Leading Teams in the Time of AI with Andrew Murphy

---

Summary

Andrew Murphy, a technology leader and founder of CTO School in Australia, joins Carl Franklin and Richard Campbell to discuss the profound impact of AI on software engineering teams and leadership. The conversation centers on the emotional journey many experienced developers are undergoing as AI tools like GitHub Copilot and Claude change the nature of coding work.

Murphy references Nolan Lawson’s poignant blog post “We Mourn Our Craft,” which articulates a sense of grief and loss among engineers whose identity is tied to handcrafting code. He frames this transition through the lens of the five stages of grief (denial, anger, bargaining, depression, acceptance), emphasizing that leaders must acknowledge and help their teams navigate these feelings. The discussion highlights that while AI can generate code, the core value of a senior engineer lies in problem understanding, architectural judgment, and domain knowledge—skills AI currently lacks.

The hosts and guest explore the practical implications for leadership. They discuss the moral responsibility engineers retain for AI-generated code, the potential for juniors with AI tools to outpace resistant seniors in output, and the danger of long-term “research projects” that become obsolete due to the rapid pace of change. Murphy stresses that effective leaders must champion smart adoption, focusing on quality controls and understanding the trajectory of improvement, rather than denying it.

The conversation also touches on broader industry trends, including the potential bursting of the AI investment bubble, the shift towards running models locally for cost efficiency, and the future where “taste” and human-centric design may become key differentiators in software. They conclude by acknowledging the very real depression and existential questioning some engineers are experiencing, urging leaders to create space for these conversations and guide their teams toward the new career shapes emerging on the other side of this technological shift.

---

Recommendations

Articles

• We Mourn Our Craft by Nolan Lawson — Described as ‘the best thing I’ve read about what it actually feels like to be a software engineer right now.’ It’s a beautifully written, honest exploration of the grief and loss of identity many feel as AI changes coding.
• Embrace the Suck by Andrew Murphy — Andrew’s own blog post about the transition to leadership, arguing that a lot of being a leader involves doing difficult, ‘sucky’ things, and avoiding them means not being the leader your team needs.

People

• Kent Beck — Referenced for his insight that with AI, ‘the value of 90 percent of my skills has gone to zero and the value of the remaining 10 percent has 1000X. The job I have now is working out what that 10 percent is.’
• Nolan Lawson — Author of the influential blog post ‘We Mourn Our Craft,’ which powerfully captures the emotional state of many software engineers facing the rise of AI code generation.

Tools

• Squad by Brady Gaster — An AI tool that gives you a development team through GitHub Copilot. You describe what you’re building, and it creates persistent, specialized AI agents (front-end, back-end, testers) that live in your repo, learn your codebase, and collaborate.

---

Topic Timeline

• 00:01:21 — Introduction and Historical Context of 1991 — Carl and Richard introduce the show and delve into historical events from 1991, including the dissolution of the Soviet Union, the Gulf War, and the release of seminal music albums like Nirvana’s ‘Nevermind’. They also touch on key tech events such as Tim Berners-Lee announcing the World Wide Web and Linus Torvalds releasing Linux. This sets a backdrop of significant change, mirroring the current AI-driven shift.
• 00:17:58 — Better Know a Framework: Squad AI Tool — Richard discusses ‘Squad’, an AI tool by Brady Gaster that creates a team of specialist AI agents (front-end, back-end, testers) within a GitHub repo. He shares insights from a demo with Jeff Fritz, noting its efficiency and surprisingly low cost due to GitHub Copilot’s request-based pricing model. This segment highlights the practical, multi-agent AI tools already changing development workflows.
• 00:20:26 — Listener Comment on AI and Vibe Coding — The hosts read and discuss a listener comment from John Suda reacting to a previous episode. The comment expresses concern about the pace of AI adoption and the notion that ‘only hobbyists will soon read, let alone write code.’ Carl and Richard agree on the need for vigilance and moral responsibility, emphasizing that engineers are ultimately accountable for any code that ships, regardless of its origin.
• 00:25:44 — Welcoming Guest Andrew Murphy on AI and Leadership — Andrew Murphy is formally introduced. He explains that despite trying to avoid it, AI dominates every conversation at his CTO School meetup. He introduces Nolan Lawson’s blog post ‘We Mourn Our Craft’ as a beautiful expression of the grief many engineers feel. Murphy disagrees with a fatalistic acceptance, arguing this is another in a long line of industry shifts, similar to the move to the cloud or the advent of IntelliSense.
• 00:33:42 — The Core Value of Engineers in the AI Era — The discussion focuses on what remains uniquely valuable about human engineers. Andrew and Carl emphasize that deep domain knowledge, understanding customer pain points, and the judgment to know when a hacky solution is appropriate versus when to architect properly are skills AI cannot replicate. They reference Kent Beck’s insight that 90% of a developer’s skills may lose value, but the remaining 10% could become 1000x more valuable.
• 00:46:28 — Managing Team Resistance and Skepticism — Andrew addresses how to lead teams through this change. He distinguishes between valuable skepticism (questioning where to apply AI) and dangerous denial (rejecting the overall trajectory). He notes that the informal leaders of a team—those others look to during a break—are often the ones struggling with this transition and are worth investing time in to help them adapt.
• 00:55:44 — Acknowledging the Depression Stage of Grief — Andrew highlights the importance of acknowledging the ‘depression’ stage in the grief model. He stresses that leaders need to create space for engineers to discuss their existential concerns about their careers and craft, which often go unspoken. Validating these feelings is crucial for helping the team move forward to acceptance and finding their new role in an AI-augmented workflow.
• 00:58:34 — Conclusion and Future Outlook — The conversation wraps up with Andrew sharing his personal focus on family and writing more authentic content about leadership. The hosts agree that change is constant in tech, citing past transitions from VB6 and ColdFusion. The key takeaway is to guide teams through the emotional journey while pragmatically adopting tools that enhance, rather than replace, human judgment and creativity.

---

Episode Info

• Podcast: .NET Rocks!
• Author: Carl Franklin and Richard Campbell
• Category: Technology Software How-To
• Published: 2026-02-26T04:00:00Z
• Duration: 01:03:00

References

• URL PocketCasts: https://pocketcasts.com/podcast/net-rocks/2b369ce0-0426-012e-f9a0-00163e1b201c/leading-teams-in-the-time-of-ai-with-andrew-murphy/52073f7c-2ddf-4732-a2b1-6debc99c7163
• Episode UUID: 52073f7c-2ddf-4732-a2b1-6debc99c7163

---

Podcast Info

• Name: .NET Rocks!
• Site: http://www.dotnetrocks.com
• UUID: 2b369ce0-0426-012e-f9a0-00163e1b201c

---

Transcript

[00:00:00] Psst! How’d you like to listen to .NET Rocks with no ads?

[00:00:04] Easy! Become a patron.

[00:00:06] For just $5 a month, you get access to a private RSS feed where all the shows have no ads.

[00:00:13] $20 a month will get you that and a special .NET Rocks patron mug.

[00:00:18] Sign up now at patreon.dotnetrocks.com

[00:00:21] Hi, this is Carl Franklin.

[00:00:24] And this is Richard Campbell.

[00:00:25] We’ve got two special shows coming up soon.

[00:00:28] Episode 1999 and 2000.

[00:00:32] For episode 1999, we’re collecting people’s Y2K stories.

[00:00:36] What did you do to help the Y2K event not actually happen?

[00:00:40] And for episode 2000, we’re going to be sharing stories about how .NET shaped your career.

[00:00:45] We have a special page at .NETrocks.com slash VoxPop

[00:00:49] where you can record messages for us that we can play on these special episodes.

[00:00:54] So tell us what you did for Y2K.

[00:00:56] And what .NET means to you.

[00:00:58] And of course, how long you’ve been listening to .NET Rocks.

[00:01:01] So go to .NETrocks.com slash VoxPop now

[00:01:04] and leave us a message before the thought evaporates like whiskey left in a glass overnight.

[00:01:09] Do it!

[00:01:21] Hey, guess what? It’s .NET Rocks all over again.

[00:01:24] I’m Carl Franklin.

[00:01:26] And I’m Richard Campbell.

[00:01:27] We’re here again.

[00:01:27] We’re here again.

[00:01:28] Again.

[00:01:28] Again.

[00:01:29] Keep doing it.

[00:01:30] You go again.

[00:01:31] 1,991 times.

[00:01:33] You keep going and going.

[00:01:35] You keep going and going.

[00:01:37] You go now.

[00:01:38] Okay.

[00:01:39] Yeah.

[00:01:39] We’re only nine shows away from 2000, man.

[00:01:42] And that’s going to be hellaciously cool.

[00:01:45] It’s going to be fun.

[00:01:46] If nothing else, it’s going to be a party.

[00:01:48] Yeah.

[00:01:48] Literally.

[00:01:49] Party with Palermo.

[00:01:51] Party with Palermo.

[00:01:51] At the MVP Summit.

[00:01:52] Okay.

[00:01:53] Well, this is going to be a good show.

[00:01:55] Andrew Murphy is here.

[00:01:57] We are talking.

[00:01:58] We’re going to be talking to him in a few minutes.

[00:02:00] And before we get to him, though, let’s do a few things.

[00:02:03] Starting with what happened in 1991.

[00:02:06] And Richard, I’m going to let you go first.

[00:02:09] Because I’m really interested to what you have to say about the dissolution of the Soviet Union.

[00:02:13] Yeah.

[00:02:13] Well, obviously, this has been, you know, things have been unraveling since the late 80s.

[00:02:18] Yeah.

[00:02:18] And so, there’s a referendum in the Soviet Union.

[00:02:22] A national referendum.

[00:02:24] Six of the republics refused to participate.

[00:02:26] Right.

[00:02:27] But the majority that do.

[00:02:28] The majority that do participate are 80% in favor.

[00:02:30] And by the end of 1991, the USSR will be no more.

[00:02:34] There will be an attempted coup on Gorbachev.

[00:02:36] There’s a whole bunch of things that happened at once.

[00:02:39] Right.

[00:02:40] Estonia, Latvia, and Lithuania, who had already declared independence in 1990 and are fighting with the Soviets at the time.

[00:02:49] There’s conflict.

[00:02:50] It’s like the old crackdowns from the 50s and 60s.

[00:02:53] Yeah.

[00:02:53] So, they were first.

[00:02:54] But it goes on from there.

[00:02:59] Moldova, which is east of Romania, also declares independence in 91.

[00:03:05] All of the stands, Kyrgyzstan, Uzbekistan, Tajikistan, Kazakhstan, Turkmenistan, all five of them declare that’s in the eastern Central Asia.

[00:03:14] And then down in the Caucasus Peninsula, so that’s Georgia, Azerbaijan, Armenia, and of course, Ukraine.

[00:03:20] Now, these were all regions of the Soviet Union before, right?

[00:03:23] They did have borders.

[00:03:25] Well, these were all originally, these were countries.

[00:03:27] Oh, they were originally countries.

[00:03:28] These had always been independent countries.

[00:03:30] But in the era of Stalinism, at the end of World War II, as the Soviet Union was already formed, they were controlled by the Soviets.

[00:03:39] And they were expanded as the Warsaw Pact into the West.

[00:03:43] That included also Poland, Romania, and so forth.

