The New, New Kingmakers: Agents, Developers, and the Future of Software with Stephen O'Grady

0:07What does the outside world already know about your company? Cyber criminals could be seeing leaked credentials, compromised session cookies, exposed infrastructure, and even impersonations of your brand and executives. Nord Layer Intelligence by Nordstellar gives security teams visibility into those threats across the deep and dark web, data breaches, attack surfaces, and brand impersonation all in one platform. Find out what attackers know before they can use it. Visit nordlayer.com slash intelligence slash enterprise AI and use code enterprise10 for 10 % off your Nordstellar plan.

0:54Today's show is sponsored by Palo Alto Networks. Every time your team deploys a new cloud workload or AI agent, another identity gets permanent access to your critical systems. Legacy tools were built to manage human employees, leaving modern machine and AI access largely unmanaged. That's why IDERA by Palo Alto Networks comes in. Human, machine, AI, one identity platform for all of them. IDERA replaces permanent permissions with dynamic access. so you can lock down every identity without slowing down your business. Secure every identity with IDERA by Palo Alto Networks. Visit paloaltonetworks.com slash IDERA.

1:32Again, that's paloaltonetworks.com slash IDERA.

1:41And we're back. Joining me today is Stephen O 'Grady, principal analyst and co-founder of Redmock. Welcome to the show, Stephen. My pleasure. Well, listen, I don't think you need a long introduction. I think people already know who you are. And we're going to talk a little bit about your great book, The New King Makers, and your recent response to your own book, if you want a second. But before we jump in, I thought I'd start with something kind of fun. Last I looked, I've seen on the social media, you were like a big fitness walker. You've been like walking many miles, and you have a bunch of blog posts about it.

2:09So I was going to say, are you still, I know you've got family now, priorities and stuff, but are you still getting your walks in? You still making around your little area out there up in Maine? I am. I am still at it. I got into it a couple years ago when I tried to get back into running and just kept like, oh, my foot hurts, your back, or everything's tweaked. And I was like, yeah, just walk a lot. I am fortunate to be in a profession where I can do a bunch of things. Certainly we'll get into the end a bit later, but even taking and listening to calls and all that, I can do well and walk around.

2:35So it's a good way to combine, in many cases, getting some work done with getting some exercise. Yeah, I agree. I think it's a great way to still. I always tell people, don't sit at your desk and watch the keynote. Put it in your phone and then walk around. You'll stay more involved. I did say one thing that I thought, did you walk something? And I remember seeing a picture of it was like very cold or it looked very cold. It was like snowing and you did like a long eight hour walk or something. Is this, am I remembering this correctly? Yeah, I've done walks here cold. I've done walks here pretty long.

3:04I did 50 miles one time, 60 miles another time. Like a serious walker. This is, I don't want people to be like, it's just, I'm just out like taking a stroll. Like you're doing the real thing. No, but honestly, I tell people all the time, man, if all you can do is get out and take a stroll, like great, man. Like any, any movement's good. So if you've got 20 minutes, half hour, whatever it is, recommend it. Do what you can do. Hopefully we always get notes that people are running, walking, walking their dogs about the podcast. So hopefully people are out right now. They're listening to us on a great new walk.

3:31Listen, as I mentioned at the top there, so you wrote a book in 2013, The New King Makers, which I think was like, I don't know, I think it's like a seminal book for the industry. Took a lot of ideas at the time and kind of condensed them into a nice narrative for all of us. and of course many things have changed there. So I thought maybe let's just start with, for those that haven't read the book, why don't you set it up? What was, take us back to 2013. What were you thinking about? What prompted you to write the book and kind of what's the core thesis of it? So what prompted me to write the book was that I got tired of having the same conversation over and over.

4:04And I thought, hey, if I document this in a short book, because I don't remember how many pages it is, but it's something like 18, 20 ,000 words. So it's not a sort of lengthy tome. But basically the gist of the argument was that for decades in the technology industry, largely as a function of budget and purchasing and all that, developers were essentially an afterthought. It was like, okay, I'm a CIO, I'm going to go buy some things and this is what developers are going to use. And over a period of years, a bunch of different things came together to change that, open source being one, cloud being another, there's a whole range of other things we could talk about, software as a service and education, sort of marketplaces for services and so on, that basically empower developers.

4:42And by 2013, when the book came out, the reality that we were seeing was materially different, which was basically that, excuse me, in many cases, when you looked around in the industry, the technologies were actually being used or the ones that developers wanted to be used. And we basically came up with a, I wrote the book to essentially crystallize and encapsulate that argument. Like I said, hopefully shortcut the conversations I needed to have. It makes sense. And so what do you think? What was it about this time period? Why did developers go from being, if you will, being directed to, and so this is what you're going to use to like, this is for them basically taking an active role in choosing technology.

