Paul Ford: Hi, I’m Paul Ford.

Rich Ziade: And I’m Rich Ziade.

Paul: And we are the co-hosts of…

Rich: Reqless!

Paul: A podcast about how AI is changing software.

[intro music]

Paul: All right, here we are. You and I are the co-founders of an AI software company.

Rich: We build anything you need much faster and cheaper because we use AI, and the framework and the platform—called Aboard, at aboard.com.

Paul: Aboard.com.

Rich: Big announcements coming in the coming weeks. I’m very excited about them. Check it out: Abord.com.

Paul: Yeah, lots of good stuff going on. Also, if you check our newsletter, we’re hosting a big event in the end of September focused on bringing climate data into our product. So we want to show that to you.

Rich: And if you’re in New York City.

Paul: Yeah, come by.

Rich: Come on by.

Paul: September 25th.

Rich: There will be meatballs on little toothpicks.

Paul: That is true. In Union Square. You’re totally welcome. I think—

Rich: It’s a great panel. We’ll talk more about it.

Paul: We’ll get a URL up for next week. But if you want to get on the list, just send an email to hello@aboard.com or check the last newsletter.

Rich: Yep.

Paul: All right. So Rich.

Rich: Yo!

Paul: Interesting things happening in the world of AI right now. And I want to read to you this post, or just sort of like, talk a little bit about it. The CEO of Amazon is named Andy Jassy.

Rich: Yes.

Paul: Okay. Not Jeff Bezos anymore.

Rich: Jeff Bezos is on a boat.

Paul: He’s literally on a boat. Jassy put a post up on LinkedIn, and it’s like this. It’s like, okay, look, maintenance is really boring and difficult and so on.

Rich: You should read it verbatim.

Paul: Okay. “One of the most tedious (but critical)”—this is Andy Jassy, not Paul Ford speaking.

Rich: Mmm hmm.

Paul: “One of the most tedious (but critical tasks) for software development teams is updating foundational software. It’s not new feature work, and it doesn’t feel like you’re moving the experience forward. As a result, this work is either dreaded or put off for more exciting work—or both.” Hilarious side note here, this is the CEO basically saying, “I can’t get my people to do the work that I need them to do.” The CEO of Amazon. [laughing]

Rich: The core foundational update work.

Paul: This is an important moment for people to understand what corporate leadership really is. Like, he’s just literally like, I—

Rich: [laughing] Nobody wants to do it.

Paul: I have so much work for people to do.

Rich: Yeah.

Paul: They won’t touch it. They will do—yeah.

Rich: Yeah, but they love the idea of the rotating sneaker module.

Paul: Exactly.

Rich: Because it’s more fun from an engineering puzzle perspective to build something new than to go in.

Paul: This guy must make, like, $100 million a year.

Rich: He cannot—

Paul: And he cannot get an engineer to do anything.

Rich: He can’t.

Paul: So, “Amazon Q, our GenAI assistant”—

Rich: Q? Just the letter Q?

Paul: Well it’s Amazon. Yeah, Amazon Q.

Rich: Okay.

Paul: Because they can’t help themselves and they also, they have to have one, they have to have one of everything.

Rich: There are 400 products inside of AWS.

Paul: They—

Rich: Do not even go there.

Paul: There really are. And so like, one of the things that he can’t get anyone to do is update the AWS icon system to make any freaking sense.

Rich: It’s unbelievable.

Paul: Yeah, it’s like Windows 3.1 met a database and just had sex for two years. Is that too much for our podcast? Are we allowed to say things like that?

Rich: Talk about icons?

Paul: Having sex?

Rich: With icons?

Paul: Yeah.

Rich: [laughing] Absolutely.

Paul: [laughing] I think—

Rich: We’re trying to get listeners here.

Paul: I think that’s—

Rich: You can only ride AI and the software revolution so far.

Paul: I think you’re right. There we go. It’s getting racy.

Rich: Keep going with the quotes.

Paul: “Our GenAI assistant for software development is trying to bring some light to this heaviness. We have a new code transformation capability, and here’s what we found when we integrated it into our internal systems and applied it to our needed Java upgrades.” So Java upgrades are not fun.

Rich: Java is…

Paul: It’s the business language.

Rich: You ever do a reno and knock down the wall and then the contractor stands there and goes, “I thought it was going to be in better shape than that. We’re going to have to replace the innards here.”

