Paul Ford: Hi, I’m Paul Ford.

Rich Ziade: And I’m Rich Ziade.

Paul: And you’re listening to Reqless, R-E-Q-L-E-S-S, the podcast about how AI is [morphing into radio DJ voice] changing the world of enterprise software. I don’t know why I have this voice today!

Rich: Not just enterprise. All software.

Paul: That’s right. It’s changing a lot of things. And so we’re going to talk about that. I think, Rich, you have this thesis about why technology exists.

Rich: Let’s not get too metaphysical.

Paul: All right, so let’s, let’s play the podcast theme here and just kind of get into it.

Rich: Cool.

[intro music]

Paul: Okay. So, Rich, what is technology for? What does it do?

Rich: I mean, it does everything, right?

Paul: Yeah.

Rich: You can play a video game, you can get stuff done, you can organize your life, you can build a business on it.

Paul: Good stuff. But what’s the point? Why, why, why use version 2 instead of version 1?

Rich: I mean, if we want to go all the way back, I could put everything in a ledger. I have a family member who still has, it’s, like, an extra-large book.

Paul: God, I know exactly which family member this is.

Rich: [laughing] And it’s a ledger.

Paul: Yeah.

Rich: And it’s, like, the accounting for the day from the business goes in the ledger.

Paul: I’m going to imagine, I’m going to say two words and you’re going to—Boar’s Head.

Rich: No!

Paul: No?

Rich: No. [laughing] He doesn’t know what a ledger looks like. [laughter] Good God, Paul.

Paul: Okay.

Rich: No, no, it’s on the in-law side.

Paul: Ohhhh…

Rich: In fact. Yeah.

Paul: Okay, okay. All right, good, good. So, a ledger.

Rich: So ledgers are nice. They worked for a long time. Clean lines on a page. But then when computers showed up, it just made life so, so much easier, right?

Paul: It’s true. And let’s be clear, like, ledgers? 1300. It’s, like, they had a good run. They made it to the ’70s.

Rich: [laughing] It was awesome.

Paul: Yeah. And then spreadsheets, VisiCalc in the ’70s, suddenly computers, people were like, “Oh my God.”

Rich: Well, the other thing that ledgers didn’t do that you had to do by hand is if you needed to add the numbers up in the column, you’d have to do it yourself.

Paul: That’s right. And then everybody got real excited when the calculators show up. But then you could put the whole ledger into the computer?

Rich: A lot of work used to be ledger, calculator over on the side.

Paul: Yeah.

Rich: I’m going to add up the numbers and we’re going to give you totals for the day and for the week and for the month. That was what accountants, like, good bookkeepers did.

Paul: The thing that we just described, by the way, that is, when people talk about “the productivity revolution.”

Rich: Yeah!

Paul: That’s that.

Rich: Yeah, yeah, yeah. I mean…

Paul: I had four volumes of leather-bound books and now I have a floppy diskette.

Rich: I mean, you could make a strong case that the word processor and the spreadsheet are the pillars of all of the productivity revolution.

Paul: Sure.

Rich: Right? Writing and sending a manuscript that’s been typed out and someone working on a ledger by hand.

Paul: Also the telephone. You no longer had to send your boy with a calling card.

Rich: Yeah, exactly, exactly. So why are we bringing this up? To me, the way I perceive technology and the advancements in technology and as the march of technology, if you want, is essentially, and this is very unglamorous sounding, but it’s the best way I can think of to boil it down. It is the elimination of steps.

Paul: I mean, it’s a good—if we are eliminating steps, we probably are moving forward. I think that’s a fair hypothesis.

Rich: That’s right. That’s right. And I think what is so profoundly revolutionary about AI right now is that the internet is in our hands. We could do all the things. We could do research on a particular legal topic, or a particular medical condition—with the web, in fact. We could do it. We could do the hard work. But what it’s done is two things. One, it’s doing the legwork. It’s actually going out to the world and doing the research for you. It’s gathering the pile of research artifacts.

Paul: Mmm hmm.

Rich: But it’s doing something else that’s even more impactful for us, which is it’s assembling it into forms that are familiar to us, and that are useful to us out of the gate. Whereas I used to cut and paste out of Google and out of websites and put them in Google Docs, often, to do my research. And then I would clean it up and tidy it up and rewrite it and put it into paragraphs. AI is eliminating such monotonous steps around all the, like, scavenger hunting that we do around working with information.

Paul: I mean, people argue about what it’s good at and what it’s not good at. But one of the better formulations for something it’s really good at is, is from Laurie Voss, who works in AI, and he said it’s great at turning text into less text.