5:18What do you think was like, what, is there one reason or is it a whole host of reasons? It's a host of reasons, but probably the sort of biggest one was open source, right? So in other words, when I was coming up in the late nineties, if you wanted to do anything as a developer, right, if I need an operating system, I needed a database, if I need, basically I want to go out and build something. The early sort of versions in the late 90s, Linux was coming up. But by and large, you would have to go and get budget from somebody. You would have to go and get budget from somebody to purchase a license for an OS, purchase a license for a database, and all that.

5:51And what ended up happening was that with open source, they didn't have to do that anymore. And all of a sudden it was, all right, I want to go build something. And there's no sort of commercial restriction of me doing that. I can go pick up a quality-free operating system, a quality-free database. Languages, obviously, were free to acquire. and even tooling. Developer tooling for years had cost a ton of money. And so anyhow, you had this situation throughout the early 2000s where developers were empowered in ways they just fundamentally hadn't been before. And then in 06, what is it? In March and August of that year, S3 and EC2 launch kicking off the cloud wave.

6:28And all of a sudden, not only do I have free tools or free software, the free as in beer software, all of a sudden I have hardware that is affordable. I can go run a cluster for 20 minutes or an hour or a day, and we're talking in dollars. That might cost me 20 bucks. I don't need to come up with thousands of dollars to go out and get physical hardware. So anyhow, like I said, I think the biggest thing for me was open source. And I remember as an analyst when I first started talking to another analyst from a commercial who is responsible for serving commercial database markets. And I had asked, hey, what about MySQL?

7:01What about Postgres? What about these sort of open source databases? And they said, oh, those aren't paid for, so we don't track those. And that's just very early on. Like I said, in my first days as an analyst, I was like, that doesn't make any sense, because that's not going to reflect usage. All you're measuring is what people are paying for, and if people are using these things that aren't being paid for, then we need to track that. So anyhow, those are the origins, and then that progressed over sort of a decade plus. And like I said, there's many other factors that came in, but that's what started.

7:26Well, I guess when I think about this book, and I think about the software, I guess I just think Linux was like the driving factor, right? I think that was the thing that, I don't know, maybe crystallized all of this, right? It was the developers essentially saying, not only do I want to use this, I'm going to use this, right? And almost dragging Linux into the enterprise to the point that then we had Google, of course, prove that commodity servers running Linux was a very powerful way to run large websites in Red Hat. So I don't know, I guess when I think of it, I almost feel like without Linux, like this maybe would not have happened.

8:00Like that was the linchpin that really broke it open. I don't know, do you think of it that way? Yeah, I think Linux is certainly key, but honestly, I would expand that a little bit. And honestly, I think the biggest example that people point to is the LAMP stack, right? So it's an operating system, a database, a runtime and a web server, right? And that just, that sort of package combination just ended up everywhere. So it's no matter what you're building, no matter how you want to build it, it's going to give you some of the requisite pieces to go out and build, whether it's a toy app for yourself as a developer, whether it's, I remember talking to a bank and they first engaged with Red Hat years and years ago, obviously, because they walked into their data center and they're like, what is all of this?

8:40I didn't know we were running this. And Martin Mikos, back when he was the CEO of MySQL, loved to tell the story of walking into, again, I think it was another bank on Wall Street and saying, hey, you folks might want to talk to us about some MySQL licensing. And they said, oh, that's open source. We're not using that. And he had to wear with all the say. It's interesting because it's been downloaded from your IPs 5 ,000 times. So it's probably in there somewhere. So anyhow, yeah, I think Linux is really at the core of all of this for sure. But I think there were other pieces that need to come along with it.

9:10Now for sure, the LAMP stack. And I think, but I also think of the company that maybe was like largely displaced, at least in the most meaningful way was Sun, right? Sun went away. And maybe we'll do a quick aside here, because this is a fun topic that we all like to talk. If you've of a certain age, Brian Cantrell recently revisited his, why did Sun fail? Right. And I think there's only maybe a 40, 50 ,000 opinions about why Sun didn't make it through. But I don't know what's your opinion. What my opinion to put it on the table is like, I think there's nothing. I don't think there's any like magic thing you could have done.

9:36I think when you, if you will, it's like a tidal wave. When the tidal wave shows up and takes away all your revenue, at least for your big revenue, there's just really no solution. So I'll put my cards on the table. What do you think? Was there some magic thing Sun could have done? Wave your wand, go back in time. What would you have advised? Or maybe you did. Maybe you gave the advice at the time. Who knows? No, I think like every company, first of all, I think we should back up and say that the history of the industry is that there are very few companies that have survived waves of disruption, period.

10:05That's just not a, most companies are not able to do that. And they're not able to do that for a variety of reasons. Clayton Christian wrote a ton about this in the Innovator's Dilemma and so on. Sun, I think, was a difficult case in that it was a bunch of sort of disparate businesses, right? In that they had the ultimate scale-up business, right? From a hardware standpoint, they obviously had the sort of operating in a core sort of operating system. And then they had the runtime business in Java and trying to have all of those weather. So in other words, on the hardware front, you have the disruption from scale out, right?