[00:03:45] It was the end of World War II, and Germany basically took those countries over.

[00:03:51] They were independent countries before World War II.

[00:03:53] Germany took them over as part of.

[00:03:55] And then it ended up them merging as part of the Soviet Union.

[00:04:00] So they’ve been kicked around a bit.

[00:04:01] As has most of Eastern Europe.

[00:04:04] Yeah.

[00:04:04] Yeah.

[00:04:05] And as Moscow loses control in the late 80s, Poland’s one of the first.

[00:04:10] And all of the Warsaw Pact, like the Warsaw Pact actually ends in 1991, but it had already fallen apart at this point.

[00:04:16] What I know about Poland is that the geography of Poland makes it very susceptible to invasion.

[00:04:21] And it’s been invaded many times by many other countries.

[00:04:23] Because of that.

[00:04:25] Yeah.

[00:04:25] And, you know, they’re still here.

[00:04:27] Yeah.

[00:04:27] But under control lots of different ways.

[00:04:31] The Central Asian states, the Stans, as you refer to them, are obviously very important.

[00:04:36] So you have it apart for some time.

[00:04:38] And so they do become part of the Commonwealth of Independent States.

[00:04:42] That referendum is really about should we have more independence.

[00:04:48] The current Soviet system will end, but something will replace it.

[00:04:51] This is also the period where…

[00:04:53] Boris Yeltsin is elected as the president of the new Russian Soviet Federative Social Republic.

[00:05:00] So they come up with a new name.

[00:05:02] Other words, Russia.

[00:05:03] Yeah.

[00:05:04] Yeah.

[00:05:04] The Russian Federation.

[00:05:05] Sort of a collective.

[00:05:07] So it is an unraveling.

[00:05:09] And it will affect space as well as we get down into that portion of the story.

[00:05:14] And obviously, depending on where you are in the world, this was a disaster or a huge success.

[00:05:19] I mean, this is also when Yugoslavia starts to break up.

[00:05:23] Right.

[00:05:23] And…

[00:05:23] Right.

[00:05:23] It goes immediately into conflict.

[00:05:25] Yeah.

[00:05:26] Serbs and Croats and the Macedonians.

[00:05:28] It turns into what will become quite a nasty…

[00:05:31] Nasty, nasty.

[00:05:32] …civil war for several years.

[00:05:34] Yeah.

[00:05:34] But on top of that, this is when the Gulf War begins.

[00:05:38] Yep.

[00:05:38] That’s right.

[00:05:39] The invasion of Kuwait had already happened.

[00:05:42] And George Bush is president.

[00:05:44] So it’s crazy to think about with all of that going on, that also the airstrikes begin for the first Gulf War.

[00:05:51] It’s also when apartheid ends in South Africa.

[00:05:53] Yeah.

[00:05:54] Like, didn’t anyone is…

[00:05:55] It’s just a lot of stuff, a lot of history going on.

[00:05:58] A lot of it happened simultaneously there.

[00:05:59] I remember David Crosby on somewhere on the news or something in 1989.

[00:06:04] He says, you think the 60s were wild.

[00:06:07] Just wait till you see the 90s.

[00:06:10] Yeah.

[00:06:12] Yeah.

[00:06:12] So it’s a very intense time.

[00:06:15] And a lot of the things that have been coming, you know, are coming to fruition.

[00:06:19] Like I said, so the USSR will be over by the end of the year.

[00:06:22] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] Yeah.

[00:06:23] The other voice you heard was Andrew.

[00:06:26] Do you want to say something about the chain, the Balkan chain?

[00:06:31] Yeah.

[00:06:31] So we were talking about this in the pre-show, and this is…

[00:06:35] I’m just old enough to remember this kind of transition away from the USSR.

[00:06:40] And I have a very vivid memory of watching the TV of something called the Baltic chain,

[00:06:46] which wasn’t actually in 1991.

[00:06:48] It was a couple of years earlier, in 1989.

[00:06:51] And it was at $2 million.

[00:06:53] And I remember seeing people across Eastern Europe held hands in an unbroken chain over

[00:06:58] 400 miles long to kind of protest the USSR.

[00:07:02] I just have this vivid memory of the news shows about it.

[00:07:06] And if you’re in Europe, you probably know about the Black Ribbon Day, which is the 23rd

[00:07:11] of August every year.

[00:07:13] And that’s kind of the remembrance of, you know, anti-Stalinism and Nazism.

[00:07:17] And the reason it’s the 23rd of August is because that’s when the Baltic chain or the

[00:07:22] Baltic Way was.

[00:07:23] Wow.

[00:07:24] Cool stuff.

[00:07:25] All right.

[00:07:26] Let me get to some other things that happened.

[00:07:29] You already talked about all of that stuff.

[00:07:32] There was a military coup that overthrew Haiti’s first democratically elected president.

[00:07:38] So Haiti.

[00:07:39] You’re going to talk about the World Wide Web, of course.

[00:07:42] So I’ll leave that to you.

[00:07:44] That last.

[00:07:44] What I want to talk about is some cultural stuff.

[00:07:48] Nintendo released the Super Nintendo Entertainment System,

[00:07:52] which was…

[00:07:53] It was amazing.

[00:07:54] There were some good movies.

[00:07:56] Terminator 2 was there.

[00:07:59] Silence of the Lambs was there.

[00:08:00] But I want to talk about the music.

[00:08:02] So the top 10 albums of 1991.

[00:08:08] And, you know, for me, 1990 was sort of the comeback of people who played their own instruments.

[00:08:16] You know, in the 80s, it was all synthesizers and MIDI and all that stuff.

[00:08:20] And it just wore on my soul.

[00:08:22] You know, I grew up.

[00:08:23] I grew up listening to 60s and 70s.

[00:08:25] And, you know, I wanted to play like Peter Frampton and Brian May and Eric Clapton, right?

[00:08:30] People who really were musicians of an instrument and practiced their craft on an instrument.

[00:08:37] And so in the 90s, it sort of came around again.

[00:08:41] So in 91, let’s talk about Matthew Sweet, Girlfriend, was number 10.

[00:08:46] And I remember I was working at Voyatro Technologies and the head engineer there,

[00:08:53] was friends with Matthew Sweet.

[00:08:54] So I got to hear his stuff before it went live.

[00:08:58] The Pixies, number nine.

[00:08:59] Trompe Le Monde, a good album.

[00:09:02] Soundgarden, Bad Motor Finger.

[00:09:05] De La Soul.

[00:09:06] De La Soul is dead.

[00:09:08] R.E.M., Out of Time, number six.

[00:09:11] Number five, Pearl Jam, 10.

[00:09:13] Number four, U2, Aktoong Baby.

[00:09:17] Hello?

[00:09:18] Like, this is amazing stuff.

[00:09:20] And number three, A Tribe Called Quest.

[00:09:23] The Low End Theory.

[00:09:25] Number two, My Bloody Valentine, Loveless.

[00:09:27] And number one, Nirvana, Nevermind.

[00:09:31] Nice.

[00:09:31] And that really started the whole grunge movement.

[00:09:34] Yeah.

[00:09:34] But it’s interesting that Neil Young had a great album on the top 10 in 1990.

[00:09:39] And he was called the grandfather of grunge, right?

[00:09:42] Because.

[00:09:43] Sure.

[00:09:43] Yeah.

[00:09:44] He was sort of the inspiration for all of that stuff.

[00:09:46] And Neil never stopped rocking all throughout the 70s and 80s.

[00:09:50] Well, and certainly a man who always played his own instruments, too.

[00:09:53] Right.

[00:09:53] Absolutely.

[00:09:54] Come on.

[00:09:54] And, you know, there was some great hip hop albums and stuff, too.

[00:09:57] But for a white kid who grew up listening to primarily white bands, but some black bands, too, of course, black musicians.

[00:10:04] But, you know, I was always just into real music.

[00:10:08] And, you know, nothing about rap and hip hop and all of that isn’t real music.

[00:10:15] It really is.

[00:10:16] But the electronic stuff really just drove me crazy.

[00:10:20] And so in the 90s, it felt like a sigh of relief.

[00:10:22] Oh, my God, this is great.

[00:10:24] Bands are back.

[00:10:25] So, yeah, we’ll talk about more of that later on.

[00:10:28] Anyway.

[00:10:29] All right.

[00:10:30] The space side is fairly short.

[00:10:31] There are five shuttle flights in 1991.

[00:10:33] Two of them are militaries.

[00:10:34] There’s not much to talk about.

[00:10:35] Columbia flies another space lab mission, which is awesome.

[00:10:39] To me, the most interesting mission of the bunch.

[00:10:41] Discovery and Atlantis both launch observatory satellites.

[00:10:45] Discovery does the upper atmosphere research satellite, which is about detecting carbon dioxide.

[00:10:49] Yeah.

[00:10:50] And Atlantis flew the Compton.

[00:10:52] Gamma Ray Observatory, one of the great observatories, which is an awesome, massive machine.

[00:10:57] It, like the Hubble, was designed to be serviced, although it never was.

[00:11:01] Ultimately, it worked as it was, and there was no reason to ever upgrade it.

[00:11:04] That was the whole idea behind this collection of things with the shuttle was to do maintenance on them.

[00:11:09] But ultimately, only Hubble.

[00:11:10] Of course, Hubble leaded the most.

[00:11:12] Hubble’s already launched.

[00:11:12] It just doesn’t work because they’ve misground the mirror.

[00:11:15] So they’re still at this point trying to figure out how to fix that.

[00:11:18] But there’s one Russian flight, Soviet flight, actually.

[00:11:22] The important one, TM-13, poor old TM-13.

[00:11:26] This was a Soyuz mission to the Mir space station, which is, of course, operating at this time.

[00:11:32] And two Russians are on Mir space station, and they go up there as Soviet Union members in October of 1991.

[00:11:44] But when they return in March of 1992, they will be the last Soviets.

[00:11:48] They’ll return.

[00:11:49] There’s no more Soviet Union when they come back.

[00:11:51] There is no Soviet Union.

[00:11:51] It’s out.

[00:11:52] So Alexander Volkov and Sergei Krikalev will be somehow impacted by that huge change, right?

[00:11:59] Okay.

[00:12:00] Yeah.

[00:12:02] What?

[00:12:02] You have no passports?

[00:12:04] The other part of this, of course, is Kazakhstan.

[00:12:07] All of these flights go out of Kazakhstan, which is now an independent nation, too.

[00:12:11] So one of the things they did on TM-13 was they flew a Kazakh astronaut or cosmonaut.

[00:12:17] So they’re trying to keep the Kazakhs engaged as they see the changes that are going on.

[00:12:22] And so the Kazakhs get their first astronaut as part of TM-13 to go up to Mir and maintain the agreement that it still exists to this day to host the space flights of the Russian space system.

[00:12:34] Okay.

[00:12:35] Computing.

[00:12:36] I mean, we have to lead with this.

[00:12:39] Tim Berners-Lee announces the World Wide Web project in a news group and provides a link to the very first website in the world, info.cern.ch, because it’s in Switzerland.

[00:12:52] Of course, the only graphical browser that exists at that time is on the Next, because TBL wrote it.

[00:12:56] So unless you have a Next, you can’t go there anyway.