Paul: That is—

Rich: That’s Java. [laughing]

Paul: That is Java. I have a tremendous soft spot for Java, because it actually—

Rich: It’s pure.

Paul: Yeah, it’s purely, it’s not a puzzle-solving language. It’s a, I put, here’s the box of tools that you need language.

Rich: It is nails—

Paul: Yeah.

Rich: And wood.

Paul: So clever, smart, creative programmers, of the kind that Amazon, you know, makes them go and whiteboard that they love to hire?

Rich: Mmm hmm.

Paul: They hate it because it’s not like doing 50 sudokus at once.

Rich: Yeah.

Paul: It’s like, hey, where is the—

Rich: It’s old.

Paul: —the business logic for how we calculate shoe taxes?

Rich: Right, right.

Paul: So anyway, all right, Java. I like how when we talk about software we’d like, I was going to read this. It was going to take 30 seconds.

Rich: Didn’t happen.

Paul: [laughing] No.

Rich: There’s just so much passion.

Paul: This is it!

Rich: In the room.

Paul: So he’s got three bullets. You ready for—bullet one: “The average time to upgrade an application to Java 17 plummeted from what’s typically 50 developer-days to just a few hours. We estimate this has saved us the equivalent of 4,500 developer-years of work (yes, that number is crazy but, real).”

So let’s come back to that, because that’s the big one, right? “In under six months, we’ve been able to upgrade more than 50% of our production Java systems to modernized Java versions at a fraction of the usual time and effort. And, our developers shipped 79% of the auto-generated code reviews without” changing them.

Rich: That’s a crazy number.

Paul: “The benefits go beyond how much effort we’ve saved developers. The upgrades have enhanced security and reduced infrastructure costs, providing an estimated $260M in annualized efficiency gains.”

Rich: That is unbelievable.

Paul: Clearly—he’s on LinkedIn talking about their new gen-AI assistant, and like, OpenAI over on the other side is saying it’s worth $100 billion. So obviously, Andy Jassy is going like, “Hey, we’re in the game, too.” There’s a big part of that in this story.

Rich: There is a big part of that. And look, it’s worth noting where we are in the AI hype cycle. The AI as assistant, like wow, wow, wow, I just ask it a question and I ask it to rewrite my headlines and I ask it to write a paragraph or tell me something about something very arcane, and it does an incredibly good job? That’s still magical. The images are magical. But, but that is now—that has settled in. That is settled in. Like, the wow factor that frankly is going, it is a fundamentally useful tool for a lot of people, is now in the world. It’s in the world and there’s a lot of competitors and there’s different ways to use it.

Paul: I would also say a cultural immune system is showing up. For instance, I’m seeing lots of tweets from professors who are like, the students are turning in their introductory essays using ChatGPT and they’re saying they use ChatGPT. So clearly they’re going to be talking about this and working with them. And I think like, a dynamic is emerging where educators are going to say, you can use it a little bit for this, but you still have to cite your sources, you still have to—

Rich: Right.

Paul: It’s finding its way into the world on the chatbot-assistant front.

Rich: Yeah.

Paul: And I think we’re going to figure out as a culture where it’s allowed, where it’s not allowed, et cetera, et cetera. But now it’s starting to land in a professional context where—have you used heavily any of these tools? So like, Cursor? Zed? Aider?

Rich: In a—no pun intended—cursory way.

Paul: Okay.

Rich: Like, I’ve messed with them.

Paul: So I’ve been going deeper and deeper on them. So Cursor is a good example. It’s the one everybody’s talking about this week. But there are a couple.

Rich: What is it?

Paul: It is a development environment with AI built in. And there’s another one that’s open source called Ader, A-I-D-E-R, that you can use. And so the experience of using them is roughly, so you hook them up to one of the big chats.

Rich: Yep.

Paul: You hook them up to ChatGPT or Anthropic. You can also use a locally hosted version of Llama from Facebook, which we talked about before. And you say, “I would like to—” I’m going to give a goofy example because I use it to figure them out and compare them. “Write me code that will write a random musical fugue.” Fugues are very like, bah bah bah—

Rich: Yeah.

Paul: And then it’s like, the next part comes in. So they’re kind of mathematical.

Rich: Layers, and yeah.