10:38So Google and everyone else is, okay, we're not buying bigger machines. We're buying tons and tons of cheap, smaller machines. The operating system and the runtime businesses are facing different disruptions, right? Obviously Linux is coming in and disrupting Solaris, a host of languages were flooding in to disrupt Java. Ultimately, I don't think there's any sort of one cause. I think it was largely a business that was phenomenally successful in one market and then just faced disruption from every potential quarter. And that's just very difficult to navigate that. Yeah, I think it's a fair assessment.

11:12Now, Sun's sort of being the big loser, if you will, at least one of them that lost out. Now, if I had gone back in time and you had asked me, because I think at the time Sun bought MySQL and I really, at that time, I was like, this is going to be the end of Oracle, right? Oracle is going to go the way of Sun. Obviously that hasn't happened. And then I think of also Cisco is related, not related, right? That was another area where I was like, I don't know, open source networking and this is getting simpler and things like that. Are you, looking back on it, are you surprised those companies have generally doing well?

11:42I guess you could put that depending on quotations exactly how you want to do it. But they've weathered the storm. Looking back, would you have made that prediction at the time, or are you surprised that they weren't as disrupted, at least as significantly, as a Sun? It's a good question. The Cisco thing doesn't surprise me as much. Networking was very, and arguably still, is largely opaque to and inaccessible to developers. At least it was up until Open Compute and so on. But largely, because it's not something that developers spend a ton of time and attention on themselves, they just count on a network stack and they take it for granted, and there's not a lot for them to geek out on.

12:17in the way that there are in operating system, database, etc., where these are tools that they're going to leverage themselves. So the Cisco side of things doesn't surprise me as much. Oracle, I think you could make an argument that, yeah, that's a little more surprising. But I always think back to talking to reporters over the years as an analyst. Every time there was an Oracle price hike, some new reporter would show up at our doorstep and, oh my God, did you see Oracle hike their prices by X, Y, or Z? And does everybody get to leave now? and we'd have to say, slow down. It's one thing to migrate other pieces of the stack.

12:51The database is the highest risk asset that you have. And even in cases where, I can't remember how many years it was, but it took Amazon years to get off their Oracle databases in spite of running, obviously, their own very competitive and stable and successful database and service offering. So I think databases tends to be stickier than people imagine. and so yeah would I have been mildly surprised possibly but again it was I don't know even at the time people I remember all these forecasts like okay open source databases are going to go out and displace all of these these commercial businesses and I can't remember what year it was but I've been 08 or 09 and yeah I was at the open source business conference used to be run by and they had a panel of banks and insurance companies and the media companies all these like Fortune 500 CIOs, and they were talking at the time about, hey, what are you doing with open source?

13:47They were using open source everywhere, right? It was like languages, tools, runtime, operating system. And with one or two exceptions, they're all like, yeah, we're not on the open source databases yet. We're just, we're not. So yeah, I guess the Oracle side of things doesn't surprise me as much. Yeah, some things are just stickier than you think. I think VMware is the latest thing. People are like, oh, VMware is going to be displaced. And it's like, no. It's like it's doing pretty well still. So as we go forward here, You dispensed this advice for probably, I don't know, at least a decade before there are big changes.

14:14What's your take? Did executives largely listen? I'm sure there's cases of people listening and people didn't. But do you think overall, like when you're talking to companies, people took it to heart and embraced developer and obviously developer relations, this goes through ebbs and flows. But that's certainly something that has been more, maybe more prevalent than it was before. What's your take? Did people listen to you?

14:37Don't panic. This podcast is a work of fiction. From the Global News Network, it's 2029. Breaking news. A hacker calling himself the Purple Yak may have used advanced AI to gain control of a Russian nuclear warhead. He's reportedly offering to launch it at a target of the highest bidder's choosing. Get the news from the end of the world. Follow the clues. Unravel the mystery as civilization collapses. 2029 The War for Our World, a science fiction news series from the future, premiering October 30th. Follow wherever you get your podcasts.

15:16Yeah, I don't really listen to me. I think so much as they it just became unavoidable. Right. In that you it was a redhead event, I think. I don't know what year would have been like 2015, maybe 16. CIO of Barclays, Karen Broadfoot, I think it was. was Ed Barclays was saying like, hey, we hand this book to everybody coming in, right? You have conservative, very large organizations like SAP talking about the new key makers at the time. Certainly there was some like, hey, having a name and having a vocabulary made it easier to talk about, I think. But really, I think what ended up happening was that the book just reflected the reality, which was, all right, developers are much more important than they used to be.