[00:12:59] Right.

[00:12:59] Or you can try to use a text form.

[00:13:01] But they tried to address that by releasing the www.commonlibrary and sort of encourage people to need to build browsers.

[00:13:08] And a bunch of folks volunteered to start making the first generation of browsers.

[00:13:11] One of them will be Marc Andreessen.

[00:13:13] The first browser I used was called Cello.

[00:13:16] There you go.

[00:13:17] Do you remember Cello?

[00:13:18] It was before Netscape.

[00:13:19] Yeah, yeah.

[00:13:20] It’s before all of them.

[00:13:20] Actually, it was Mosaic.

[00:13:21] Mosaic before Netscape, yeah.

[00:13:23] And yeah, that’s the one that Andreessen will be involved in out of the Supercomputing Center in Arborena.

[00:13:27] Yeah, yeah.

[00:13:27] The National Science Foundation, which now owns that network, because it’s the military’s ARPA that’s moved on.

[00:13:33] It’s now the NSF.

[00:13:34] But this is the year that NSF announced the restrictions, removing the restrictions on commercial utilization of the internet.

[00:13:41] So arguably, this is the catalyst, although things will go crazy until Netscape comes along.

[00:13:47] But these are the elements of making what we now know as the internet.

[00:13:50] I remember.

[00:13:51] I remember there was a lot of people complaining about the commercialization of the internet, you know, on the newsgroups and all in the.

[00:13:57] Oh, yeah.

[00:13:57] Even on CompuServe before that, you know, they were really worried that the commercial thing.

[00:14:03] However, porn was totally cool, right?

[00:14:06] That’s funny.

[00:14:07] You know, porn drove the internet.

[00:14:08] This is also the year that one young programmer named Linus Torvalds releases a version of Unix.

[00:14:16] He calls Linux the kernel on Usenet.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:19] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:20] Yeah.

[00:14:21] Yeah.

[00:14:26] Microsoft releases a product called Visual Basic.

[00:14:30] That’s really funny because when I started working with it, and Andrew, you were a VB programmer too.

[00:14:37] I wonder how they got through passport control.

[00:14:39] Were you in there in the beginning, 1.0?

[00:14:42] No.

[00:14:42] No.

[00:14:43] I have a few fewer gray hairs than you, Carl.

[00:14:48] A few.

[00:14:49] My first.

[00:14:49] It looks brown to me.

[00:14:51] Mine looks practically white.

[00:14:54] Sorry for the segue.

[00:14:55] I was putting sun cream on my face

[00:14:57] and I thought I hadn’t rubbed all the sun cream in

[00:15:00] and then I realized my beard is just turning white.

[00:15:02] There’s a bit of gray on the sides there.

[00:15:06] That’s how it starts, my friend.

[00:15:08] I was a VB4 developer.

[00:15:10] My first ever programming language was BASIC on a ZX Spectrum

[00:15:15] and then I progressed from that to VB4.

[00:15:18] Nice.

[00:15:18] I remember VB1, I was a happy clipper programmer

[00:15:22] making lots of money.

[00:15:23] Business was good, but people kept wanting Windows stuff.

[00:15:25] I didn’t understand why.

[00:15:26] I thought taking your hand off the keyboard was stupid.

[00:15:30] But with Windows 3 and device contexts, that was the thing.

[00:15:34] Yeah, DCs.

[00:15:36] Trying to write MFC apps and crashing Windows all the time sucked.

[00:15:42] So VB looked to be a window through there.

[00:15:45] Although I didn’t have a lot of love for one.

[00:15:46] One was pretty primitive.

[00:15:47] You couldn’t do a whole bunch.

[00:15:48] I got it.

[00:15:49] Two would be the version that I’m like, okay, this is the way.

[00:15:52] I liked one just because it was the proof of concept, you know?

[00:15:55] And you couldn’t really do anything with it.

[00:15:58] It was, you know.

[00:15:59] No, it was just the beginning.

[00:16:00] Just the beginning.

[00:16:01] This is the year that Microsoft releases MS-DOS 5.0.

[00:16:04] Yeah.

[00:16:04] And also the separation from IBM is complete

[00:16:07] and they stopped calling their Microsoft OS 2.

[00:16:10] They start calling it Windows NT.

[00:16:12] Dave Cutler.

[00:16:13] What else?

[00:16:14] The PowerPC processor, a joint development between IBM, Motorola, and Apple

[00:16:18] for the new Power Mac is released.

[00:16:20] I don’t know whether to feel a kinship with you guys

[00:16:22] or just sad at how old I am, you know?

[00:16:26] This video call is like a live demo of entropy, isn’t it?

[00:16:30] Yeah.

[00:16:34] Beautiful.

[00:16:35] All right, you can stay.

[00:16:38] Creative Labs releases the Multimedia Kit.

[00:16:41] Remember this?

[00:16:41] It comes with your Sound Blaster Pro.

[00:16:45] It’s also this thing called a CD-ROM drive.

[00:16:48] So this is the beginning of multimedia on your computer.

[00:16:51] Yeah, that’s right.

[00:16:52] One that I particularly love and remember at the time

[00:16:55] was a new game for the Sega Genesis called Zero Wing.

[00:16:58] And it is the origin of the phrase,

[00:17:02] all your base are belong to us.

[00:17:04] Oh, all your base are belong to us.

[00:17:08] Yeah.

[00:17:08] This is where it comes from.

[00:17:09] It’s this terrible Japanese translation of this story of a…

[00:17:13] Somebody made a…

[00:17:14] I think we did it on .NET Rocks a long time ago.

[00:17:17] Jeff Meeseel.

[00:17:17] Huge.

[00:17:17] Maybe play.

[00:17:18] That there was like a music loop that was done with all your base are belong to us.

[00:17:23] It was like a techno kind of thing.

[00:17:25] It was a techno thing.

[00:17:26] Yeah.

[00:17:26] That’s where it comes from, 1991.

[00:17:28] This is also the year that id Software and is formed.

[00:17:31] Their first game is called Commander Keen,

[00:17:33] but you know about them for Doom and Quake and all of those games.

[00:17:37] And a company called Silicon and Synapse, which now we know as Blizzard.

[00:17:41] There you go.

[00:17:41] You care about World of Warcraft or any of those lines of games.

[00:17:44] All right, that’s what I got.

[00:17:45] All right, let’s move on to Better Know a Framework.

[00:17:48] Roll the music.

[00:17:49] Awesome.

[00:17:54] All right, man.

[00:17:58] What do you got?

[00:17:58] All right.

[00:17:59] So I just recorded an episode of Code It With AI today with Jeff Fritz.

[00:18:05] And he pulled out this thing called Squad.

[00:18:08] And Squad is a tool by Brady Gaster.

[00:18:11] Let me read it from the repo.

[00:18:13] Yep.

[00:18:14] Squad gives you an AI development team through GitHub Copilot.

[00:18:18] Describe what you’re building.

[00:18:20] Get a team of specialists, front-end, back-end testers that live in your repo as files.

[00:18:27] Wow.

[00:18:27] They persist across sessions.

[00:18:28] One of them, their job is just to remember things and write it down.

[00:18:34] They persist across sessions, learn your code base, share decisions,

[00:18:38] and get better the more you use them.

[00:18:41] It’s not a chatbot wearing hats.

[00:18:43] Each team member runs in its own context, reads only its own knowledge,

[00:18:48] and writes back what it learned.

[00:18:50] Get this.

[00:18:51] It will write skills.

[00:18:53] It’s like really smart.

[00:18:56] And I saw a demo of that with Jeff Fritz today.

[00:19:00] It’s Code It With AI episode 18, and we’ll provide a link to that website.

[00:19:05] And it just blew my mind.

[00:19:07] Here’s the other thing that blew my mind.

[00:19:09] I said, isn’t this with all these multiple agents going to rack up a lot of money, a lot of tokens?

[00:19:16] He says, yes.

[00:19:16] But my Copilot.

[00:19:17] License is not by the token.

[00:19:20] It’s by the request.

[00:19:22] Interesting.

[00:19:22] And so, yeah, it turned out that he did this demo of adding, oh, what did he add?

[00:19:31] He added a testing framework.

[00:19:32] He added a SQL expert to analyze the code that was generating SQL.

[00:19:37] This is for AVN Data Genie.

[00:19:39] He did this.

[00:19:39] He added a GitHub CICD pipeline, all of this stuff.

[00:19:47] Only took 18 requests, and that’s because it was really only six requests, but it was a three times license, essentially.

[00:19:58] The thing that he was using was a 3X agent, essentially.

[00:20:04] So, yeah, it only took 18 tokens.

[00:20:06] We did the math, and it turned out to be about a buck.

[00:20:09] Wow.

[00:20:09] So, it’s very cost efficient if you use it the right way.

[00:20:13] Okay.

[00:20:13] Yeah, I’m very impressed.

[00:20:14] Well, it sounds like you should be at .NET Rocks about it at some point.

[00:20:17] I think there will be, as a matter of fact.

[00:20:19] Yeah.

[00:20:19] We just had Brady on the show, too.

[00:20:21] But, you know, we’ll figure it out.

[00:20:23] Very cool.

[00:20:24] So, anyway, that’s what I got.

[00:20:25] Who’s talking to us, Richard?

[00:20:26] Grabbed a comment off a show, 1989.

[00:20:30] So, that’s the one we just did with Ben when we talked about the role of AI.

[00:20:34] Lit off a whole bunch of comments that I thought we should address.

[00:20:37] This is from a longtime listener.

[00:20:38] John Suda has been listening to the show for a very long time, so I grabbed his comment.

[00:20:42] He said, hey, not so long ago, you guys mocked the concept of vibe coding.

[00:20:46] Sorry, dude, I still do.

[00:20:47] If you’re talking about the original concept, which, you know what, Andrea, we talked about,

[00:20:51] where you don’t adjust any of your own code, you just keep feeding it back and so forth.

[00:20:55] But now it’s down to, and I’m paraphrasing, only the hobbyists will soon read, let alone write code.

[00:21:00] Wow.

[00:21:01] Who said that?

[00:21:02] Well, I’m not sure, but this is from that conversation of you now managing a bunch of agents generating code, right?

[00:21:10] Although I think…

[00:21:11] Well, I didn’t say that.

[00:21:13] I think maybe…

[00:21:14] No, I didn’t think you did.

[00:21:15] But there is that point…

[00:21:16] Maybe the guest said that.

[00:21:17] Which you do in agent management, and then there’s also this point of there’s some things that the agents just aren’t good at.

[00:21:22] Right.

[00:21:22] At least for now.

[00:21:23] One thing I said in this conversation with Jeff Fritz on Squad is that, you know, you have to be more vigilant than ever because we have this…

[00:21:35] Now that we’re doing, we’re in this role of, you know, just barking out commands and letting things happen, we have a moral obligation to double check, triple check that code.

[00:21:47] Right.

[00:21:47] Because our name’s on it, you know what I mean?

[00:21:50] And if the company comes back and they’re looking to people to blame, you can’t say, oh, the AI did that.

[00:21:57] You can’t…

[00:21:57] No, you’re responsible for it.

[00:22:00] You can, they’re just not going to take it.

[00:22:02] You’re responsible for it.

[00:22:03] So, and people can get hurt.

