Paul: Hi, I’m Paul Ford.

Rich: And I’m Rich Ziade.

Paul: And we are coming to you from Aboard’s offices in the heart of New York City.

Rich: Did you say Aboard’s offices?

Paul: Yes. And you are listening to Reqless, the podcast about how AI is changing software. And we’ll explain all those things that we just discussed right after this theme song!

[intro music]

Paul: Okay, so Aboard is our company. Aboard.com. Go check it out. Click “Try it”. It’s an AI-powered software builder. But then once you’ve messed around a little bit with the AI, humans come in and help you actually build the software. And we’re doing that for our customers right now. So check it out. That’s the end of the marketing portion. So, Rich.

Rich: Mmm?

Paul: Co-founder Richard. So, first of all, everybody, we are in a new office. We like to connect with humans. We love the robots, but we also like the people.

Rich: Mmm hmm.

Paul: It’s a little echoey in here. We’re going to figure out our acoustics as we go, but we’re going to be here for a little while.

Rich: Yup.

Paul: So good for us. Now something interesting has happened because, look, the AI conversion experience for us really started about a year ago, where we realized it didn’t just draw pictures and write bad college essays and make all the pictures look kind of blurry around the edges. Which was interesting that it could do it, but didn’t seem valuable in the same way as some other stuff might. But then we saw it write code.

Rich: Sure.

Paul: And it wrote pretty good code. And it was funny, because people had a lot of opinions about the other stuff, but very few people had opinions about the code.

Rich: Eh.

Paul: And we started to incorporate it into our product, and we kept finding new ways that it worked. And what’s happening in this world, the world—I’m going to, let’s just narrow in. There’s AI, by which we mean LLMs, large language models, companies like ChatGPT and Anthropic. You can tell them to write very specific kinds of code. The more granular, the better. And it’ll do a pretty good job. It’ll build things—

Rich: Yeah.

Paul: —it’ll create data, et cetera. And the more specific you are, the more you kind of emulate the software-development process, sometimes when you’re just really polite to the bot, the better the output. And so I started showing you this stuff and I definitely, like, for me, alarm bells started to go off. I’m like, “Oh, wow. Okay. Some of the stuff that we’ve built in the last couple of years isn’t super relevant. A lot of things that happen in the consulting industry aren’t super relevant.”

Rich: Mmm hmm.

Paul: And what I’ve been watching, as we’ve been experimenting more and more, because you and I used to run an agency. An agency is a collection of processes and people, right?

Rich: Sure.

Paul: Like people first, then the people embody the processes. So I go to an agency and I say, “Hey, I need to build a thing that makes 3D hats and sends them on the internet as emails or whatever.”

Rich: Mmm hmm. Mmm hmm.

Paul: And somebody says, “Okay, I’ll write down what you said and you tell me if it’s right.” “Yeah, it looks right.” “Okay, I’m gonna turn that into a high-level specification and a product-requirements document.” “Okay, well, that looks pretty good, too.” “Okay, well, those are nice documents. Now I’m gonna turn it into a technical specification. I’m gonna write all these pieces down, and the technical specification outlines what we gotta build. And now I’m gonna build an even more detailed plan and give it to an engineer and they’re gonna build it as a prototype and blah, blah, blah.”

Rich: Mmm hmm.

Paul: So I’m describing all these different kind of phases and artifacts. It actually turns out that what we keep learning is different parts and different phases of this process can be done with AI. It’s not just the code, it’s not just the document.

Rich: Sure.

Paul: And actually, if you do them in the sequence that you used to do them at the agency, you get kind of better results and more interesting results out of the AI.

Rich: Mmm hmm.

Paul: So you’ve been kind of following along with me and you’ve also been very engaged learning about it, reading about it. But I turned to you yesterday in a meeting. This is what I want to talk about.

Rich: Uh oh.