Rich: Yes.

Paul: Right?

Rich: Yes.

Paul: And I think that, like—

Rich: That’s a great way to put it.

Paul: Where can I trust this technology implicitly? If you give it 10 pages and say, “Give me 10 bullet points.”

Rich: It’ll do it.

Paul: It will do it and it will be consistent and the bullet points will be related to what was in the 10 pages.

Rich: It’ll be pretty credible, right?

Paul: Yes.

Rich: And I’m just talking about information retrieval and the distilling down of a lot of information. But if you told it to write a poem, it’ll do that, too.

Paul: Yeah.

Rich: It’ll do a lot of other things.

Paul: Sure. But no, no but I mean, we’re learning what it’s good for. It’s not good for everything, no matter what they say.

Rich: And you should—and we say this a lot on the podcast—everything should get vetted by human eyes. If you’re writing a legal memo or you’re actually diagnosing someone’s condition, you should not just take it wholesale by any means. You should actually interrogate what it’s putting out, obviously.

Paul: So look, today’s podcast, we’re just kind of talking about prompts in general.

Rich: I want to talk about the prompt for a minute, I think.

Paul: Talk about the prompt for a minute. Okay, so we’re skipping steps. That makes a good technology. Good technology lets you skip steps.

Rich: When you say skipping, when you say you’re skipping steps, what you’re essentially saying is, I’m going to require you to move the mouse pointer around less.

Paul: Sure.

Rich: And click on less things and understand less interfaces. Because when you type in a box, it is raw human communication. And that is the fundamental—that’s why prompt, the prompt is like such a big part of the AI conversation right now.

Paul: I mean, let’s compare an old way to a new way, right? What, give it, let’s come up with a task. Maybe something legal. You were a lawyer.

Rich: Yeah. “Hey, Rich, we need an NDA document for our partners. I don’t have one.”

Paul: We’re going to assume one thing, which is that you are the decider as to when this document is ready to go out for signature.

Rich: Yes.

Paul: Okay? So tell me the old way of creating the NDA document.

Rich: I would look around on the web. What you’ll often find, if you type in “NDA template” in Google, what you’ll find is law firms that want you to, like, discover them.

Paul: Yeah—

Rich: Give away a document.

Paul: Sure.

Rich: They’ll give away a PDF. There’s also, like, LegalZoom and other resources out there, where you can pay $20 and get an NDA draft.

Paul: Mmm hmm.

Rich: What often happens, though, as a lawyer, I have a law background, is that the particular NDA that I want isn’t just sitting there on the internet. I’m going to have to essentially compose it by grabbing chunks that make sense to me from a lot of different drafts. So I’m opening 12 tabs.

Paul: Let me ask you what sounds like an ironic question, but actually isn’t. Has any legal document been started from scratch in the last, like, hundred years?

Rich: [sighs] I mean, just because of law of numbers. Yes.

Paul: I mean, sure.

Rich: But frankly, rarely.

Paul: Okay, so this is absolutely—I don’t know if everybody knows this. I assumed it. It’s a cut-and-paste culture and has been for a long time.

Rich: It has been. What often happens, though, is things get negotiated and discussed and the lawyers, what they—the real skill is editing that clause to account for the change.

Paul: I get it. I get it.

Rich: But yes, you’re right. No one’s staring at a blank page.

Paul: Okay, so that, that is already we’re turning, we’re kind of back a little bit. We’re taking—

Rich: We’re talking about contracts, though. I mean, the interesting thing about legal opinions, court opinions, is they actually get drafted.

Paul: Yeah, well, you can’t cut and paste that, if you’re a Supreme Court justice.

Rich: That’s a bad scene.

Paul: Yeah.

Rich: That’s not cool.

Paul: That would be, that would make a whole other kind of news.

Rich: Yeah.

Paul: No, but, okay, so we are in a world where we’re kind of, when we’re saying we’re converting, we are converting more text to less text, even without AI. Like, we’re taking all the contracts that we have kind of around.

Rich: I’m taking, you know, 200 pages of NDAs across 20 documents, 40 documents, and I’m trying to distill it down to four pages.

Paul: And then you’re applying literally what you learned at Brooklyn Law and going like, “Here’s the one I want.”

Rich: This is irrelevant, this one makes sense, but I’m going to change it. And so I’m essentially piecing it together. So now—

Paul: Okay, so you’re like in Microsoft Word doing this in ye olden times.

Rich: Hell yeah.

Paul: You might go Google, you might go, you got some photocopy in the drawer, and you OCR it.

Rich: Absolutely.