15:55They're much more valuable. And they are not, for years, it was, they were treated kind of like factory workers with pocket protectors and it's a fungible resource. And all of a sudden it was like, oh wait, no, this is a competitive differentiation. And looking around the industry, it's like, think of aqua hiring, right? It's like, okay, I'm going to go out and pay a premium for the startup. I don't even want the tech. I'm just hiring a team. And that, I think that was the, maybe the most extreme example of the value of this persona, if you will. So yeah, I don't, Like I said, I don't know that we or I or the book persuaded sort of anybody at scale.

16:33I think people kind of looked around at the reality around them and said, okay, yeah, no, this is a very important audience that we need to court. Yeah. I've always thought this book pairs, I don't know, maybe intentionally, unintentionally a little bit with market entries and software eats the world or eating the world, whatever that was. And it's like, you can almost just give somebody like these two things. It's a slightly different take, but it's in the same area, right? And if you didn't, I feel like if as an executive, if you didn't take some of that to heart. I don't know. You probably didn't do well.

17:00Maybe you did. I'm pretty sure. I don't think it's been so long since I read it. I'm pretty sure that is, in fact, referenced in the book. Right. And that was, again, just the idea that for those who are unfamiliar with it, it's like, hey, software businesses are making intrusions into all these traditional categories. Right. He goes through the examples of the Brian Cantrell loves to point out that if you go back and look at the actual examples from the essay, the examples themselves haven't aged so well. But the concept itself is, I think, and the idea is an accurate one, right? Which is that digital businesses were really becoming challengers.

17:33And these digital businesses were run and fueled by developers, right? And yeah, I think it's just another example of how important that audience is. All right, so we're all cruising along here. You're out doing stuff. We're all building software. Then, of course, the world starts to change. So maybe the simplest side is on the analyst side, were you already tracking LLMs? I always think of LMS and I guess the data is like the release of chat GPT, right? When it went, I guess, whatever it was, 3.5 or whatever it was. So like, that's the demarcation line I like to use. Were you immersed in this world at all?

18:02Did you see this coming in a way that maybe others, at least myself, didn't necessarily see it coming? No. So I think there, so there's two ways to look at that. The short version is that do we see the world that we're living in right now? No. Right. And I think there are so many times over the years at RedMonk where it's like, I'm just going through technologies like open source or SaaS or even something like distributed version control. It's like, okay, this is a thing. This is where the puck is headed. It's not hard to see if you're talking to the right people. But then LLMs just burst on the scene.

18:36We've heard, probably like yourself and probably a lot of people in the industry, we've heard behind the scenes rumors of, hey, man, there's some tech at these companies where it's borderline creepy, right? Where it's like, hey, you can't really tell if it's human. I think, I can't remember if it was, somebody from Google. Anyhow, there were all these sort of rumors and they'd pop up in the press occasionally. So we'd heard about this, but our sort of my first exposure to some of the potential was, I think, I'm pretty sure, we were the first analyst company that GitHub talked to with Copilot the year before, ChatGPT.

19:04And they came and briefed us and we were like, how does this work legally? And they had to go off and talk to lawyers and get their ducks in a row and all that. But so we got an early look at that. And the example I always use is that we were on heating oil at the time. And so I was like, okay, I'm going to use Copilot to generate a script to go out and scrape the site so I can get better pricing, blah, blah, blah. And so I got to put in a comment, I'm like, okay, I'm going to scrape for pricing. And it auto-completed the URL for the very obscure, I can't remember, it was like mainoil.com, anyway, whatever the, I was like, how the hell did it know that?

19:40That is, that's a pretty trick. I guess prior to ChatGPT, yeah, we had some idea of, oh wait, this is very potentially transformative. But I'd be lying if I said, no, I would know where we were today. Because really up until the inflection point, right, which was, what was it, Opus 4.6 and GPT5, I can't remember what the actual versions are at this point. Even up until that point, it's like, okay, these things are really useful. They have a lot of sort of important potential applications. And then post that inflection point, which was what? November 24, I think that's right. Or November 25. Anyhow, whichever one it was, it's, oh, wait, no, this is materially different.

20:25This is light years. As much as, as impressive as that co-pilot, that initial co-pilot demo was and experience was for us, post the inflection point is just another step function change in terms of, all right, this is a totally different deal. We're in a totally different world. I think that's a really good summary. And I think of it as like we kind of did a speed run from like kind of what you're talking about, like Copilot, which was, oh, this is like really good autocomplete. So even a little bit better than what I think it was. So like Cursor, there's that moment where all, you're still in the IDE and it was like, oh, it could build, like it could do a whole bunch of stuff.

20:55It can do all the files, right? And then I always think of it as Cloud Code is sort of the one where the Cloud Code people came in, but you don't even need to be in the IDE. And for a while, I think a lot of us at the same time, it was like, that doesn't make any sense. And then you had to do something with it to be like, Yep. Now, like now I get it kind of thing. So what's your take now that we've like at least that little section, we've gotten to the point where most code by different, different measures, a lot of code, or if not most code is generated now by LLMs in some form. Do you have an opinion on it?