Paul: It’s a comprehension test. Will you make—I’m not expecting it to make good music. I’m just like, I’m going to say, no, use more quarter notes. Change the rhythm.

Rich: Right.

Paul: Add a section with more contrast. How far can it go.

Rich: Okay.

Paul: It can go really far. And what it does is it gives you a basic, and it kind of runs and you’re like, all right, well, that’s not really what I want. And then you say, actually, could it do this? And then it drops it into the world of, like—in the case of Aider, it actually does like a GitHub commit for you, meaning it tracks the change. So now it’s kind of your little programmer buddy who is, who, all the changes are incremental.

Rich: Okay.

Paul: The experience that the Amazon CEO just described is in no way surprising to me using these tools. I’m like, yeah, of course that would happen. I would be 20 or 30 times more productive.

Rich: Sure.

Paul: Because it used to be, Java in particular, like, one or two classes of Java a day, like maybe 40 lines of Java that were really good in a big enterprise system? That was a good output for an engineer.

Rich: Yeah. Yeah.

Paul: And I would guess that example, that that number goes up 10, 20 times with these tools.

Rich: I think, you know, you’re clearly fascinated and enamored with sort of the possibilities of these things, right, of these platforms. So you went right to the engineer’s desk.

Paul: Well—

Rich: You went from Jassy’s quote—

Paul: Yeah.

Rich: Which is frankly, an executive who is looking at the net benefits to a business, a very large business, and then you zoomed in on the tools themselves. And I want to zoom out for a second.

Paul: Okay.

Rich: Because I think you’re touching on a lot of different things at once.

Paul: I guess what I’m actually confirming, I think this is very real. I don’t think this is Amazon PR. I think he really did save that much time and money and get better code.

Rich: Oh no, seeing it. We’re building a platform right now that we’re seeing, if you are smart about the guardrails you put in front of it, you could make magic happen and you can make it happen in an incredibly fast pace. That is shattering a lot of our notions about the preciousness and the processes necessary for software to be durable and reliable.

Paul: You know what else it’s really amazing at? Is you say, hey, take this and translate it into a name, another language. So like Lisp—COBOL, the great language of banking.

Rich: It’ll port it. And I think you’re touching on another thing that, you already said it. This isn’t about are humans capable, it’s more about do humans want to do it, and no one wants to do it.

Paul: These are syntactic tasks. They’re not—

Rich: They’re not innovative—

Paul: —interesting.

Rich: They’re not interesting.

Paul: Compound-interest calculation.

Rich: I think that’s right.

Paul: And actually that’s worth noting. Like, these tasks are really—it’s, like, I need to upgrade from Java 11 to Java 15 because that’s our new standard. And, like, telling the computer to do that is actually, has been the promise that it would do it for like 25 years, and it always needed humans.

Rich: Always needed humans.

Paul: And now it does all the work and you kind of give it a thumbs up. Like, we’re kind of, we’re getting to what we promised.

Rich: I think the thing that is so promising here is the ability to finally go back into legacy software, and rather than cake on over the issues, and rather than replace it with something new, you can actually renovate that software. And there’s so many implications to all of this. Like, it’s, it’s, it’s actually profound. And I think it’s interesting that we’re talking about it because this is not interesting. What’s interesting is funny pictures and oh my God, it knew the answer to that really weird question.

Paul: I actually think that’s going to be remembered as the least interesting part of this technology.

Rich: I think so much of the status quo is about to get reset with this stuff. And I don’t just mean engineers. I mean the business stakeholders. They have not gotten the news yet. [laughing]

Paul: Well, this is, let me throw this out, because I actually think, let’s say this happens fast.

Rich: Yeah.

Paul: Okay? What did Andy Jassy say? He was saying, hey, this is amazing and powerful and we were able to do so much, so much more quickly using our tool Amazon Q. Okay? He also said 5,000 developer-years.

Rich: Yeah.

Paul: That’s 5,000 developers, that, like…

Rich: I mean look, Amazon’s code base, right?

Paul: Yeah, but—

Rich: It’s absolutely massive.

Paul: Yeah, but they have a lot of developers. Could they maybe do without, could they cut 5,000 developers and just keep using this technology?

Rich: Well, you know, I think, well first off, they’re unique. They’re Amazon. So the code base and what exists there is so vast and so big, right?