[00:22:05] That’s the thing about it.

[00:22:06] Like, you really have to think about this as a moral issue, I think.

[00:22:10] Anyway.

[00:22:10] Yeah, I don’t disagree.

[00:22:11] And this is basically what John’s saying.

[00:22:13] It’s like the pace at which things are happening scares me at times.

[00:22:15] We almost fatalistically accept.

[00:22:17] It is inevitable.

[00:22:17] But is it always an unmitigated good?

[00:22:19] No.

[00:22:20] Yeah.

[00:22:20] And we’re not fatalistically accepting it.

[00:22:22] We’re testing it and trying it and trying to figure out if it’s valuable and understanding the issues around it.

[00:22:27] Like…

[00:22:27] Right.

[00:22:28] These things are in front of us.

[00:22:29] And my first reaction was, show me the money.

[00:22:32] Like, does this work?

[00:22:33] Right.

[00:22:33] And how do you use it responsibly?

[00:22:35] And the genie is not going back in the bottle.

[00:22:37] Like, you know, it’s only ever going to get better.

[00:22:40] It’s not going to get worse.

[00:22:42] And, you know, we just can’t undo what’s being done.

[00:22:44] We have to find a way to deal with it.

[00:22:46] Although the scope…

[00:22:47] The scope of badness is still fully yet to be fully explored.

[00:22:50] There’s more harm to be done without a doubt.

[00:22:52] Right.

[00:22:53] So, and he goes on to say, change is good.

[00:22:55] There’s no two ways about it.

[00:22:56] But sometimes I wonder, can there be preferably too much of a good time?

[00:22:59] Absolutely.

[00:22:59] Too much of a good thing?

[00:23:00] Absolutely.

[00:23:01] Right.

[00:23:01] And that’s part of why we keep digging into these things is because I am concerned.

[00:23:06] Right.

[00:23:06] And we do have real goals to make.

[00:23:08] We’re not just writing software for a front.

[00:23:11] Are you delivering the quality that the customer needs?

[00:23:13] And are you, you know, is your product good enough?

[00:23:15] And do you understand the scope?

[00:23:17] Of what’s been created?

[00:23:18] Right.

[00:23:18] I don’t find this all that different from when we started outsourcing development into other countries

[00:23:22] because the bandwidth was cheap and the labor was cheap.

[00:23:24] And we had similar kinds of problems.

[00:23:27] Yes.

[00:23:27] You know, with quality and understanding and so forth.

[00:23:31] Communication.

[00:23:32] Yeah.

[00:23:32] The whole thing.

[00:23:33] But, you know, I’m not going to argue with Andrew.

[00:23:35] Genie’s definitely out of the bottle.

[00:23:37] And we just keep on looking at it.

[00:23:38] I mean, last year was the year we were often tired of talking about AI.

[00:23:42] Yeah.

[00:23:43] This year is the year we seem to have a stack of results sufficient to say,

[00:23:46] okay,

[00:23:47] have we got here and what are the right things to do?

[00:23:49] Right.

[00:23:50] Well, at the same time, feeling like I really don’t need the hype anymore.

[00:23:53] I just need to get to work.

[00:23:55] Right.

[00:23:55] It’s finding the signal in the noise.

[00:23:57] That’s what it is.

[00:23:58] Like, there’s so much stuff out there.

[00:24:01] Some of which is absolute and complete horse fodder.

[00:24:06] Yeah.

[00:24:07] Yeah.

[00:24:08] But, you know, some of it is meaningful and different.

[00:24:11] And, you know, we need to find that signal in the noise.

[00:24:14] Yeah.

[00:24:14] So maybe that’s the thing.

[00:24:15] Last year, we were trying to figure out if there was signal.

[00:24:17] Now we’re pretty darn sure.

[00:24:18] And we’re just scraping away the noise to focus on what actually is real here,

[00:24:23] for better or worse.

[00:24:24] Yeah.

[00:24:24] Right.

[00:24:25] John, I’m pretty sure you have a Coffee Music Code by,

[00:24:27] but you’re always welcome to another.

[00:24:29] Just send me an email and we’ll figure it out.

[00:24:32] So thank you so much for your comment.

[00:24:34] And if you’d like Coffee Music Code by,

[00:24:36] write a comment on the website at .netrocks.com or on the Facebooks.

[00:24:39] We publish every show there.

[00:24:40] And if you comment there and I read it on the show,

[00:24:42] we’ll send you Coffee Music Code by.

[00:24:43] And if you want to just go get music to code by,

[00:24:45] we have 22 tracks.

[00:24:47] They’re 25 minutes long to keep you in a state of flow and focus

[00:24:51] while you’re writing code.

[00:24:52] And that’s at musictocodeby.net.

[00:24:54] And we have them in MP3, FLAC, and WAV formats.

[00:24:59] Okay.

[00:25:00] Should we finally get to our guest?

[00:25:02] I feel embarrassed here.

[00:25:03] It’s like.

[00:25:03] I’m going to get this, yes.

[00:25:04] This show has taken a real turn for, you know, that’s 20.

[00:25:08] What is it?

[00:25:08] 25 minutes so far?

[00:25:10] Something like that.

[00:25:10] I’m here for it.

[00:25:11] I’m here for it.

[00:25:12] All right.

[00:25:13] Well, we’re going to cram it in.

[00:25:14] Andrew Murphy started his career as a VB programmer,

[00:25:17] but after almost two decades in technology leadership,

[00:25:19] he decided to focus on teaching the skills that he learned the hard way.

[00:25:24] When he moved into leadership, there was no support.

[00:25:27] So he had to make all the mistakes and a lot of them and learn from them.

[00:25:32] His goal is now to make sure that other tech leaders don’t have to do things

[00:25:36] the hard way and to make them happy, confident, and effective leaders.

[00:25:41] All right.

[00:25:41] A formal welcome to .NET Rocks, Andrew.

[00:25:44] Thank you.

[00:25:45] Great to be here.

[00:25:46] Great to have you.

[00:25:46] Thank you.

[00:25:47] So what are you thinking besides all this AI stuff just, you know, flooding your brain?

[00:25:52] Overwhelming the conversation.

[00:25:54] Yeah.

[00:25:55] So that’s kind of what I wanted to talk about is, you know,

[00:25:58] I run this meetup event, Brains Trust type thing in Australia called CTO School.

[00:26:06] And I see it as my job as host to kind of, you know, curate the topics of conversation

[00:26:12] and, you know, just kind of keep things interesting.

[00:26:16] Mm-hmm.

[00:26:16] And for the past year, I have tried every single month to not have AI dominate the agenda.

[00:26:23] And I have failed every single month.

[00:26:25] And I’ve just come to an acceptance in the past kind of, I don’t know, quarter or so

[00:26:31] that it is just going to be an important part of what we do as professionals and especially as leaders.

[00:26:37] And so, you know, I was doing a bunch of research on it.

[00:26:41] And I came across this blog post by a guy called Nolan Lawson.

[00:26:45] I don’t know if I have the…

[00:26:46] Whether you have read it.

[00:26:47] It’s called We Mourn Our Craft.

[00:26:50] And it’s…

[00:26:50] Oh, yeah.

[00:26:51] I’ve read that.

[00:26:51] Yeah.

[00:26:52] It’s incredible writing.

[00:26:53] It’s absolutely beautiful.

[00:26:56] It’s gorgeous.

[00:26:57] Honest.

[00:26:58] It’s probably the best thing that I’ve read about what it actually feels like to be a software engineer right now in the industry.

[00:27:05] Wow.

[00:27:06] And it talks about a whole bunch of stuff.

[00:27:10] But I just want to quote one line from it because I think it kind of sums up exactly…

[00:27:16] Exactly the feeling in the industry.

[00:27:19] So this is him talking about acceptance of where we are with AI.

[00:27:21] I don’t celebrate the new world, but I also don’t resist it.

[00:27:26] The sun rises, the sun sets, and I orbit helplessly around it.

[00:27:30] My protest can’t stop it.

[00:27:31] It doesn’t care.

[00:27:33] It continues its arc across the sky regardless, moving but unmoved.

[00:27:38] Beautiful poetry.

[00:27:40] Absolutely beautiful poetry.

[00:27:42] But I disagree with Nolan.

[00:27:46] I think there’s a whole bunch of changes that we’ve had in our industry.

[00:27:52] And this is one of them.

[00:27:54] It’s going to be a fundamental shift in the way we write code.

[00:27:59] But so was a whole bunch of other things that happened in the industry.

[00:28:02] That’s right.

[00:28:03] You know, all three of us remember the days before IntelliSense, you know, when you actually had to remember the names of methods.

[00:28:12] That’s right.

[00:28:13] When the cloud emerged, I got email.

[00:28:16] I got emails on the run-ass side from folks mourning spinning screwdrivers, racking and stacking.

[00:28:22] You know, stuff I used to do, too.

[00:28:24] And we really took a lot of pride from building out of 42.

[00:28:27] It was a great gig.

[00:28:28] It was a day of work.

[00:28:31] It was fun.

[00:28:32] And it’s over.

[00:28:33] Yeah.

[00:28:34] Yeah.

[00:28:34] And I think there’s so much of this that has happened in our industry that, you know, this could be a fundamental shift.

[00:28:43] But it’s not going to get rid of, I think, the core.

[00:28:46] It’s going to get rid of the core of what it means to be a software engineer.

[00:28:48] And so that’s why I wrote my article in response to Nolan’s around, you know, looking at that grief that Nolan identified through the common model everybody has heard of grief, which is the five stages of grief.

[00:29:01] And I called it the five stages of losing our craft because it is grief.

[00:29:06] We are losing a part of what it means to be a software engineer where our identity is being challenged.

[00:29:13] But, you know, the reason why the five stages.

[00:29:16] So grief exists as a model is because you don’t just give up.

[00:29:21] You know, if you if you’ve lost a loved one, the wallowing in that in that grief is not productive.

[00:29:27] It’s moving through it and looking at what it means.

[00:29:30] And I wanted to kind of look at the grief that Nolan identified through the angle of the five stages of grief.

[00:29:37] Sure.

[00:29:37] Yeah.

[00:29:37] We all have to move on.

[00:29:39] That’s what it’s all about.

[00:29:40] And if you don’t, you know, then go work at Walmart.

[00:29:45] Yeah.

[00:29:46] Yeah.

[00:29:46] I mean, it’s a valid option.

[00:29:48] Is it?

[00:29:49] I don’t know that it is, but OK.

[00:29:54] Well, the collective and I say collective personally, the knowledge that I’ve collected about systems and about customers and about what they want, what they need and users and stuff is all really valuable.

[00:30:09] And an AI just doesn’t have that.

[00:30:10] That’s it.

[00:30:11] And especially especially in the domains that I’ve worked in, which tend to be.

[00:30:16] The domains that I keep working in, you know, people will hire you because you have that experience and that knowledge.

[00:30:24] Yeah.

[00:30:24] And, you know, it’s it’s it’s not just about building great system architecture.

[00:30:30] It’s also about knowing when not to do that.

[00:30:33] Like one of the things that I’ve I feel like I’ve learned and honed over my career is knowing when the hacky solution is the right solution and when it’s worth while investing in building a great solution.