21:23Are we at a point that should anyone be writing code by hand? And if not, then how much should even people be reviewing code? DHH recently gave a speech, the inventor of Ruby on Rails, where he basically, at his own conference, but at the Ruby conference, said it was dead. Nobody should be writing code. Everyone should just use the LLMs and just do it in Rust. What's your take on all of this? I think that there's a couple different things I think are going on. One, I don't think we have an industry consensus on that. My own feeling is that I think it's likely to be pretty variable on a case-by-case basis.

21:59In other words, we have a class of quote-unquote vibe coding. right? Probably like yourself and a lot of people listening to this, using CloudCode and Codex and all these other tools. I could not tell you how many little toy applications I have running around. Half of them I'm like, oh, I built that. I forgot. And I've never looked at a line of that code. I don't care. It's not particularly important. I needed to do what I needed to do. And if it doesn't work, I go back to the machine and I'm like, okay, go take a look at this. At the opposite of the spectrum, there's a class of application where it's like, all right, I'm controlling the ILS systems for aircraft, or I'm doing medical imaging, or I'm doing, oh, I don't know, defense, right?

22:37Where it's like, yeah, we probably need to know exactly what's going on there. We probably can't just trust two test suites. And I don't know. Like I said, we'll see what the consensus ends up being, but I think we'll have a spectrum. And a lot of that will also depend on like, all right, how much, what's the visibility level, right? So in other words, if you're rewriting Rails and Rust, it's probably going to have a lot of eyes on it, virtual or otherwise. Versus if I'm going back to medical imaging, you're not going to have a ton of eyes on that. And the cost for failure is very high, right?

23:15And I don't remember the, I wish I could remember off the top of my head. I should have thought of it ahead of time. But there was a medical issue with medical imaging devices, I want to say x-rays, and it ended up killing people. And it was basically because you had a, if I remember it right, or approximately it's like a decimal point off. And people receive much higher doses, right? And those are the kinds of things where it's like, yeah, you're probably going to want some sort of awareness, supervision, et cetera. But I think the simple takeaway I think is going to be we are going to have dramatically less code that is viewed and reviewed by, and frankly even understood by humans.

23:52Because even today, and I think this will just be that much more true in the future as people keep vibe coding things, a lot of software is disposable, right? A lot of software is not, the vast minority of software are those medical imaging or loan applications or where it's like, hey, the real world consequences of this stuff is very significant. We're talking scheduling and marketing and media transformation, like all these different things where it's like, okay, I remember talking to somebody at one of the movie studios, right, about their workflow for rendering trailers and bursting them up to the cloud and so on.

24:25that has real-world consequences in terms of if you get it wrong, right? You're potentially talking higher bandwidth bills, but nobody's going to die, right? And yeah, I don't know. I'm very curious to see where the consensus ends up, but I think we're likely to have a spectrum of, yeah, this is not, no one's ever going to look at this. No one cares to, okay, no, this actually matters a lot. Somebody's got to take a look at it. Yeah. And so in your piece, you talk about developers, right? And I think you give not so much predictions, you give some possible outcomes. Like there's a very optimistic take and maybe less optimistic.

24:54So why don't you give us those takes? What are the possible outcomes for, if I'm a developer, been doing it a long time, maybe I'm an architect, senior kind of type of person. What do you see the future or the possible futures are for that kind of person? Yeah, so I think there's, again, I think there's a number of different pathways, right? And the way that I typically talk about it just as a sort of generalization is that I worked construction in high school and college as a summer job. and to me it's the difference between basically up until the inflection point developers were essentially like I'm going to put together this wall I'm going to string stairs I'm going to cut the lumber to size I'm going to fit it, I'm going to assemble it and then set it up now we're more in a world where you're basically looking at I'm not going to build the wall framing I'm not going to build the stairs but I'm going to decide where they get placed and I'm going to decide what that building looks like and how it gets constructed and what the features are and so on.

25:55Essentially, you're going from essentially like a builder sort of laborer job to more of an architect type role. That's the sort of best case, right? The worst case, if you are essentially just human in the loop, right? That is a much reduced role, much reduced significance and so on. And I think a lot of it will depend on ability, right? Because as I mentioned, we have the inflection point. There's a step change in function. And yet, I can't speak for your experience, but the experience, my own personal experience using the tools and certainly from the majority of people that we talk to is that they're really good, but there are limits to what they can do.

26:32There are limits to how maintainable these applications are going to be moving forward. And I would personally probably lean towards, we're still going to have human supervision and a human role for taste and judgment and creativity, if nothing else, because God knows they're not great at that. I think it's TBD. Like I said, a lot of it's going to depend on where you're working and how you're working, what you're working on. Because look, if you're churning out effectively disposable software, if that's your job and your role, that's something the machines are very good at. And there's not likely to be a role for humans in that.

