters, and all things enterprise AI
Community discussion