Paul: But, like, so I’m going to make them digital. I’m going to glue it together and then I will edit. And then I’ll send it over to their lawyer who will send back some red lines. And here we go.

Rich: I mean, to be frank, we ran a business on exactly this process until I had the, you know, a folder of templates that made sense.

Paul: Totally get this. Okay, so that was the process.

Rich: Yes.

Paul: Now AI, now you have a prompt. You sit down. Which one is it today? ChatGPT or Claude? Choose one.

Rich: Claude for technology, code, and, you know, anything that’s related to software, and GPT for everything else. Now I have this prompt and I can type in and, you know, there’s a lot of tips on how to use prompts well. The more verbose you are, the more you set it up, the more you sort of give it a persona, like, “You are an expert legal thinker who’s very experienced—”

Paul: “The world’s greatest NDA lawyer.” Yeah.

Rich: A lot of compliments and whatnot. And then off it goes and it will start to assemble—

Paul: A non-disclosure agreement.

Rich: A non-disclosure agreement, what will look on its face, and this is a key piece of advice for anybody using these things, is it looks really good out of the box.

Paul: You know what, it’s interesting, because I’ve been an editor and I’ve seen a lot of raw text in my life. You know how you can tell, even without reading a paragraph? The balance. Truly organized, well thought-out documents don’t have natural parallel structure. Certain things are longer, sentences—like, there’s a variation in how they look and feel.

Rich: Uh huh.

Paul: And when ChatGPT produces output, it is like architecture. Like, it’s just everything’s kind of the same length, same formatting.

Rich: Yeah.

Paul: And it makes me very suspicious, because contracts are jagged too. They don’t actually, they might have lots of bullet points and so on, but they’re not, they’re human artifacts. They’re not robotically the same.

Rich: Oh, a lot of my follow up prompts are, “Tighten the language and make it shorter,” very often.

Paul: Okay, so you say—what’s your prompt, actually? Go ahead.

Rich: “I have a small software business and I need a relatively flexible NDA agreement that I can give to new hires as well as partners and clients.” Hard stop.

Paul: Okay. What comes out?

Rich: It’ll give it a go. It’ll just write. It’ll just start writing.

Paul: It will look roughly like a legal document.

Rich: Will look roughly like a legal document. Now, it probably, it doesn’t know what jurisdiction I want to be in. So, “If there’s a conflict on this based on this NDA, it’s under the jurisdiction of the state of New York.” Doesn’t know to do that.

Paul: It is learning to ask. That’s different—

Rich: It is learning to ask.

Paul: Not always, but sometimes it’ll ask.

Rich: That’s right. Sometimes it’ll say things like, “Okay, non-disclosure. Do you want it to also include non-compete?” And you’re like, “Yeah, sure, go ahead.” And then off it goes. Now—

Paul: Would you use the deep research mode for ChatGPT for this? Have it go out and look.

Rich: No, the reason I wouldn’t is because it’s, it’s an NDA is a, is a relatively small artifact.

Paul: We actually want a generic starting point.

Rich: You want a generic starting point. I think, look, deep research is like if you’re in litigation and you’re in a complicated case with a lot of fact patterns and a lot of discovery material and you’re not, you know, you’re looking for strategies. I haven’t used deep research. I’m assuming it’s for something more complicated.

Paul: Critically, also, you’re not feeding it any documents.

Rich: No, no, no. An NDA template is a basic thing. No, I’m not. I’m not.

Paul: So no. But this is, it’s really good at bureaucratic artifacts that are very common.

Rich: I will say if there was an agreement, a contract for services that had a statement of work, and I had a two-page memo of what we’re going to do for the client, you could feed it that and be like, “Put this in an agreement,” and it’ll do it.

Paul: This is—now we’re back to taking lots of text and making it and tightening it up. Right?

Rich: Yeah.

Paul: And it’s really, it’s quite good at that.

Rich: And to me, why are we talking about this sort of corner of using AI? I think it is about the interface, I think the inter—the ability to not run around, save stuff in folders, paste things into documents to get work done, and instead use this tool by essentially writing a fairly verbose request in a box.

Paul: Mmm hmm.

Rich: And that’s a wild, wild thing. And it is also, here’s the other thing I would add the accessibility. And I don’t mean accessibility in terms of, like, vision and hearing. The accessibility of capabilities like this for people who are not savvy with, with computers is incredible. Right? Because you can essentially tell someone who doesn’t really know how to use their phone very well to just ask for what they want.

Paul: No, this is real. It’s also, people who where English isn’t their first language or speak a dialect that isn’t, like, the business dialect. They have access to formal, proper communication in ways that they didn’t.