27:07Yeah, I keep thinking to myself, is the role of like product management, software developer, software architect, is that just like collapsing into like one, right? Kind of one person or just a couple of people that maybe have very similar schools, maybe have one person a little bit more technical, a little bit, but they can kind of do everything. They can kind of run the agents themselves. Because I think it works. The one thing I've seen, the backlash a little with AI sometimes. It's just there's at the individual level, you feel like you're super productive, like you feel like you're super human, right?

27:35But then as executives, as you roll this up, companies and executives will often say, I don't really see the, we're not seeing that up. And I think part of the problem there is like, if you still are running like teams and I'm an executive and I'm not building it, I'm having a team do it. You still have all this coordination, which creates a lot of that versus if it's just you and a couple of people, like that's when it gets fast. So I think that's going to be like a real, I don't know, almost like organizational change. It's, I don't know, what's the role of the executive used to be like, get budget, help build the team, run a big team, do all the HR stuff.

28:07But now it's like, the more we have that, the more that you're slowing it down. So I don't know. I don't have any thoughts on organizational change or how things should be get built going forward. Yeah. I think basically what we've seen in retrospect, we should have expected it. But I I remember we were sitting on a call with the Dora folks in 24, it might have been. And so you had that split that you're talking about where the individual reporting from the survey standpoint in terms of individual developers was like, oh, we're superhuman. We're cranking this stuff out. And the organizations, to your point, were saying, like, we're not seeing the benefit, so I don't know where the split is.

28:39And you kind of think about it and it's, okay, obviously the act of creating code is one part of a very long and complicated lifecycle. It's a critical part. and that got dialed up to 11 and none of the rest of the downstream the processes downstream of that were equipped for this flood so you talk to open source maintainers for example and it's in many cases made their lives like living hell right where it's like i'm getting an order of magnitude more prs and submissions and feature requests and so on i'm getting basically a bunch of slides taking me longer to go through it and yeah i think it's a i don't know I think one of the things I always think about, I go back to when you think about, okay, what are going to be the implications for human employment?

29:27Excuse me. So that it's just unpredictable, right? And there's a great example from Dr. Jim Besson, who kind of can remember if he's B or BC, I always get this wrong. But anyhow, he's a professor down in Boston, and he studied automation for years. And he looked at, there's a good graph. If you go Google Dr. Besson ATM, you'll probably turn it up. and you look at the sort of employment for human bank tellers. And you would figure, hey, we have these machines, they do it faster, more accurately. The employment for bank tellers is going to implode. And instead of went way up, right? And why did it go up?

30:00It's, oh, wait, if I can make it easier and faster to do this, I free the human up to do actually more interesting work than counting money, which a machine can do very easily. But then also I can open away more branches, right? And the net employment was counterintuitively actually grew, right? So that's the thing I think. I think a lot of conversations about employment, what the implications are going to be for developers and teams are fairly reductive and are based on assumptions that may or may not be accurate. I was talking to Adam Jacob a while ago, and the question I was asking, and I think about this a lot, is we are thinking in, we're operating in a particular context where there's a certain amount of software, right?

30:39Both individually, consumer, and enterprise. and what happens if that goes and what if that's an order of magnitude bigger right so you know back to the bank branch example if we had the same number of bank branches or a fewer jobs when there's more bank branches there are more human jobs what if there's way more software what if there are way more software companies what does that look like from an employment standpoint so i think it's difficult to predict yeah no absolutely a lot of unknowns now one of the things all the vendors are trying to figure out is how do they make sense of it? And I think in your piece, you got a lot of good examples.

31:16And I think one of the things that trends you're seeing, like I've seen a lot of people write this up, essentially the headless application, maybe Salesforce is the most recent one, where it's, Hey, really, even if you have a system of record, it's kind of like, we don't really want the UI anymore. It would be great if we could at least interact with these or have an option to interact through our favorite LLM and just have it do stuff, right? Just ask it to do it, create the report rather than hunt and peck in the UI. And you kind of, I kind of think in your piece, you kind of draw a distinction of like, here's some current companies, right?

31:44And how are they going to adjust? And then for lack of a better word, I'm going to call it agent first. Companies that have built their entire, and maybe even this is almost a callback to the open source, kind of whatever, a decade ago, if not longer, right? It was like, do you embrace open source? Is that the right position? Or can you, if you will, take your thing and make it, if you will, agent native? So as you go forward, how do you think these companies and what are you advising companies and vendors that are trying to make this shift from kind of like traditional enterprise SaaS or GUI-based to agent-based?

32:14I don't think there's necessarily a simple heuristic, right, in terms of what should the appropriate sort of stance be, like headless sort of UI or human agent. Again, I think it's largely going to depend on market. And I think there will be entire businesses that are agent-only, right? And I think there will be some businesses where, okay, it's either predominantly human or certainly like human AI workflow. So I think you'll see both. And I think the real difference is going to be the scale, right? Because it's obviously gone on to implode. But the Gastown, which is Steve Yeggie's initial sort of attempt at an IDE to harness teams of agents, it's just so easy to take one human developer and turn them into essentially a virtual team, right?