Paul: Yeah. And I said, “Rich, it’s come for your job.” And by your job, I mean, you are good at meeting people, talking to them, finding out what they want, and then really, in, like, a minute, telling them what it will take. Because you’ve been doing it for 20 years. Saying, “Here’s the platform you need to build. It’s going to cost somewhere in this range,” you know, usually…

Rich: Yeah.

Paul: “And it’s going to take you roughly this team and this amount of time.” And what I said to you, and you weren’t shocked when I said it, was—and if I’d said this a year ago, you would have been totally confused—”I really do think that we could get one of the large language models to do a pretty good job at everything that you used to do that was a major driver for growth for our company.” And how did you react when I said that?

Rich: I sobbed.

Paul: You didn’t, though.

Rich: I didn’t. No, I’m fine with it. I’m excited about it. Yeah, look, I think what’s fascinating here is that there’s a bit of sleight of hand going on. AI is very presumptuous.

Paul: Boy, is it.

Rich: It’s extremely, it’s like, “Well, Paul, oh, I know exactly—” Like, you haven’t finished your sentence yet. “I know exactly what you need. Hold on, I’ll be back in a sec.”

Paul: Well, because it’s a statistical model that is designed to just keep going.

Rich: It just keeps going. Right?

Paul: Yeah. Okay.

Rich: The truth is, what you rarely see from AI, is it saying, “Huh? All right, Paul, you’re an expert in a particular industry. Tell me a little more before I really jump the gun here.”

Paul: To be clear, you can prompt engineer to do that, but yes, there’s a reason, which is just like, that’s not actually—it doesn’t introspect. It doesn’t think or make a model of you.

Rich: You went through all those artifacts that we produce before we start coding.

Paul: Yes.

Rich: Why do we produce those artifacts?

Paul: To create consensus in the organization before the very, very expensive process of actually building the software started.

Rich: Exactly. I’m going to say it differently.

Paul: Well, but also, why does an architect design a house? Instead of destroying a bunch of, like, bricks in a yard and saying, “Have a good time.”

Rich: Why create a small scale model?

Paul: Yeah.

Rich: Right? And what you’re doing is you are essentially communicating back.

Paul: Okay.

Rich: To the person to just confirm—you called it creating consensus. What you’re doing is you’re trying to say, “I think I understand your intent. And so I’m going to show you what I think it looks like.”

Paul: That’s right. You are a translator. LLMs are a translator. You are a translator. A business person could come to you—I did this, too. They could come to you and they could say, “I have this problem.” And you could translate it into not just software, but a software project. You could say, “Six months. A lot of money. Needs these five people.”

Rich: But even before that, I’ve had conversations with people who thought they wanted Software A.

Paul: Yeah.

Rich: And I was like, “Well, if you really look at the root of this, you don’t—software A is just a band-aid. What you really need is Software B.”

Paul: Yeah.

Rich: And what that is is a dialogue based on listening to them and understanding what their intent is. Very often humans don’t articulate what they actually need. In fact, a lot of times, they’ll sign off and then three months later they’ll be like, “Well, wait, wait, there’s one more thing.”

Paul: Look, if there’s any function that we have on earth, it was that people didn’t know how to translate their needs into enterprise-y kind of software.

Rich: No. And so when you’re listening, what are you doing? You’re translating pain points and asks and requests. Sometimes—you’ll hear them all out, but sometimes the asks are actually ill-advised, because they’re not really getting at the root thing.

Paul: Let me go back to your original point, because there’s a thought that popped into my head. People love astrology. They love their horoscopes, right?

Rich: They do.

Paul: And honestly, if you switch out horoscopes, people tend to just be like, “Yeah, that’s me.” You could be like, you could switch Leo for Libra. And they’ll be like, “Ooh, boy, nailed me.”

Rich: Yeah.

Paul: Right? And I think that there’s, and that’s, that’s just humans. That’s how we are. When somebody describes, you know, “You’re an empathetic person, but sometimes you can be very shy.” [laughter] And you’d be like, “Aw, oh—”

Rich: Threads the needle.

Paul: “Ooh, boy, you really nailed it.”