Rich: A new business owner that just isn’t that familiar with how to, like, you know, how to communicate with their employees, and they need help with that. Like, they would spend hours reading up on stuff or buying books and to ask and get advice that quickly without—

Paul: You know who this is probably bad for is the local community fixer for a particular immigrant group in New York City. [laughter] There’s always that one guy who’s, like, going to be able to negotiate for you.

Rich: Yeah, yeah, yeah.

Paul: And he takes his cut. Like, you know. it’s not a good technology for him.

Rich: So I have a question for you. I’ve been doing most of the talking.

Paul: Okay, great. It’s been great.

Rich: I think, and I’m not a marketing guy, but it feels like “prompts” is an incredible way to convey the value of anything that has to do with this stuff.

Paul: So, look, I, you know, what are our jobs at this company, Richard? You are the operator, forward-motion person. You tend not to sleep, and you often talk to me at night in Slack, even when I’m asleep. [laughter] And that’s—you’re good at that. You are true, you are a forward-motion—

Rich: I don’t need you to respond for me to talk to you.

Paul: No. God help you if something has to move backwards. But definitely like that shark-like forward quality.

Rich: Yeah, yeah.

Paul: In droves. I tend to kind of sit along, and then what I do is I sort of say the same thing 7 or 8,000 times.

Rich: Yeah.

Paul: And then people repeat it back to me to tell me what I’m actually saying. And that’s how I know I’ve succeeded as a manager.

Rich: An idea went through.

Paul: Yeah, that’s it. Once they’re actually educating me as if I don’t know?

Rich: Yeah.

Paul: Then I know that it actually worked.

Rich: Yeah.

Paul: I just, I assume that that’s how it’s going to go. So I made little stickers and I put them around the office, and they have two words on them.

Rich: Uh huh?

Paul: Prompts. Logos.

Rich: Okay. Explain that.

Paul: It’s very simple. We’re building a system that helps you build software quickly with AI. Humans stay involved, but that’s what it does.

Rich: Okay.

Paul: Okay. Two things matter to us as a company. One is for the user and one is for us. The first one is that we know what prompts we can handle. We can communicate the prompts and we can say—so a prompt might be, “Build me an HR system.” “Build me a whatever.” Right? We’ll get into more detail as this thing comes online relatively soon.

Rich: Okay.

Paul: The most important tool for communicating in the organization is, and we’re building this, and I’m going to slowly build this in my leadership role, is to know what prompts we can handle. Can we handle—what level of complexity? Can I put five paragraphs of description of a software system in and get most of what I asked for? Or is it really just, like, one or two sentences, but then we kind of need to take some humans and keep them involved right there?

Rich: Mmm hmm.

Paul: And, you know, it’s funny with this technology, because sometimes it can be one or the other. I can’t consistently tell you exactly what’s going to happen—

Rich: Right.

Paul: —every time you sit down to build software.

Rich: Are you saying—when you say “prompts and logos,” do you mean that’s the way to market what, like, first off, this is becoming about us, so, like, let’s take 10 seconds. We’re building a platform that lets you type in a box and get software.

Paul: Correct.

Rich: Like, that is, as a baseline, what we’re doing. We’re doing—there are a few players out there that are doing it. We’re doing it a bit differently, but… Is this about marketing?

Paul: Business software that’s fully articulated, not so much that, like, “Hey, I made a clone of Airbnb.” We’re doing something a little different.

Rich: Zing.

Paul: Yeah, no, but for real.

Rich: Yeah.

Paul: No, this is not about marketing. This is about how to organize our thoughts about the value we’re delivering to the customers. A prompt, to me, represents a reproducible unit of thinking that I need to share with people. So it’s, like, okay, here’s the kind of prompt that my system can handle right now today. It will help me build an HR system.

Rich: So this is about communication.

Paul: Yes, because the thing with AI is if you can build an HR system for a pet shop, you can build it for a grocery store, a software company, and so on. Like, customization gets really easy on the edges.

Rich: In fact, if you tell it, “I need an HR system for my pet shop,” it’ll be more specific about that tool.

Paul: One of our use cases is “HR system for alpaca farm.”

Rich: Right. Right.

Paul: Because then it actually goes and figures out—

Rich: It thinks about alpacas.

Paul: We need to identify each individual alpaca.

Rich: Your words, I want to get to logos in a second.

Paul: Mmm hmm.

Rich: But your words are a way to sort of help us focus on a goal.

Paul: Well, let me nail it down. This world keeps changing very, very rapidly.

Rich: Sure does.

Paul: And the only way to know if you’re succeeding in using this technology is getting better results, either by making better prompts or using systems that take the prompts and turn them into better results.