[00:30:46] And that’s something that I have historically just seen AI fail and fall flat on its face on doing is should this be the the big architected solution or should this be something I just I just hacked together?

[00:31:00] I think those decisions are something that you can only only learn and sharpen from from doing this.

[00:31:06] And, you know, AI will probably get there in 10, 20, 30 years.

[00:31:11] But, you know, the difference between the people who have jobs in 20 years and those who.

[00:31:16] Who don’t or be the people who understand that that’s the value that they had.

[00:31:20] Ken Beck wrote an article on this on his sub stack where he said, I just realized that the value of 90 percent of my skills has gone to zero and the value of the remaining 10 percent has 1000 X.

[00:31:37] The job I have now is working out what that 10 percent is.

[00:31:42] Yeah.

[00:31:42] Right.

[00:31:42] That’s good.

[00:31:43] Yeah.

[00:31:44] You’re talking about Kent Beck, the father of extremism.

[00:31:46] Programming and all that.

[00:31:47] Yes.

[00:31:47] Yeah.

[00:31:48] Yeah.

[00:31:48] Yeah.

[00:31:49] He came to Yao in Australia and did the I can’t remember if it’s a keynote or the lock note there.

[00:31:54] And he talked about it.

[00:31:56] And it’s yeah, it’s it’s I think it’s really powerful because you’re you’re right, Kyle.

[00:31:59] All of those things around understanding the customer.

[00:32:03] This is this is one of the things early in my career.

[00:32:05] I was really, really lucky in that I worked for a company.

[00:32:09] They manufactured fireplaces.

[00:32:11] It doesn’t necessarily matter what they did.

[00:32:13] But I had a great boss.

[00:32:15] At the time, and what he would do is every time I had to write code to help somebody with their job, he made me go do their job for a day or so.

[00:32:26] Right.

[00:32:27] So if if if I was writing this, the the warehouse management software, I would be in my overalls in the warehouse doing a stock tech.

[00:32:35] Yeah.

[00:32:35] You had to go out and feel the pain.

[00:32:37] Exactly.

[00:32:38] And empathize and understand the users.

[00:32:41] And AI is never going to do that.

[00:32:44] You know, it can’t.

[00:32:45] Sure.

[00:32:45] At least.

[00:32:45] Any any time soon.

[00:32:47] And I think those are the pivotal moments in people’s careers that help you really understand that the job was never about producing the code.

[00:32:56] The job is about understanding the problems that need to be solved and how software can help solve those problems.

[00:33:02] Yeah, sure.

[00:33:03] Yeah.

[00:33:04] Yeah, you’re right.

[00:33:05] We’ve been through this before.

[00:33:06] Yeah, exactly.

[00:33:07] No, we should be writing our own garbage collector or encryption libraries or, you know, they they we’ve the pieces of our job have been peeled off for a while.

[00:33:13] This just seems like a big one.

[00:33:15] Yeah.

[00:33:15] It’s very interesting.

[00:33:23] Yeah.

[00:33:23] Because, you know, if you’ve got a senior engineer who is resisting this transition, then it’s it’s, you know, a problem for them.

[00:33:30] If you’ve got a CTO who’s resisting this transition, then it’s a problem for the whole company.

[00:33:36] Yeah.

[00:33:37] And it becomes existential for whichever organization you’re in.

[00:33:42] And, you know, that’s the thing that I’m worried the most about is, you know, if you don’t have the right people, then you can’t do it.

[00:33:45] if you don’t have those leaders who can kind of champion this change,

[00:33:49] and I’m not talking about it, you know, I’m, I’m not,

[00:33:53] I’d say I’m an AI moderate in that, you know,

[00:33:56] I’m not an AI optimist and I’m not an AI pessimist.

[00:33:58] I’m more of an AI realist and, you know, I’m not banging the,

[00:34:02] the drum of, of, of, you know,

[00:34:04] this is the best thing that ever happened and you’ve got to jump on the

[00:34:06] bandwagon or, you know, you’re going to lose your jobs.

[00:34:09] But like we said earlier, the genie is out of the bottle and it’s,

[00:34:12] it’s less about, you know, what,

[00:34:15] what’s the existential threat to the industry and more about what can you do

[00:34:19] in the next few years to make sure you’re, you know,

[00:34:22] catering with what’s happening and dealing with the industry.

[00:34:24] How can we use it smartly?

[00:34:26] Exactly.

[00:34:27] Right.

[00:34:27] And not just go off the deep end.

[00:34:30] Yeah.

[00:34:30] And, you know,

[00:34:31] one of the mistakes I’ve seen some leaders make in this area is like,

[00:34:35] okay, well let’s do a research project.

[00:34:37] Let’s, you know,

[00:34:37] let’s spend six months finding out where, you know,

[00:34:40] what we can use it for and how we can use it.

[00:34:42] And then they get to the end of those six months and they decided what their

[00:34:45] AI policy is.

[00:34:47] And then they realized that where they were making decisions based on where AI

[00:34:50] was six months ago.

[00:34:52] Right.

[00:34:52] Yeah.

[00:34:54] So,

[00:34:54] I mean,

[00:34:55] I saw folks do that with mobile in the back of the day as well.

[00:34:58] Right.

[00:34:58] And if you,

[00:34:58] whatever tools were in front of you at the time,

[00:35:01] especially,

[00:35:01] you know,

[00:35:01] you think about 2009,

[00:35:03] 2010,

[00:35:03] the early days of mobile of,

[00:35:05] you know,

[00:35:05] the iPhone,

[00:35:06] you took those six months by the time you come to the other side of it,

[00:35:10] everything you’re doing was a waste of time.

[00:35:11] The tools have totally changed.

[00:35:12] Yeah.

[00:35:13] Yeah.

[00:35:14] Yeah.

[00:35:14] I think the way you have to think about it is,

[00:35:16] is kind of what we said earlier,

[00:35:19] which is you’re responsible for the code that ends up in the code base.

[00:35:23] Sure.

[00:35:23] You know,

[00:35:23] it doesn’t matter how that code ended up in the code base.

[00:35:26] You could have,

[00:35:27] you know,

[00:35:27] handcrafted it while drinking a Zinfandel or you could have copied it from

[00:35:32] stack overflow or an AI could have generated it.

[00:35:36] Or you asked your,

[00:35:37] you know,

[00:35:37] your junior engineer to write it.

[00:35:39] Whatever the reason is that the code ended up in your code base,

[00:35:42] you’re still responsible and accountable for it.

[00:35:45] Yeah.

[00:35:45] And,

[00:35:45] you know,

[00:35:45] I think focusing on that side of things and building quality controls into,

[00:35:51] you know,

[00:35:51] into how you generate the code and what ends up in prod,

[00:35:54] that’s the answer.

[00:35:56] Not these big research projects that,

[00:35:58] you know,

[00:35:58] are just rapidly out of date.

[00:36:00] Are you finding that teams like the junior people are okay?

[00:36:04] They’re embracing the new tools fine.

[00:36:06] Like we always stood talking about,

[00:36:07] this is going to wipe out juniors.

[00:36:09] Yeah.

[00:36:09] So I,

[00:36:10] I think there’s an interesting,

[00:36:12] interesting debate in this.

[00:36:13] And I don’t know where I kind of fall down on this line because there’s,

[00:36:19] there’s two,

[00:36:19] there’s two ways you could look at that problem.

[00:36:21] You could look at that pop the problem and go,

[00:36:24] okay,

[00:36:24] well,

[00:36:24] you know,

[00:36:25] juniors are irrelevant because the AI is just the junior.

[00:36:29] Or you could look,

[00:36:30] you could look at it from the other lens and you could go,

[00:36:33] well,

[00:36:33] actually we don’t need senior engineers anymore because a junior with AI is a

[00:36:37] senior engineer.

[00:36:39] And I think,

[00:36:39] I think it’ll be interesting to see where we end up.

[00:36:42] But I’m,

[00:36:43] I’m seeing a lot of juniors just jumping on this,

[00:36:47] you know,

[00:36:47] with two hands and just,

[00:36:49] just grasping it and running circles around the seniors in terms of code

[00:36:53] output.

[00:36:55] Now,

[00:36:55] again,

[00:36:55] you can ask the questions on.

[00:36:57] Yes,

[00:36:58] exactly.

[00:36:58] Exactly.

[00:36:59] And,

[00:36:59] and I think that’s,

[00:37:01] that’s the,

[00:37:01] you know,

[00:37:02] that’s the thing that is the differentiator is a senior that knows how to

[00:37:06] use these tools is going to produce so much better outcomes.

[00:37:10] Although a junior,

[00:37:11] you know,

[00:37:11] is probably going to produce a lot more code.

[00:37:14] Well,

[00:37:15] this seems like a good place to take a break.

[00:37:16] So we’ll be right back after these very important messages.

[00:37:21] Hey,

[00:37:22] Carl here.

[00:37:22] You probably know text control is a powerful library for document editing and

[00:37:27] PDF generation,

[00:37:28] but did you know they’re also a strong supporter of the developer community?

[00:37:32] It’s part of their mission to build and support a strong developer community

[00:37:37] by being present,

[00:37:38] listening to users and sharing knowledge at conferences,

[00:37:41] across Europe and the United States.

[00:37:44] So if you’re heading to a conference soon,

[00:37:46] check if text control will be there and stop by to say,

[00:37:50] hi,

[00:37:50] you can find their full conference calendar at www.textcontrol.com and make

[00:37:57] sure you thank them for supporting.net rocks.

[00:38:05] And we’re back.

[00:38:06] It’s.net rocks.

[00:38:07] I’m Carl Franklin.

[00:38:08] That’s my friend,

[00:38:09] Richard Campbell.

[00:38:09] And that’s our friend,

[00:38:11] Andrew Murphy.

[00:38:11] And we’re talking about AI leadership and all that goes with it.

[00:38:15] The challenges,

[00:38:16] all the fun things,

[00:38:17] the,

[00:38:17] the,

[00:38:19] the thrill of victory and the agony of defeat.

[00:38:23] There’s plenty of senior people that have changed tools enough time.

[00:38:26] They’re like,

[00:38:26] Oh,

[00:38:26] we’re changing tools again.

[00:38:27] And they just dive in.

[00:38:29] Yeah.

[00:38:29] But there’s a certain group of people that are resisting and I’m trying to

[00:38:31] figure out who they are.

[00:38:32] I genuinely think that they’re the people like me who didn’t make a

[00:38:40] transition into leadership.

[00:38:41] And what I mean by that is like,

[00:38:43] I’m a nerd and a geek.

[00:38:45] I learned to code on a ZX spectrum.

[00:38:47] My dad taught me how to code.

[00:38:50] I’ve,

[00:38:51] I’ve been coding since I was eight years old and I just love it so much,

[00:38:55] but you know,

[00:38:56] I was kind of thrust into this leadership role without really wanting it.

[00:39:01] If I hadn’t have done that,

[00:39:02] I could see myself being one of those people because I just love coding so

[00:39:07] much.

[00:39:08] Like I get a huge amount of joy in the creative,

[00:39:11] the education aspect of it and the craft aspect of it.

[00:39:15] So I have,

[00:39:16] I have empathy for those people because I,

[00:39:19] you know,

[00:39:19] in another life I would have been them.