Rich: Yeah.

Paul: And I feel that that is actually, like, AI produces a credible product-requirements document after I said, “I need to sell shoes online.” And it’s pretty good.

Rich: It’s not bad.

Paul: And I go, “Wow, it really understood me.”

Rich: Exactly.

Paul: Right? And I think that that dynamic is playing out over and over again, and it’s a lie we’re telling ourselves.

Rich: It is a lie, because when you really get into it, the very particular needs and motivations of an individual are not expressed in three sentences. They’re just not.

Paul: No.

Rich: Now look, the truth is, AI is, if you’re willing to put the time in and step, we’ve talked about this in the past, step through what you want, take it phase by phase—

Paul: And translate. You’re translating from one concept to the next concept.

Rich: It’s better and better, right?

Paul: Right. But it is not actually trying to create consensus and understanding between individuals. It’s simply extrapolating and it’s doing a database query and producing output.

Rich: Correct. But what you’re doing is by being more and more descriptive, you are narrowing the possibilities.

Paul: That’s right. That’s right.

Rich: It’s a limit—you’re limiting the vectors, right?

Paul: You’re simulating consensus-building.

Rich: You are.

Paul: Because it’s about getting—what is consensus? Consensus is everyone agreeing to something specific, where before it was more broad.

Rich: I mean if someone pitched me for a minute.

Paul: Yeah.

Rich: I wouldn’t hard sell them back. I would say, “That’s interesting. Why don’t you come by the office? I’d love to hear more.”

Paul: Right.

Rich: Like, I don’t have enough information.

Paul: That’s right.

Rich: To really make it—I don’t understand you yet. I have not internalized your needs yet.

Paul: By the time I’ve made five or six little deliverables translated in my little mini virtual agency that I’m building inside a chat.

Rich: Mmm hmm.

Paul: I now have created enough to simulate that kind of specificity.

Rich: Yes.

Paul: So it’s accelerated. I can accelerate the fake consensus-building that AI lets me do.

Rich: You absolutely can. And I know there’s a lot of anxiety out there about what this thing is going to do to different professions and skills and whatnot, right? And the truth is, I think as long as we keep in mind that humans really connect and really differentiate when another human internalizes their needs and desires? And that could be anything. That could literally be a doctor hearing you out and telling you, “Yeah, you think it’s that, but it’s actually just this. And just do these things and you’ll be better.”

Paul: Yeah.

Rich: Versus a printout saying that?

Paul: It’s real.

Rich: That’s a big difference. There’s a big difference, right?

Paul: It’s real. There’s utility in these tools. I mean, you know, I think what is tricky, there are sections of the economy, like, remember Fiverr?

Rich: Yeah. You could ask somebody to do anything.

Paul: You’d be like, “Do me a voiceover.”

Rich: Yeah.

Paul: Right? I feel that things like that where you had fully commoditized human beings, and they’re like, “Okay, put money in the coin slot, and I will give you your thing back. You type in the box and I will say those words.”

Rich: Yeah.

Paul: It’s gonna—like, that is very, that is vulnerable. If humans were already, like, pre-commoditized by the system, there is probably not—and the fantasy of those systems was like, all right, you know, Mechanical Turk, everybody will get it. You don’t hear about Mechanical Turk anymore.

Rich: No.

Paul: But everybody will get in there and they’ll do, like, a dollar’s worth of work. But that’ll be really meaningful in certain countries, blah blah blah.

Rich: Yeah.

Paul: And that’s gone.

Rich: Yeah.

Paul: It is just—because honestly, you can’t tell who’s on the other side. Most likely what will happen is like if you ask somebody to do the voiceover for you, there’s a very high chance they’ve got their voice cloned.

Rich: Yeah. They’re probably punching it in.

Paul: They easily could. [laughter] Right? And frankly, you—

Rich: Don’t own a microphone.

Paul: You’d never know. And so there’s a minute maybe where they could do that, but then the next minute somebody goes, “I’ll just do it myself.”