Paul: Yeah, but Bank of America probably has 10,000 developers.

Rich: They probably have 10,000 developers.

Paul: Do 5,000 of them not need work anymore?

Rich: Can we go back in time?

Paul: Go back in time.

Rich: And talk about why Bank of America can afford 10,000 developers.

Paul: I mean, I don’t know that they have, they definitely have more—

Rich: Thousands.

Paul: —than 1,000.

Rich: They have thousands, right? And so does JP Morgan and so do all the big banks and everybody spent, invested—and Cognizant is made up of 400,000 people and—

Paul: Yes.

Rich: There are armies of people out there. And the reason those armies exist is because technology—the payoff was so massive over the last 30, 40 years that the math worked still spectacularly, because the efficiencies brought by it, the ability to process transactions and the ability to sort of track work.

Paul: I mean when’s the last time, I used to have to go to a bank and talk to a teller when I was very young to get my money for the week.

Rich: Yeah!

Paul: And then, you know, ATMs were everywhere at a certain point.

Rich: Yeah. And here’s what I would put forward. I would put forward a huge percentage of those armies of people is tending to the code base. They’re not upgrading it. They are not fixing it. They’re fixing issues.

Paul: Actually what I would say is they are upgrading it just to bring it into compliance with what’s happened. Like, sort of Java 11 to 12.

Rich: Barely, due.

Paul: Yeah.

Rich: I wouldn’t even give them that.

Paul: Yeah.

Rich: I would give them more like this. There is a bug here and this is, this is how I’m able to reproduce it. And guess what they do with that bad code? They write more code.

Paul: Yeah.

Rich: They don’t go in. They don’t go in.

Paul: Yeah.

Rich: Right? And I’ve seen it—you ever, I logged into and I don’t, I’m going to use the name because I don’t, I don’t really care, it was Aetna insurance.

Paul: [noises of despair] Ugh.

Rich: I logged into it. And then I wanted to get to like my prescriptions and you could just see, you ever see, like, those kaleidoscopic fractal videos where like you—

Paul: Oh, God—

Rich: I opened up a whole other world. I saw 400 people leave the room and 300 others replace.

Paul: I don’t even need—you don’t even need to use that example. When we, when we booked the conference room we’re in now at the co-working space.

Rich: Yeah. You could see—

Paul: We hit seven different websites.

Rich: That’s right.

Paul: So you know, and each one was like a whole other team.

Rich: And so what you’re seeing is this. It’s easier to build shims and software fixes on top of software rather than fundamentally address the core. And entire economies are based on that. The whole support infrastructure consulting world is based on that. Now what’s happened is you have people saying, am I ever going to go into the cave or do I just keep 30,000 people supporting it?

Paul: Well, and this is also where companies like Salesforce and actually a big chunk of Microsoft, they come in and they say, hey, don’t even worry about, we’re going to just seal up the cave.

Rich: Seal up the cave. We will build pipes and we will draw data from the cave.

Paul: Yeah. And we will install Microsoft Dynamics or Salesforce or whatever and we’re just going to throw it all away. And you’ll have a commoditized version that is, that we will take care of and make better every year and so…

Rich: And then that becomes legacy.

Paul: Yes, that is right.

Rich: I mean, the truth is Salesforce has office towers in every major city.

Paul: Dude, but I got to tell you, I have never felt it all under threat like I feel with this technology.

Rich: Okay, so let’s talk about that.

Paul: Like, we’ve seen all of it. We’ve seen Metaverse, and blockchain, and object-oriented programming, and Agile will change everything—I’ve seen 50 different, like, ways. The reality is—what the Amazon guy just described is the reality. 5,000 hours that people would normally have been paid for, and they would—his point is they’d never do the work anyway.

Rich: They would never do the work anyway. And look, why do the work? Amazon, massively profitable, successful company. Why do the work? I’ll tell you why you do the work. I guarantee you, if you look at the feature upgrades of Java 17.

Paul: Yeah.

Rich: A lot of it is like, that one component that you’ve been living with for 15 years? It now runs at 10% of the CPU cycles.

Paul: Yeah, that’s right.

Rich: We do that times 1,000. That’s a lot of money.

Paul: I will save $50 million on my electric bill—

Rich: [laughing] Yeah.

Paul: —if you three people sit in a room and solve this for two weeks.