[00:39:22] Uh,

[00:39:22] but you know,

[00:39:23] it’s,

[00:39:23] it’s kind of similar to all the skills that have,

[00:39:27] have,

[00:39:28] have disappeared,

[00:39:29] you know,

[00:39:29] throughout the years we’ve as you know,

[00:39:32] we know not many people weave by hand anymore.

[00:39:36] We use looms to do it.

[00:39:38] And then,

[00:39:38] you know,

[00:39:39] we use automated looms.

[00:39:40] People still do that stuff for fun because it is insanely enjoyable.

[00:39:46] But you know,

[00:39:47] it’s not,

[00:39:48] it’s,

[00:39:48] you can’t build an industry off doing it by hand anymore.

[00:39:51] No,

[00:39:51] you’re not making a living from it is tough.

[00:39:54] Yeah,

[00:39:54] exactly.

[00:39:55] And if you do,

[00:39:56] it’s going to be some hugely specific niche thing where people are willing to,

[00:40:01] to pay for,

[00:40:02] you know,

[00:40:03] the fact that this has had human hands on it.

[00:40:05] Um,

[00:40:05] and there might be something like that in our industry.

[00:40:08] Like I could see,

[00:40:09] you know,

[00:40:10] uh,

[00:40:10] a certain group of people going,

[00:40:12] let’s handcraft this thing for fun.

[00:40:14] Just like people go,

[00:40:15] let’s assemble a car for fun.

[00:40:16] Like it’s,

[00:40:17] it’s,

[00:40:18] it’s,

[00:40:19] you know,

[00:40:19] it’s a hobby.

[00:40:19] It’s going to be a niche.

[00:40:20] Yeah.

[00:40:21] I wonder if digital handcrafted digital is going to be a thing as opposed to

[00:40:25] handcrafted physical.

[00:40:26] Like that’s certainly a thing.

[00:40:27] Yeah.

[00:40:28] Yeah.

[00:40:29] Yeah.

[00:40:29] Handcrafted digital.

[00:40:32] And then how do you prove it?

[00:40:33] Like,

[00:40:33] that’s the interesting thing.

[00:40:34] Like how do you,

[00:40:35] how do you prove that it,

[00:40:37] that it wasn’t generated by an AI?

[00:40:39] Yeah.

[00:40:39] Yeah.

[00:40:39] Yeah.

[00:40:39] You have to make me,

[00:40:40] you have to make typos.

[00:40:41] There has to not be comments in some areas.

[00:40:44] There you go.

[00:40:45] I don’t pull requests.

[00:40:46] Can’t be too coherent.

[00:40:47] I don’t know when,

[00:40:48] but I’m going to make a prediction.

[00:40:49] And I think I have said this before that.

[00:40:51] I don’t know if it’s five years,

[00:40:53] 10 years,

[00:40:54] but pretty soon,

[00:40:55] uh,

[00:40:56] everybody’s going to be sick of AI generated content and talking to bots

[00:41:01] and all that stuff.

[00:41:02] And there’s going to be a revolution in human to human contact activity,

[00:41:08] the arts.

[00:41:10] Um,

[00:41:10] going to see real people perform,

[00:41:12] whether it’s dance or,

[00:41:13] or music or,

[00:41:15] you know,

[00:41:15] whatever.

[00:41:17] And I think that that’s,

[00:41:18] that’s going to be revolutionary.

[00:41:19] I don’t know when it’ll happen,

[00:41:20] but I think that’s the next turnover event that will happen after all this.

[00:41:26] I’ve been thinking so many,

[00:41:27] something similar about SAS as a,

[00:41:30] as an industry,

[00:41:31] you know,

[00:41:31] the,

[00:41:32] when,

[00:41:32] when the cost to produce SAS reduces almost to zero SAS as an industry is

[00:41:38] going to really struggle.

[00:41:39] Yeah.

[00:41:40] You know,

[00:41:40] when you’ve got companies that are paying thousands,

[00:41:43] tens of thousands of dollars a month for Trello and you know,

[00:41:47] you give an engineer a weekend and they can basically build Trello.

[00:41:51] What’s going to be the differentiator?

[00:41:53] I think it’s going to be taste.

[00:41:56] Taste is,

[00:41:57] is going to be the new features.

[00:41:58] You don’t,

[00:41:59] you don’t choose a piece of software because it can do something.

[00:42:03] You choose a piece of software because it feels a certain way,

[00:42:06] acts a certain way and it kind of fits your brain,

[00:42:10] better.

[00:42:10] This is kind of similar to the proliferation of like to do managers and that kind

[00:42:15] of stuff.

[00:42:15] You know,

[00:42:15] there’s a thousand to do lists,

[00:42:17] but the one you choose is not the one that has the features you want.

[00:42:20] It’s the one that kind of fits the way your brain works.

[00:42:23] Yeah.

[00:42:23] Well,

[00:42:24] and you’re right,

[00:42:24] not everybody has a strong opinion about all things like we were seeing with

[00:42:27] tools like cloud bot and the like,

[00:42:30] the emergence of this idea of surrounding yourself with custom software that

[00:42:33] there is no,

[00:42:34] there is no definitive software per se.

[00:42:37] Everything will be unique to you,

[00:42:38] but most,

[00:42:39] most people don’t have that.

[00:42:40] Strong of opinions on that many things.

[00:42:41] So,

[00:42:43] you know,

[00:42:43] the same way in theory,

[00:42:44] we could all make our own clothes or custom all as our,

[00:42:47] all our clothes,

[00:42:48] but people don’t,

[00:42:48] most people don’t care enough.

[00:42:50] We look to these influencers and these thought leaders in that space.

[00:42:55] I hate that phrase,

[00:42:55] but you know what I mean?

[00:42:56] Yeah.

[00:42:57] Yeah.

[00:42:57] That there will be a taste element to this.

[00:43:00] Did you guys see the movie her?

[00:43:02] Sure.

[00:43:02] Yes.

[00:43:02] Yeah.

[00:43:03] So what did you think the main message of that movie was?

[00:43:07] Was there a message or was it just to kind of enjoy a little romp?

[00:43:10] I mean,

[00:43:10] it was a great exploration of a digital assistant and they,

[00:43:15] you know,

[00:43:15] the,

[00:43:15] you know,

[00:43:16] the what’s the term now?

[00:43:19] It’s just jumped out of my head that that social stratum of,

[00:43:23] you know,

[00:43:24] you projected your feelings onto a piece of technology.

[00:43:27] Right.

[00:43:27] And it was kind of an,

[00:43:29] it’s not the same.

[00:43:30] It was kind of a cautionary tale,

[00:43:31] wasn’t it?

[00:43:32] Yeah.

[00:43:32] And we didn’t really have these people getting emotionally involved in chat

[00:43:36] bots,

[00:43:37] but maybe a little bit before that.

[00:43:39] But I mean,

[00:43:39] the author of that movie and that book really saw the future that,

[00:43:45] wow,

[00:43:45] this is a problem and,

[00:43:46] you know,

[00:43:47] this could,

[00:43:47] could really hurt people.

[00:43:50] So I saw,

[00:43:51] I’m just watching,

[00:43:53] I don’t know what I was watching the super bowl or something or the

[00:43:55] Olympics.

[00:43:56] And there was an ad where this woman is woken up by her digital

[00:44:01] assistant.

[00:44:02] And she has this jovial kind of,

[00:44:06] you know,

[00:44:07] a rapport with it,

[00:44:09] that you would have with your partner and she’s waking up alone.

[00:44:13] Right.

[00:44:14] And it it’s celebrating the AI as your,

[00:44:19] you know,

[00:44:19] your,

[00:44:20] your emotional and you know,

[00:44:22] your emotional partner really.

[00:44:24] And I just thought,

[00:44:25] wow,

[00:44:25] how far we’ve come.

[00:44:26] Like,

[00:44:27] you know,

[00:44:27] this is something that people are really looking forward to in there.

[00:44:31] Is it,

[00:44:31] or is it just the person who made that ad?

[00:44:33] Well,

[00:44:35] well,

[00:44:35] they wouldn’t make it if people didn’t respond to it.

[00:44:37] Right.

[00:44:38] Uh,

[00:44:38] I’m not necessarily,

[00:44:39] they’ve got a lot of money.

[00:44:42] Do you know the last time?

[00:44:44] So when you look at all the ads at the super bowl,

[00:44:47] AI dominated it huge.

[00:44:48] Yeah.

[00:44:49] Do you know the last time when technology dominated to the ads at the super

[00:44:53] bowl was 2000,

[00:44:56] right.

[00:44:57] Just before the doctor come boom.

[00:44:59] Yeah.

[00:45:00] So I think that maybe tells you something about,

[00:45:02] you know,

[00:45:02] they’re just spending money trying to grab market share.

[00:45:05] Yeah.

[00:45:06] I think,

[00:45:06] you know,

[00:45:07] that there’s a whole heap.

[00:45:08] I don’t want to get too much,

[00:45:09] in the industry it’s been done to death,

[00:45:11] but there’s so much in,

[00:45:12] in,

[00:45:13] in the,

[00:45:13] where,

[00:45:14] in where we are right now with the investment in AI,

[00:45:17] where it is 100% a solution trying to find the problem.

[00:45:21] Like they’re,

[00:45:22] they,

[00:45:22] they just have this big hammer,

[00:45:24] which is the LLMs and they’re just hitting it everywhere.

[00:45:26] Right.

[00:45:26] And that doesn’t mean that there isn’t a huge value in this stuff.

[00:45:29] There is.

[00:45:31] And,

[00:45:31] you know,

[00:45:31] I think one of the big problems people have in this kind of mental

[00:45:36] transition of how they utilize LLMs,

[00:45:38] is they kind of fixate on the things it does badly,

[00:45:42] rather than the things it does well,

[00:45:43] because there’s so many companies that are just using this LLM hammer in

[00:45:48] places it shouldn’t be used.

[00:45:49] Right.

[00:45:49] Yeah.

[00:45:50] Everyone’s desperate to catch,

[00:45:51] you know,

[00:45:51] this,

[00:45:52] the line I used the other day on,

[00:45:53] on,

[00:45:54] when I was a guest on a show was there’s only so many chances in your

[00:45:58] lifetime to become a billionaire.

[00:46:01] And this is one of them right now.

[00:46:03] And,

[00:46:03] uh,

[00:46:04] and so for a certain group of people,

[00:46:05] like this is your golden ring moment.

[00:46:07] And you will put,

[00:46:08] put aside arguably all ethics of morality for this chance and you’re,

[00:46:13] you’re seeing the consequences of it.

[00:46:15] I do want to get back to your post and this,

[00:46:18] you know,

[00:46:19] you’re leading a team in the midst of this change.

[00:46:23] So what are,

[00:46:26] I mean,

[00:46:26] do you let go of the folks that aren’t willing to embrace this?

[00:46:28] Like,

[00:46:29] I got to know,

[00:46:29] I think there’s teams that are doing that.

[00:46:30] It’s like,

[00:46:31] we’re going this way and those are unhappy.

[00:46:33] There’s the door.

[00:46:33] Yeah.

[00:46:34] I’ve just told you there’s a better way.

[00:46:35] There’s the guy in the back on it’ll never work.

[00:46:38] And I’ll never get off the ground.