Rich: Yeah.

Paul: Yeah.

Rich: So, I mean, and the truth is, that was on a path anyway, before AI.

Paul: It was always—

Rich: The commoditization of certain deliverables and skills was on a path anyway.

Paul: We knew it was bad and we knew the margins were always going to get tighter and that people were going to get more and more vulnerable. It is a pretty big tidal wave for a chunk of the world. That’s real.

Rich: That is, that is happening.

Paul: Yeah.

Rich: There’s no way around that. And so what can I do? Let’s actually be the Fiverr person.

Paul: Okay.

Rich: Okay.

Paul: I’ll be a Fiverr voiceover guy.

Rich: Yeah.

Paul: Okay.

Rich: Let me post on Fiverr.

Paul: Okay.

Rich: For $20, you give me a script and I will read it back in a masculine voice.

Paul: Sounds promising.

Rich: With enthusiasm, and I’ll put emotion in it.

Paul: Great.

Rich: Okay.

Paul: Okay, good.

Rich: Okay, that’s getting replaced. You can punch into a machine.

Paul: Well, I mean, Google Gemini can do a podcast based on, like, a two-sentence prompt.

Rich: Exactly. Now let me pitch something different, right? Which is, “Looking for ideas and personality to really elevate your product? I’ll help you write the script, and we’ll have a conversation about what’s really going to reach your audience.”

Paul: You know, the reality is, if you pitch that to me, and you charge me for it, but you tell me that AI is going to read it, but I got my choice of voices, I’ll probably lean in. Like, that’s worth $20, 30.

Rich: What I did there was, is I specialized my offering, right?

Paul: Well, you actually, no, what’s funny though, is you defined the process and said, “I will do more of the process.”

Rich: I will do more of the process, but also I’ll be your collaboration partner, or not.

Paul: You generalized, but you generalized in a way that is headed towards a more market-savvy outcome.

Rich: It’s something I used to call the entry point at the agency.

Paul: Okay.

Rich: There’s an entry point where it’s like, I know I have the blueprint—

Paul: [laughing] Every now and then you say something where it’s like, “I used to call it the entry point.” Wait, the point of entry? You came up with that?

Rich: No, not the entry—

Paul: I know. Just helping the listeners.

Rich: Sometimes, somebody would come to the agency and be like, “I know exactly what I want. I’m looking for bidders.”

Paul: Yeah. That’s right.

Rich: It’s very hard for me to compete with that.

Paul: They’ll send you, literally, an RFP, a request for proposals, and 10 people are writing proposals. And the person who secretly wrote the RFP and gave it to them is the one who’s going to get it.

Rich: [laughing] There’s that.

Paul: Yeah.

Rich: That’s true. And they’re probably going to lowball, right?

Paul: Yeah.

Rich: They’re going to pay the least amount, because they don’t see a difference in the delivery capabilities.

Paul: That’s right. That’s right.

Rich: Versus a different entry point, which is like, I have a problem. Can you help me solve it? Meaning there is ambiguity there. That ambiguity means you and I have to have a dialogue.

Paul: Yeah.

Rich: About what you really need. Now look, can I start a prompt with, “I have a problem”? You could!

Paul: I mean, this is—

Rich: You actually could, and it might get you somewhere, but that’s very different than, essentially, the commoditized delivery of a thing that you’ve already specified, which computers are going to do. Like, hard stop. That RFP is going to get fed into a computer one day.

Paul: Yeah, this is real. [laughter] Everything’s going to get analyzed and turned into a table and reviewed, and the decks will get made automatically, et cetera, et cetera. So the thing that you do. Okay, let me, let’s go back to the one specific thing you do, and then you tell me how it’s going to go in the future.

Rich: Mmm hmm.

Paul: Because what it used to be is like, “Rich, I heard from a friend that you have a good agency. You’ve got that weird co-founder. And I sell over 5 million bushels of wheat a year from my giant wheat business, and I need a way to manage where all that wheat is and who the buyers are. And I need a system to do that.”