Rich: That’s right.

Paul: That’s real. At that scale, that’s real. I’m gonna, I’m gonna flip it real quick because Amazon is going to be fine because they sell all sorts of stuff and they actually, the transaction, like Amazon will be fine. Maybe some of the Amazon programmers will go home. My guess is they’ll actually find more stuff for them to do because they can never feed their growth. They always have more growth.

Rich: Yeah.

Paul: So things are going to shift around this technology in there. The one that is interesting to me, the one I think about: The government has enormous code bases and most of them can only be served by very few approved partners and vendors.

Rich: Yes.

Paul: The big one would be like Booz Allen Hamilton.

Rich: They have a large Washington, D.C. office.

Paul: Name a giant government bureaucracy, and Booz Allen is probably in there.

Rich: They like government.

Paul: Helping them with their code base.

Rich: Yeah.

Paul: And I’m not going to just pick on them. We also have McKinsey, they’re across the street. They have a big wing called Digital McKinsey.

Rich: Yeah.

Paul: And so all of these companies, they make an enormous amount of money, along with the Cognizants and the Accentures, going into these orgs and turning their code bases around. And it is always slow, and a lot of times there’s like, a lot of government code is legacy and messy and it’s very hard for them to hire talented people internally and so on.

Rich: These people don’t sell software licenses.

Paul: No. No.

Rich: Right? What do they sell?

Paul: They sell services and consulting and long-term projects to—

Rich: What is the atomic unit of what they sell? What does the invoice look like?

Paul: A human hour.

Rich: A human hour. So they have an inverse incentive to reduce those hours. They’re like, you have very large systems, you need very large numbers of people, and they send those people and those people live there.

Paul: Yeah.

Rich: They will be on these gigs for years, right?

Paul: But you could point a talented person at a legacy—five years from now, let’s say five years from now, two years from now, today, you can go in if you are cool with the tool running, because it can run locally, you can feed it to government code base and it can improve the government code base. It can increase the speed, it increase the quality, it’ll save you money, and it’ll run at scale. And that is that—if it can happen at Amazon, it can happen in the government.

Rich: You think Booze is pitching that?

Paul: No, and I think they should. [laughter] No, I’m serious. I think it’s coming. Right? Like, there is—

Rich: It is coming.

Paul: There is a reality here that is—

Rich: Somebody’s going to pitch it. If it isn’t them, somebody will.

Paul: Yeah.

Rich: Right? That’s coming. And everybody’s standing up their AI practice. And let’s talk about agents.

Paul: Yeah, yeah. They’re seeing, they’re seeing growth here, and I actually, for literally the first time in my career, I’m like, yes, there is growth here. Absolutely. We’re building a product. I believe in it. It’s good.

Rich: Yeah, yeah.

Paul: But, like, you should see some fear.

Rich: I think that’s normal, right? Like, I think that’s not, you know what I think of?

Paul: Well, honestly, we reap what we sow in this industry. Like, all we’ve done is optimize other people’s jobs. So here we are.

Rich: I think about cities that were, that used to be hubs because the trains came through and then planes showed up and commerce changed, and those cities kind of still exist, but they were not, they’re not thriving as much because the trains aren’t the lifeblood anymore. And that is life.

Paul: I mean, let’s be clear. I don’t think ten years from now, I’m going to look back on this podcast and there will be no engineering jobs. I’m just saying that, like, I just think the sea change is coming in that the computer is now as good as the junior engineer. They can get the thing up when, if they see—

Rich: It’s also extremely motivated and very high-energy. It’s willing to work.

Paul: [laughing] It’s always like, well, I love their attitude. They’re just like, I’m sorry, that was a bug.

Rich: Yeah, yeah, yeah.

Paul: Let me fix it for you!

Rich: You’re never wrong. [laughing]

Paul: Oh, never, never. And it’s—so we are entering that world, right? So all the engineering jobs don’t go home. Product manager’s still going to have a lot to do because there’s still just a lot of humans, a lot of data, and a lot of stuff to work with. But a lot of on-ramps are going away, and a lot of low level jobs are going away, and they are going away.

Rich: I got to tell you, I think the profession that will finally have its moment is the product manager.

Paul: Yeah.

Rich: Because the product manager—

Paul: [laughing] They’ve been telling you it’s their moment for 15 years.