[00:46:41] Well,

[00:46:41] and you’ve also got the folks who are just pointing out all the,

[00:46:43] all the problems,

[00:46:44] which sounds like a kind of denial.

[00:46:46] It’s exactly what it is.

[00:46:47] And you know,

[00:46:48] and,

[00:46:48] and,

[00:46:48] but you want those people around,

[00:46:50] don’t you?

[00:46:51] Well,

[00:46:52] I think,

[00:46:52] I think,

[00:46:52] you know,

[00:46:53] a certain amount of skepticism is valuable.

[00:46:55] Yeah.

[00:46:56] You,

[00:46:56] you,

[00:46:57] you want those people who are saying maybe we shouldn’t let the AI run

[00:47:02] wholly,

[00:47:03] you know,

[00:47:03] our entire code stack and you know,

[00:47:05] we’re never going to manually review things like you want those

[00:47:08] people who are pushing back on it.

[00:47:09] Agreed.

[00:47:10] Yeah.

[00:47:10] But,

[00:47:11] but you know,

[00:47:11] I think the,

[00:47:12] the issue becomes when people deny the trajectory of where it’s heading.

[00:47:18] So I think,

[00:47:19] you know,

[00:47:19] didn’t talking about what it’s good at and what it’s bad at right now is

[00:47:24] exactly the conversations we should be having.

[00:47:26] Right.

[00:47:26] But anybody that,

[00:47:27] that looks at the trajectory we’ve had over the last 12 months and goes,

[00:47:32] okay,

[00:47:32] well,

[00:47:33] you know,

[00:47:33] that’s going to stop on,

[00:47:35] you know,

[00:47:35] whatever today’s date is.

[00:47:37] And it’s never going to get,

[00:47:38] it’s never going to get better or it’s going to get worse.

[00:47:39] I think those are the risky kind of opinions to have.

[00:47:44] And,

[00:47:44] you know,

[00:47:44] I think it comes from the fact that it is bad at a whole bunch of

[00:47:48] stuff.

[00:47:48] And I think it comes from the fact that the models,

[00:47:52] the models have incrementally got better.

[00:47:56] So,

[00:47:56] you know,

[00:47:56] we’ve,

[00:47:56] we’ve had a bit of a diminishing of returns in new model versions,

[00:48:00] a GPT three versus two was just game changing GPT four versus three was

[00:48:07] really,

[00:48:08] really,

[00:48:08] you know,

[00:48:09] really mass.

[00:48:09] Yeah.

[00:48:10] Yeah.

[00:48:10] But GPT five versus four is,

[00:48:13] is more so much.

[00:48:14] Yeah.

[00:48:14] It’s incremental.

[00:48:16] It’s,

[00:48:16] it’s,

[00:48:16] it’s meaningful.

[00:48:17] Sure.

[00:48:17] Don’t get me wrong,

[00:48:18] but it’s not exponential,

[00:48:19] but it’s also a point where you realize like the Moore’s law doesn’t apply here.

[00:48:24] There isn’t an exponential additional data set available.

[00:48:27] We’ve kind of indexed everything,

[00:48:29] including things we probably shouldn’t have been next.

[00:48:31] Yeah.

[00:48:32] And there is an exponential compute,

[00:48:34] right?

[00:48:34] You know,

[00:48:34] but what has got better has been the,

[00:48:37] the tooling around those models.

[00:48:39] Like this is,

[00:48:40] this is something we’re really good as engineers at is taking something that is

[00:48:44] pretty crappy and,

[00:48:46] you know,

[00:48:46] putting guard rails and tooling around it to make it good.

[00:48:50] You know,

[00:48:50] w we,

[00:48:51] we tricked locks into thinking and now we kind of make those rocks do slightly

[00:48:56] smarter things.

[00:48:57] I think that’s,

[00:48:58] that’s what we’re good at as engineers.

[00:49:00] Uh,

[00:49:00] we,

[00:49:00] we,

[00:49:00] we talked about it earlier of,

[00:49:02] of where we got to with IntelliSense and those toolings around it.

[00:49:07] Programming languages have,

[00:49:09] haven’t got meaningfully better in the past 20 years.

[00:49:12] They have a little bit,

[00:49:14] you know,

[00:49:14] there’s,

[00:49:15] there’s,

[00:49:15] there’s not,

[00:49:16] I would resist that.

[00:49:17] Like,

[00:49:17] I think the emergence of cloud specific languages like go fair point.

[00:49:22] And even the emergency of rust,

[00:49:23] like there is definitely a new generational language that were more tailored to the next generation of problems that,

[00:49:30] um,

[00:49:30] yeah,

[00:49:31] fair point.

[00:49:31] Fair point.

[00:49:32] Yeah.

[00:49:32] But you know,

[00:49:32] the,

[00:49:32] but the reality is lots of people also haven’t embraced those languages.

[00:49:35] Like adoption of your language is slow.

[00:49:37] Oh yeah.

[00:49:39] And yeah.

[00:49:39] And C sharp has continued to improve,

[00:49:41] but it’s also a 26 year old language.

[00:49:45] Yeah.

[00:49:47] Though that ability to iterate and improve on something that isn’t,

[00:49:52] isn’t perfect is what we do as engineers.

[00:49:55] And that’s what we’re going to see over the next few years.

[00:49:57] Like,

[00:49:57] I don’t think we’re going to see a land break improvement in LMS,

[00:50:01] but what we are going to do is see things like open claw and,

[00:50:04] and you know,

[00:50:05] those,

[00:50:06] those ways of using,

[00:50:07] the models to be meaningfully better.

[00:50:10] And so I think that that denial of the trajectory is the problem,

[00:50:14] not,

[00:50:15] not the kind of realism of,

[00:50:17] of,

[00:50:17] of where we are and that those,

[00:50:19] those people,

[00:50:21] it is,

[00:50:22] you know,

[00:50:23] on us as leaders to invest in those people and help them see and help them understand what’s happening,

[00:50:29] which takes a lot of time.

[00:50:30] But those,

[00:50:31] those people often tend to be,

[00:50:34] sorry,

[00:50:35] I know you don’t like that word,

[00:50:36] but those people tend to be the thought leaders.

[00:50:37] Yeah.

[00:50:37] Yeah.

[00:50:37] They tend to be the thought leaders in the team.

[00:50:38] Like,

[00:50:39] yeah,

[00:50:39] I,

[00:50:41] I,

[00:50:41] I,

[00:50:41] I joke about this in,

[00:50:43] in the workshops that I run in that,

[00:50:45] you know,

[00:50:45] it’s really easy as engineers for us to work out who the real leader of a team is.

[00:50:51] You,

[00:50:51] there’s a really simple way to work out who the leaders in your team are.

[00:50:54] Do you know what it is?

[00:50:55] What’s that?

[00:50:55] Break pod.

[00:50:57] Who,

[00:50:57] who does everybody turn their chairs around to look at when you break pod?

[00:51:00] Interesting.

[00:51:01] Those are the real leaders of the team,

[00:51:03] not the person with lead or manager in the title.

[00:51:06] Yeah.

[00:51:06] And those,

[00:51:07] those are the people that are often stuck in this place.

[00:51:09] And so they’re,

[00:51:10] they are worthwhile investing in.

[00:51:13] Yeah,

[00:51:13] that’s great.

[00:51:14] I appreciate that.

[00:51:14] Fair.

[00:51:15] But,

[00:51:15] and yeah,

[00:51:16] I also think we’re nowhere near the end of this.

[00:51:19] Like I do feel like we’re in this bubble and it’s going to end.

[00:51:22] And a lot of folks that have been resisting or go see when the,

[00:51:26] when the money strips away.

[00:51:28] Yeah.

[00:51:28] But I also feel like we’re about to get more efficient too,

[00:51:31] because when the money goes away,

[00:51:32] people learn to be more efficient.

[00:51:34] They don’t need to right now.

[00:51:35] So they’re not.

[00:51:36] And the hard,

[00:51:37] where is going to get more efficient.

[00:51:39] You know,

[00:51:39] I think,

[00:51:39] I think that’s a big thing.

[00:51:40] Even if the models don’t get appreciably,

[00:51:45] you know,

[00:51:45] exponentially better,

[00:51:46] what will happen is the ability to run those models on local hardware will become,

[00:51:53] you know,

[00:51:53] just what we do.

[00:51:54] That’s where,

[00:51:54] like that’s,

[00:51:55] that’s,

[00:51:55] that’s where my money is.

[00:51:56] I’m currently literally a new machine that,

[00:52:01] that I’ve set up to run local stuff.

[00:52:03] And I’m betting on the fact that local models in the way that we can,

[00:52:07] interact with them with our software tools are really going to change,

[00:52:13] be a game changer.

[00:52:14] Yeah.

[00:52:14] Yeah.

[00:52:14] Yeah.

[00:52:15] I think you’re absolutely right.

[00:52:16] I think that’s going to be,

[00:52:17] you know,

[00:52:18] the change of the next five years is going to be the stuff that we have to pay hundreds of dollars a month.

[00:52:24] Like I,

[00:52:25] my,

[00:52:25] my bill to anthropic right now is about 2000 us a month.

[00:52:29] You know,

[00:52:30] I could,

[00:52:30] I could build a piece of a piece of hardware that,

[00:52:34] that would run that in a year.

[00:52:35] Yeah.

[00:52:36] You know,

[00:52:36] so like,

[00:52:37] I think we’re going to get to the point when,

[00:52:39] you know,

[00:52:40] we’re just,

[00:52:40] we’re just running this stuff on machines.

[00:52:43] The 96 gig VRAM.

[00:52:46] That’s the problem.

[00:52:47] It’s 50,

[00:52:47] 90 RTX is about 10 grand us.

[00:52:50] So in less than six months of an anthropic bill,

[00:52:53] you’ve got the month mother of all cards.

[00:52:56] Well,

[00:52:56] that’s just now.

[00:52:56] I mean,

[00:52:57] if there’s a bubble that bursts in the,

[00:52:59] you know,

[00:53:00] the,

[00:53:00] the AI providers and the LLM providers,

[00:53:03] the,

[00:53:03] you know,

[00:53:04] the,

[00:53:05] you know,

[00:53:06] what is it?

[00:53:07] The one,

[00:53:08] what did,

[00:53:09] what did you call it?

[00:53:09] Richard?

[00:53:10] All you can eat model.

[00:53:12] Yes.

[00:53:13] Yeah.

[00:53:14] The,

[00:53:14] all you can eat model for a certain number of dollars a month that isn’t

[00:53:17] paying for it.

[00:53:18] No,

[00:53:18] that isn’t covering its costs.

[00:53:19] Like we have not paid the real costs yet.

[00:53:21] No,

[00:53:22] no.

[00:53:22] And when that happens,

[00:53:23] you know,

[00:53:24] all this,

[00:53:24] all these graphics cards are going to be for sale on Craigslist.

[00:53:29] Yeah.

[00:53:30] For very cheap.

[00:53:31] I have,

[00:53:31] I have the real evidence of that.

[00:53:33] So that $2,000 that I spend a month.

[00:53:35] So I,

[00:53:36] for my own use,

[00:53:37] I have a,

[00:53:38] well,

[00:53:38] whatever the Claude pro max 20 X subscription is.