Rich: Okay.

Paul: And you would have a conversation, and you would be able to, very quickly, way before Accenture could even respond to the email, you’d be like, “Here’s what we’re going to do, and I can do this for you.” And you would be pretty accurate most of the time.

Rich: Yeah.

Paul: Okay. So that was where we were. And now I could type in, “I need it to do this thing.” And it would give me a really nice bullet point list, with more detail than you’d usually provide at that meeting. So now what? What are you going to do? What’s your value?

Rich: Well, first off, I mean, it’s a great question. First off, this thing that I viewed as the enemy, as a threat?

Paul: Yeah.

Rich: I’m bringing it into my toolbox.

Paul: Okay.

Rich: Hey, you want the Bushel of Wheat Management System? I’m going to get that for you. And guess what? I’m getting it for you a lot cheaper because I’ve got tools now that are AI-powered that are going to get it to you way cheaper.

Paul: Okay.

Rich: But we still need to talk.

Paul: Okay.

Rich: I still need to understand how your business works. I need context, and—

Paul: The thing is, if I’m talking millions of bushels a wheat, no problem. We do have to talk.

Rich: We have to talk!

Paul: Yeah.

Rich: We have to talk. Also, I think—

Paul: That’s a lot of wheat.

Rich: Let’s be realistic about humans and risk. If I’ve got tens of millions of dollars running through a system, I’m not going to type a few prompts in, spin up another system, and funnel those tens of millions dollars through some new system.

Paul: Hold on a minute, let me blow that up for one sec. What if I already had, you know, SAP ERP Wheat Tracker.

Rich: Mmm hmm.

Paul: And I had a prompt-based software builder that would let me do some new nifty thing? Do you think I would do it then?

Rich: Wait, are you, are you getting rid of SAP?

Paul: No, I’ve already got, like, it’s on top of SAP. SAP provides me with a software builder.

Rich: Maybe?

Paul: Maybe. Yeah, I…

Rich: Are you going to write a few prompts and then allow it to, like, shave off 11 cents off of every transaction? Eh… It’s a scary world, right?

Paul: I think the reality is, like, Joe isn’t going to get to do that. The entire IT infrastructure that manages the giant SAP deployment for the wheat people is going to be able to say, “Hey, if we did these three things quickly, we might be able to save a penny here and there.”

Rich: Dude, IT has been fleecing business for 50 years.

Paul: They love when you say that.

Rich: I think if we can now tell them 50% off, not 95% off, that’s cool.

Paul: I swear to God.

Rich: Let’s take it one step at a time.

Paul: We’re ready to have art revolution for people who, like, do pencil sketches and that the AI can now do the pencil sketches. [laughter] And there are like, there are 18 trillion IT consultants. There actually are, like, 20 million.

Rich: Yeah.

Paul: Who are a little bit on the block here. I haven’t seen a tear.

Rich: No.

Paul: [laughing] No.

Rich: No.

Paul: No one is—

Rich: Also—

Paul: No one’s upset about Mike in IT.

Rich: They’ll morph into something else, don’t worry about that.

Paul: Yeah, it is true.

Rich: I wouldn’t worry about them.

Paul: Somehow the software still won’t ship.

Rich: Yeah, but I don’t think anyone is coming out and saying, “Oh my God, my costs are about to go down 90% because computers can do it.”

Paul: Yeah.

Rich: No one’s saying that. How about we, like, actually show value by folding in these accelerants, essentially, and be smart about being the dialogue so that you don’t go too crazy. Look, it’s funny when it screws up on a prompt.

Paul: Yeah.

Rich: It’s not funny if, like, millions of dollars got redirected the wrong way.

Paul: Yeah, or like the X-ray machine just keeps shooting X-rays into your lungs.

Rich: [laughing] Yeah.

Paul: Okay. The reason I did this for the podcast is, like, I literally looked at you, I said, “That’s it for your job.” Because we were getting to this point in the conversation about how this technology is changing our industry.