Rich: I know.

Paul: But okay.

Rich: But the product manager, what has happened is this—

Paul: The technical product manager.

Rich: Yes, but that’s going to get, that, that bar is going to come further and further down as these tools get better, I’m sure.

Paul: Mmm hmm.

Rich: Actually, the product manager, to date, has been the liaison between business and technology. They’ve been sort of the go-between, right?

Paul: Yes.

Rich: I’m going to translate business interests and business needs into a set of requirements.

Paul: This is what we’ve been talking about. This thing is the perfect translator.

Rich: But I think the product manager’s power and knowledge hits a wall because engineers—and I don’t mean this in a negative way, but their expertise creates this incredible barrier—

Paul: It’s a territory they defend.

Rich: It’s a territory, right?

Paul: This is another aspect of, like, hey, I can translate from JavaScript to COBOL or, like, from Java to whatever.

Rich: Exactly.

Paul: Like, those were very fiercely defended territories. And the computer is like, “Cool! It’s all code! We’re all good!”

Rich: Exactly. And so I think let’s, let’s close this with a piece of advice to our business listeners.

Paul: Okay.

Rich: Because we tend to be technologists who talk to the business side of things.

Paul: Yes. That’s real.

Rich: The business stakeholders should look at this world as a huge opportunity to finally say, well, why not? Why not in three months? You could never say that. When you say to an engineer, when an engineer tells you 18 months and you’re like, well, why not three? They laughed you out of the room and you had no other options. You have options now. You do need product management. You need technical product managers to sort of light the fuse there.

Paul: There are nowhere near enough of those.

Rich: I’m going to give a piece of advice to engineers out there who are junior, people are graduating from school. Go be a product manager.

Paul: Yeah.

Rich: Because it is going to be a sea change.

Paul: I got to tell you, too, if I’m reading that Andy Jassy thing and then Accenture is giving me a quote for an 18-month Java re-engineering process and my big chain of retail stores, I might call and ask some questions.

Rich: Yeah!

Paul: And those are big public companies. Like, here’s a big chunk of this economy that is built on engineers needing a lot of time.

Rich: I think they’re going to be called technical product specialists. There’ll be new titles, it’ll be very exciting, Paul. Very exciting titles are coming.

Paul: You still do need to know how to read the code. Like, you’re not, it’s not, you have to be able to actually be in the world with the computer.

Rich: I think that is a great piece of advice to add on to what I just suggested, which is you should know just enough. There are about eight or nine concepts that pretty much make up all of programming.

Paul: Yeah.

Rich: Putting syntax aside, you should know them. They’re good to know.

Paul: Well, actually, what you need to know is enough syntax, because there are still errors. You need to know, like, hey, this bug happened and I need to—

Rich: Yeah.

Paul: Or a good example is like, I was getting set up with Aider and it was like, go get your API keys and so on. And like, just that process of like, finding—

Rich: What is that? Yeah yeah yeah.

Paul: Yeah, exactly. Like, that is, that’s a week for someone who doesn’t really know what an API key is.

Rich: Right.

Paul: So, so we’re not there yet. You can’t flip a switch. But hoo doggies, if you are tired of people who telling the humanists to learn to code over the last 20 years—

Rich: It’s a wild time.

Paul: Yeah. The coders are going to get taught to learn to do lots of other things.

Rich: Well, there’s a lot more to talk about on this, I’m sure. we’re going to come back to it.

Paul: We will. But that’s, that is just, here we go. Here we go. That’s my feeling.

Rich: Check out, we’re building a platform that actually represents a different approach around building custom software, right up to the UI and all the bells and whistles that come with it, and accelerating that process through AI. And Aboard is that platform. Check out aboard.com. We’ve got a lot of exciting announcements coming. And reach out: Hello@aboard.com.

Paul: You know, and I’m going to correct you on one thing. You keep saying, we are building. We built!

Rich: We built.

Paul: Yeah, it’s working.

Rich: Yeah.

Paul: People are using it.

Rich: Yeah.

Paul: It is working. And there’s still lots, lots we’re going to do. I’m excited about it.

Rich: Never ends.

Paul: All right, friends. Hello@aboard.com. If you want to come to an event, send us an email, and we will talk to you soon.

Rich: Have a great week.

Paul: Bye.

[outro music]