33:02And what's the, what's the podcast and like shell game where it's like a company of run by human and made up of agents and all that. Yeah. I think it's, it's, like I said, I think we're going to have both, but I think the real growth is going to be at agents, right? Strictly because they're digital, right? It is not a, all right, I need to go find and hire and recruiting human, it's a, hey, I'm going to, it's actually, it's a little bit like cloud in that respect, right, in the sense that, hey, I can scale and I can go out and talk to Dell or HP or IBM or whomever and get a bunch of physical machines and get them shipped.

33:32Or I go to Amazon, I can press a button and have thousands of them online in seconds. And that's where the scale is going to come from. And certainly I think a lot of the vendors are going to, I don't know, claim or rush to meet that scale. Yeah, I think the one thing I do like to see is I think Cloudflare is a good example because they have, they basically have, I think they have an API or their command line utility used to be called Ring or it is called Wrangler, but they've come out and they basically said, hey, we're rewriting it. And like Wrangler, I don't know, Wrangler had, let's say 10 % of the coverage of things you can do in Cloudflare.

34:02I'm making that up, whatever it really is. And now, but they've explicitly said this new API or their new utility is like, it has everything, right? And it's like, you can, and I think that's like the simplest thing for every company to do is, and you don't even have to, I don't even think you have to make it overly complicated with like an MTP service. And it's like, just give an API, Make an API that I can do everything and then turn the LLMs loose, obviously with appropriate security precautions, and see what people do. Because I think that's going to really be the driving force. It's almost like product management by agent.

34:31Once I give all these APIs and give them access, what do they start to do? And that probably starts to drive your roadmap, which I think is fun. Yeah, I think that's right. And I think we'll see that. And I think part of it is just that giving them sort of keys to the kingdom, I think the security angle of that is easier said than done. And the example I use with people is that one of my tasks in terms of testing these tools historically has been there was a great, I think he works for CNN. He's on the sort of web team at CNN.com. And he wrote it. He built this sort of Dodgers. He's a L.A. Dodgers baseball fan.

35:01And he built this dashboard for the Dodgers, tracking their data and all that. And he released this as a project on GitHub. And I was like, okay, great. And I historically have taken sort of that open source and turned different tools on it. And to your earlier point about Cloud Co was the first one. and was like, got it, here's your Red Sox version. Boom, done. I was like, whoa, okay. And it's like RedSox.bot and all that. But one of the things it did in the process was that it, so one of the things on the site is they have a roster page and he, Matt Stiles had all these, I don't know, cartoons or caricatures or whatever, presumably drawn up by the players.

35:36I don't know if he used AI or not. But Cloud Code was like, you know what? It'd be much better if we had the actual headshots from MLB.com. And I was like, yeah, no, you can't actually do that. and I'm going to get in trouble is the terms of service violation. And that's the thing is that there are things that are very easy to foresee from a production standpoint. Don't go delete databases. And obviously sometimes they do and all that. But it's the subtle things where you're like, if you're not paying attention to that, you're going to get yourself in trouble. And I don't, yeah, I'm digressing here.

36:03But I think the sort of gist of it is that we're going to see, we are seeing, as you mentioned in the piece, we're seeing a lot more people target this market. And we're going to see a lot more functionality open up to the agents and appropriately. at least. But those guidelines, I think, are going to be harder than people imagine them to be. Absolutely. All right. Listen, as we wrap up here, I always like to do the fun questions everyone's talking about. Give you a chance to just give us your hot take if you feel like you want to. If not, you can obviously opt out. Obviously, there was the, I don't know, a few weeks ago, someone said there's 10 % chance all humanity dies because of AI.

36:33So where do you come down? Is this like I put the sustaining technology to Skynet? Do you have an opinion of like where we're going to end up? I just, I think the thing that I come back to is that But I think, you know, so I was a history major in college, not a comp sci major. And I maintain that there's a lot of utility in having a background, educationally speaking, not just in history, but just outside of the digital realm. Because, look, these systems do pose clear and present dangers, right? To, in other words, can you use these tools, for example, to go out and have massive real world implications in terms of going out and hacking power grids?

37:13Yes. That is undeniably a sort of present risk. But it's one of these, I know Brian Kancherl and Adam Leventhal talked about this on the Oxide podcast a while ago with Simon Wilson, where I shared their, I don't know, sort of concerns, objections, critique in that the AI world in many cases just, in my opinion, does not draw a sufficiently bright line between here's what can be done digitally and here's what can be done physically. right so in other words think about all the anthropic people are like okay bioweapons and nuclear weapons and so on it's like okay but you need raw materials for that i'm gonna go build okay so cloud tells me how to build nuclear weapon okay like where do i get the radioactive elements like how do i physically work that so i don't know and i think the sort of the other piece of this and lots of people made this point which is if you legitimately thought that there was a 10 % chance that this would wipe out humanity.