Rich: Yeah.

Paul: Inside the company.

Rich: Yeah.

Paul: And I actually, I just thought that, like, I think that this is—okay. Yes.

Rich: Radiology.

Paul: Yeah?

Rich: It’s probably going to do a phenomenal job at some point.

Paul: Oh, it’s already doing a lot of that stuff.

Rich: Okay.

Paul: And the radiol—it talks to the radiologist.

Rich: That’s, I think that’s really key. And I think as humans, I’d prefer it to go that way.

Paul: For quite a long time. Yes.

Rich: For quite a long time.

Paul: We’re not there yet.

Rich: Yet we’re not there yet, right? It’s a fascinating world to navigate through right now.

Paul: Yeah.

Rich: I think when you embrace the threat and understand how it can empower you, and how you can level up because of it, I think that’s what it’s about. Look, many won’t, and they’ll just be angry. Okay. Good luck.

Paul: We’ve been there before. I do need you to reapply for your job.

Rich: [laughing] Totally fair.

Paul: That’s fine. CEO, but—

Rich: Hey, let me tell you a little bit about myself.

Paul: Yeah. You’re going to be enhanced. You’re enhanced CEO.

Rich: Yeah, yeah.

Paul: All right, friends.

Rich: All right.

Paul: So that is a way to look at this. I’m seeing my stuff that—you know, what it can’t do right now? It can outline things pretty well, but it just can’t do the writing. It can’t do the, like, the jazz hands, when I’m writing something and I’m putting a little spin on it?

Rich: Yeah.

Paul: It just doesn’t have any of that in it.

Rich: Yeah, yeah, yeah. It’s…

Paul: It’s because it’s a lot of little tiny decisions made at the sentence level.

Rich: Yeah.

Paul: And they actually. What happens is you get the product—I mean, I guess, you know, I should experiment. I should go back and be, like, “Okay, make this more fun or something.” Just like… But the iterative process of editing, it doesn’t seem to have captured yet.

Rich: Well, it’s, it’s drunk.

Paul: Yeah. Yeah. Well, it’s a database.

Rich: It’s  smart and drunk.

Paul: The data—

Rich: I once asked it, “Give me ideas for a Lebanese restaurant name.”

Paul: Oh…

Rich: And for some reason it like pivoted on “banana”.

Paul: Really?

Rich: And it just give me, like, Banana-Rama Shack. And it was just like, “Well, how did we end up here?” And it’s like we ended up here because one of the vectors went left instead of right. [laughing] And here we are!”

Paul: That is real. This the dice keep rolling. All right, so here we are in the middle of this world. It is a puzzling world. But even, you know, the CEOs now, even the CEOs got to start looking out.

Rich: Yeah.

Paul: That’s where we are.

Rich: There we are.

Paul: And so you’re going to have to figure out what your real value is, Rich, and you’re just like, “My value is, I’ll just talk to the guy.”

Rich: It’s an exciting and humbling moment. I think we do want to talk to each other. I think we’ve been holed up for too—that’s a separate thing. I think we like to connect. Humans like to connect.

Paul: I think that maybe, you know, humans exist in culture. We’re actually economic being second.

Rich: Exactly.

Paul: Yeah.

Rich: And I think we keep going back to that, right?

Paul: Yeah.

Rich: The restaurants are packed everywhere.

Paul: Boy, are they. People just want to eat.

Rich: Check us out at aboard.com. We have a tool that lets you sort of take a little sampling. It’s like a little…

Paul: That’s right. You just give us a minute to see what we’re doing with these tools, and then once you’ve taken that minute, get in touch. Give us, like, a little more time, and boy, we can, we can solve your problem for you.

Rich: Hit us up. Aboard.com. Have a wonderful week—and reach out, if you have podcast topic ideas. Let us know.

Paul: Hello@aboard.com. That’s an email address.

Rich: Sure is.

Paul: Bye.

[outro music]