[00:53:41] So basically the biggest one that Claude offer,

[00:53:43] and that’s about 300 us a month.

[00:53:46] Yeah.

[00:53:47] And then one of my engineers,

[00:53:48] he uses Zed as his editor and Zed doesn’t allow you to,

[00:53:53] to use all open or anthropic doesn’t like,

[00:53:56] well,

[00:53:56] or whatever it is.

[00:53:56] Basically you cannot use the subscription with Zed.

[00:54:00] You have to use a raw API key.

[00:54:03] And so the other 1700 a month,

[00:54:05] his tokens,

[00:54:07] the raw key.

[00:54:08] Yeah.

[00:54:09] So that’s the difference between the two.

[00:54:11] Yeah.

[00:54:12] It’s literally a five,

[00:54:13] a five X difference.

[00:54:14] Yeah.

[00:54:15] And you know,

[00:54:15] and the thing is at two grand a month,

[00:54:19] that’s still cheaper than an intern developer.

[00:54:22] Like it’s not actually that much money if you’re putting real value,

[00:54:26] producing real results from it.

[00:54:28] Exactly.

[00:54:28] And that’s,

[00:54:29] that’s the way to think about it.

[00:54:30] Yeah.

[00:54:30] It’s six months.

[00:54:31] You could build your own machine with that.

[00:54:32] That would be pretty rock and sweet possibly,

[00:54:34] but you couldn’t.

[00:54:35] So,

[00:54:35] you know where we are right now with the models,

[00:54:37] you couldn’t run Claude Opus 4.6 on the hardware.

[00:54:42] Like to run that you need multiple,

[00:54:44] you know,

[00:54:45] a six hundreds or whatever they are.

[00:54:46] Like you can’t be two hundreds.

[00:54:48] Yeah.

[00:54:49] Yeah.

[00:54:49] But you can run deep seek and you can run a llama and there are things that you can run in.

[00:54:53] Then there’s rag and you could teach it stuff.

[00:54:55] And Laura,

[00:54:57] well,

[00:54:57] and that’s the other side of this is when the,

[00:54:59] when the engineers actually focus on efficiency and we start,

[00:55:02] if especially we’re talking about software development,

[00:55:04] like narrowing the scope of these models,

[00:55:07] down to the job that we’re doing rather than the generalized models.

[00:55:11] Yeah.

[00:55:12] Yeah.

[00:55:12] Yeah.

[00:55:13] That’s why these,

[00:55:14] these AI agents and the stuff like a squad is great because each of these team members can use their own models for the stuff that they’re good at,

[00:55:24] you know,

[00:55:25] their own meaning a different model.

[00:55:27] I do before,

[00:55:28] you know,

[00:55:28] I’m aware of where we are in the recalling of this and I want to make sure we touch on one thing that’s really important to me about this,

[00:55:37] this kind of stages of grief model that I discussed,

[00:55:40] which is the depression stage.

[00:55:44] Cause I think it’s,

[00:55:45] it’s really real and we don’t talk about it enough.

[00:55:48] Like,

[00:55:48] you know,

[00:55:49] that that senior engineer,

[00:55:51] which we kind of,

[00:55:52] you know,

[00:55:53] joked about earlier,

[00:55:54] they’re probably are going through some kind of existential depression around what it means to be an engineer.

[00:56:02] And you know,

[00:56:03] they don’t talk about it at stand up.

[00:56:06] They don’t,

[00:56:06] they don’t put it in their slack status,

[00:56:09] you know,

[00:56:09] morning,

[00:56:10] my career,

[00:56:10] like they,

[00:56:11] they,

[00:56:11] they don’t,

[00:56:12] they don’t discuss this stuff,

[00:56:13] but I think it is important,

[00:56:15] especially as leaders to talk about it with them because I’m,

[00:56:20] I’m feeling,

[00:56:21] I’m feeling the same way.

[00:56:22] I’m sure lots of people are,

[00:56:24] but we’re just,

[00:56:25] you know,

[00:56:25] we get so excited or at least I do about the possibilities of this that we kind of push down that feeling of,

[00:56:33] you know,

[00:56:34] depression and grief and a morning.

[00:56:36] And,

[00:56:36] and don’t acknowledge that it’s there.

[00:56:39] That’s fair.

[00:56:39] Yeah.

[00:56:40] Yeah.

[00:56:40] I mean,

[00:56:41] nobody has 10 years experience with Claude.

[00:56:44] Yeah.

[00:56:44] Right.

[00:56:45] You know,

[00:56:45] as,

[00:56:46] as hap has happened repeatedly,

[00:56:48] we get new tools or we get new libraries and we’re all beginners at it.

[00:56:53] And so part of that is just like,

[00:56:54] can you get up to speed?

[00:56:56] Is it worthwhile for you?

[00:56:58] But I do,

[00:56:58] you know,

[00:56:59] get the point and Nolan wrote about that as well.

[00:57:01] It’s like there was a thing you used to do that the shape of these new tools,

[00:57:06] I mean,

[00:57:06] it really doesn’t make a lot of sense to do that anymore and that’s,

[00:57:10] and that’s what you’re missing.

[00:57:11] That’s the,

[00:57:11] that’s the main thing that makes you sad.

[00:57:13] And if you’re feeling that it’s normal,

[00:57:16] you know,

[00:57:16] you’re,

[00:57:16] you’re not broken.

[00:57:18] Your career isn’t over,

[00:57:20] you know,

[00:57:21] it’s,

[00:57:21] it is just a transition and,

[00:57:23] uh,

[00:57:23] and a phase.

[00:57:24] And there is light at the other side.

[00:57:26] There is a career at the other side.

[00:57:28] It’s going to be a career that shapes differently.

[00:57:31] The things that you spend your day doing are going to be different,

[00:57:35] but there is,

[00:57:36] there is a career that like we,

[00:57:37] you know,

[00:57:37] we’ve,

[00:57:37] we’ve seen this with,

[00:57:38] uh,

[00:57:40] you know,

[00:57:40] rapid application development stuff.

[00:57:42] We’ve seen it with Excel.

[00:57:44] We’ve seen it with,

[00:57:45] you know,

[00:57:45] low code,

[00:57:46] no code stuff.

[00:57:47] I’m reminded of the book who moved my cheese.

[00:57:51] You remember that book?

[00:57:51] I’ve not read that.

[00:57:52] What’s that?

[00:57:52] It’s a little booklet.

[00:57:53] It’s just about,

[00:57:54] you know,

[00:57:54] you going along and the mouse is eating the cheese that shows up in the same

[00:57:58] place every day.

[00:57:58] And then they move the cheese.

[00:57:59] Oh my God,

[00:58:00] somebody moved the cheese.

[00:58:02] Right.

[00:58:03] So,

[00:58:03] you know,

[00:58:04] it’s about having to find your way to the next cheese.

[00:58:06] Yeah.

[00:58:07] Yeah.

[00:58:07] Yeah.

[00:58:08] And I think,

[00:58:08] you know,

[00:58:08] those of us who have been in the industry a while know that it is,

[00:58:13] it is just part of what you do.

[00:58:14] You know,

[00:58:14] I don’t write VB six anymore.

[00:58:17] I don’t write cold fusion anymore.

[00:58:19] I,

[00:58:19] I,

[00:58:19] you know,

[00:58:20] I have,

[00:58:20] I have moved on who,

[00:58:22] who would have anticipated that we were writing C sharp that actually runs in

[00:58:27] the browser when C sharp came out 26 years ago,

[00:58:31] or we would never have envisioned that.

[00:58:33] Yeah.

[00:58:34] Well,

[00:58:34] what’s next for you,

[00:58:35] Andrew,

[00:58:36] what’s in your inbox?

[00:58:37] Uh,

[00:58:37] so I,

[00:58:38] I had,

[00:58:38] I had another baby 10 months ago,

[00:58:40] so not a lot.

[00:58:41] Oh,

[00:58:42] congrats.

[00:58:42] Thank you.

[00:58:42] Oh,

[00:58:43] wow.

[00:58:43] No,

[00:58:43] you’re played his whole friend.

[00:58:44] We love it though.

[00:58:45] Yeah.

[00:58:46] Yeah.

[00:58:46] Yeah.

[00:58:47] So the,

[00:58:47] uh,

[00:58:48] the,

[00:58:48] the,

[00:58:48] the family dial is definitely turned up to 11 right now,

[00:58:51] which is kind of great,

[00:58:52] you know,

[00:58:53] taking everything out of other places.

[00:58:54] But no,

[00:58:55] I I’m awesome.

[00:58:56] I’m trying to write a bit more.

[00:58:58] I’m trying to,

[00:58:59] I’m trying to do blog posts.

[00:59:00] Like I did,

[00:59:01] I did this blog post.

[00:59:02] I did one a couple of months ago called embrace the suck.

[00:59:06] Um,

[00:59:06] which is,

[00:59:07] is basically all about that transition to leadership and how,

[00:59:11] you know,

[00:59:11] a lot of being a leader sucks.

[00:59:14] And you know,

[00:59:15] if you,

[00:59:15] if you try and not do the sucky things,

[00:59:18] then you’re not being the leader you should be and needs to be for your team.

[00:59:21] So I’m trying to do more of that kind of stuff,

[00:59:23] which is more authentic to my tone of voice and the way that I think about things,

[00:59:28] which is hitting you in the face with the truth.

[00:59:31] And then,

[00:59:32] you know,

[00:59:32] helping you see why it’s not as bad as it,

[00:59:34] as it,

[00:59:34] as it seems.

[00:59:35] That’s great.

[00:59:36] Well,

[00:59:36] Andrew,

[00:59:36] thanks for spending this hour with us.

[00:59:37] It’s great to hear your ideas and your,

[00:59:40] and it’s a great conversation.

[00:59:41] We should keep it going.

[00:59:42] Thank you both.

[00:59:43] I had a lot of fun.

[00:59:44] All right.

[00:59:45] And we’ll talk to you next time on.net rocks.

[01:00:02] .net rocks is brought to you by Franklin’s net and produced by pop studios,

[01:00:13] a full service audio video and post-production facility located physically in new London,

[01:00:18] Connecticut.

[01:00:19] And of course in the cloud online at p w o p.com.

[01:00:24] Visit our website at D O T N E T R O C K S.com for RSS feeds,

[01:00:29] downloads,

[01:00:30] mobile apps,

[01:00:31] comments,

[01:00:32] and access to the full archives.

[01:00:34] Going back to show number one recorded in September,

[01:00:37] 2002.

[01:00:38] And make sure you check out our sponsors.

[01:00:40] They keep us in business.

[01:00:42] Now go write some code.

[01:00:43] See you next time.

[01:00:44] I got transmitted by the FCC.

[01:00:50] Yes,

[01:00:50] I’m a toy boy.

[01:00:52] Life is hard.

[01:00:54] Pay my taxes with my credit card.

[01:01:02] I get paid every time I have a�

[01:01:03] And then you can get your shirt off,

[01:01:04] right?

[01:01:05] Of course.

[01:01:05] All right.

[01:01:05] See you next time.

[01:01:05] Bye.

[01:01:05] Bye.

[01:01:05] Bye.

[01:01:05] Bye.

[01:01:05] Bye.

[01:01:05] Bye.