38:14I think most people would immediately pave the earth and salt it so that nothing would ever grow. That is not an acceptable level of risk. And yeah, I'll take the under on sort of the 10 % chance of humanity being wiped out with the caveat that, like I said, they do have the capability because so much of the world is digital now, right? They do have the capability to wreak some havoc, but it's just the, I don't know, Skynet. They keep talking about AGI. And my response is always, okay, cool. Define for me human consciousness. Define that. People have been working on that for literally a century and we still don't have a working definition for what it is and how it works.

38:57And I was talking to somebody and they're like, yeah, but it went out and hacked sort of, it broke out of the sandbox and hacked all these systems. I'm like, because it was trying to follow instructions originally given to it by a human. It didn't spontaneously wake up one day and say, I am now sentient, I want out of here, and I'm going to go do this. So, I don't know. We'll see. I could be wrong, but I will 100 % take the under on that. I'm with you. Let's definitely take the under. Okay, and then the final question here, just because the IPOs are, in theory, they're around the corner, Anthropic and OpenAI.

39:28And then, so ultimately, do you feel like these companies are at their stated values? Are they overvalued, undervalued, no opinion? What do you think? Of course, we're going to put the caveat here. We're not giving investment advice. Don't listen to us. Do what you want, right? I have said, I think, I've been pretty consistent with this for a while in that I don't understand some of the math from a valuation standpoint. I 100 % don't understand and cannot fathom the level of circular investments in many cases that are being made here, where it's like, I'm going to loan you money that you're going to pay me back.

40:04I think at one point I ended up, there's a Beavis and Butt-Head episode where the two of them buy, quote unquote, all of the candy that they're supposed to sell for this charity thing using the same$2. And I 100 % think there's quite a bit of that. And that being said, these are incredibly capable products. They're delivering very high value in many cases. But it's just, you look at some of the math and it's like, all right, I can't remember the study, but it's like, it was assuming that an enormous number of people were going to pay the same for these services they were going to pay for food.

40:40And I don't see how that works exactly. So yeah, I could be wrong, but I'm very skeptical of some of the sort of touted valuations for some of these products, particularly because the sort of open models are getting better and better and they're getting better faster. Where it used to be six to 12 to 18 months leads for the frontier models. And it's now in many cases down to six months, three months, or in some cases, one month. And there's the caveat like, okay, it's a fantastic model, but also you need serious hardware to run that, right? So, and the frontier models are in many cases demonstrably better.

41:17I asked the question, I think last summer, like what is good enough? Right. So like, I think a fable when fable came out, I was like, oh, it's better. But I'm like, it's not better enough than Opus that I'm going to pay some dramatically increased price for it. And what happens six months from now, 12 months from now, where we have open weight models that are small enough to run locally that get us a acceptable level of performance. Again, that's an interesting market pressure that I'll be curious to see how they weather. Yeah, and where we began is when we look back at the sun and Solaris, I'm sure there's people still out there that Linux still misses this great Solaris feature.

41:56But at some point, the world and developers and people, they're going to bring something into a price point, maybe it's free or maybe that they want. The market just will not allow it to go on forever. Now maybe that'll take a while, but eventually it happens. So it'll be fun to watch, see what happens with these IPOs. and I'm sure there will be many ups and downs. I mentioned at the beginning that everyone knows what RedMonk is, but probably not everyone does. Why don't you give us a little bit of what exactly do you do at RedMonk? And then most importantly, how do people find you online if they want to talk to you more?

42:23Sure. So we at RedMonk are a developer slash practitioner focused analyst shop, which means that we try to understand what practitioners like developers, DBAs, admins, operators, whomever, actually with AI now. That probably includes everybody, just what people are using, why they're using it, and how they're using it. And so if you want to understand that, obviously you can go find our research on redbank.com. And then we do a lot of work with companies in terms of, okay, hey, I want to, I have a product, I have a service, I have whatever. And I want to make sure that I'm optimizing for that practitioner audience.

42:55And we do a lot of work in terms of trying to help people understand, all right, what do you want to build? How do you talk about it? How do you message it and all that? So yeah, redbank.com, that's the best place to find us. Yeah, definitely check out all the writing. There's a great blog, Stephen. Your posts are up there. There's always fun. And there's plenty of free stuff to read. Plenty of free stuff. All our research is free. And check it all out. Yeah, it's all being consumed by the models as well. So we have that. All right. Fantastic. All right, Steve, thanks a lot for being on the show today.

43:19I really appreciate it. No, my pleasure. Thanks so much. All right. And everyone else, really appreciate you listening or watching us on YouTube. Of course, it'd be great if you subscribe to the podcast if you haven't already or give us a quick, that always helps. And with that, we'll talk to you next time. Thanks for listening. Check us out at theenterpriseaishow.com for past shows, newsletters, and all things Enterprise AI.