Rich: Yes.

Paul: Like, it’s, you either are finding better tools that take ideas and turn them into the thing you want.

Rich: Yep.

Paul: Or you’re making your inputs better and smarter because you’re learning more about how the system works.

Rich: Yes.

Paul: Sometimes putting two paragraphs in means you get a better piece of output. Sometimes switching from ChatGPT to Claude or DeepSeek gives you a different kind of, or a different kind of flavor of output. Claude is better, in my experience, at coding.

Rich: All right, explain the logos part.

Paul: That’s just for us, which is success in our business comes from logos on a website.

Rich: People can recognize that we have actually proven this thing out and delivered value to customers.

Paul: Mmm hmm. Nobody, you know, we’re—

Rich: That’s old-school.

Paul: We’re strangers.

Rich: Yeah.

Paul: We’ve been having a lot of conversations about next steps. I’m like, “Where can I provide the most value?” And it’s literally just by saying, “Hey, what prompts can we handle here?” And then going, “How’s that going to get us logos on the website?”

Rich: Yeah.

Paul: Over and over and over again.

Rich: Yeah.

Paul: And, you know, we’re headed towards launch. I’m not beating the hammer too much. But as we go out into the world, I will be beating the hammer. I am trying to develop a set of like five or six prompts that I can use to test and understand any one of these systems. Like, anytime a new version comes out, you should put the same thing that you put in six months ago.

Rich: See what’s different.

Paul: What’s changing? Yeah, because it’s actually just, it’s so overwhelming and it’s really easy to lose track of what’s real progress and what’s marketing.

Rich: In fact, there are tools out there for, like, prompt management. It’s, like, prompt clipboard tools.

Paul: God, they’re confusing, though. I’ve tried a few of them.

Rich: They’re crazy.

Paul: They’re just, yeah.

Rich: We’ve all agreed, and we have a whole podcast on the topic of AI: This has undoubtedly been a breakthrough. The prompt is just this wickedly powerful thing. Is this the end of how we interact with AI in terms of interface?

Paul: I don’t buy it. You and I actually were wrestling literally with this conversation yesterday about our tool.

Rich: Yeah.

Paul: Which is, do you think people will go back to the prompt? I think that people tend to go back to one interface modality—I’m going to be real nerdy for a minute—over and over again, so, and I actually think that’s dangerous. What happens is if you’re, like, “Hey, I gave it some words and it made a thing. Now I need to change that thing a little bit.” You want to give it more words. You want to stay in one box.

Rich: You definitely do.

Paul: Yeah. But the reality is you might actually want to go over to another artifact that it created and point and click and use your mouse like it was, I don’t know, 2022.

Rich: You want to go in yourself.

Paul: You may need to.

Rich: You may not want—

Paul: If we’re focused on acceleration and cutting steps, sometimes the prompt does not cut as many steps as simply moving your mouse around.

Rich: Sometimes you should go do it yourself.

Paul: Yeah. Because otherwise you spend all your time trying to…

Rich: Isn’t it just a both thing? You know, the classic basic/advanced paradigm.

Paul: Absolutely. But I do think that once a person is in a system, it’s really hard to switch from one to the other. So you have to guide them. Right?

Rich: Yeah.

Paul: Like, literally, you know what, you know a pattern I saw, you know the product Bolt?

Rich: Yes.

Paul: Okay. Bolt is a prompt-based software-development tool.

Rich: Yup.

Paul: And you say, “Build me a clone of Airbnb.” And it kind of does. It looks like Airbnb.

Rich: Yeah.

Paul: But then you can click on an interface element it creates and it pops up—in the chat window—

Rich: Uh huh?

Paul: A generic old-school edit and manipulate this interface—

Rich: Like a DOM editor.

Paul: Yeah. Like, change the border radius and the color and you click around.

Rich: Right.

Paul: Like old-school software.

Rich: Right. It’s sort of throwing its hands up and saying, “You can go in and actually do this yourself.”

Paul: Yeah.

Rich: “Don’t just type.”

Paul: But it’s right in the chat window. I thought that was a really interesting place to put it.

Rich: It is interesting. Yeah, I think, I think, look, also we are so early days with these capabilities is that, as you iterate and try to refine on a fairly large artifact, like a pile of code, it isn’t very reliable. It gets less reliable, strangely. It tends to veer off and touch things it wasn’t supposed to touch.

Paul: Yeah.

Rich: And I think they know that and I think we know that as well. And sometimes you don’t want to—hey, you know what? I am savvy enough and I’m going to come behind the counter and make my own sandwich this time.

Paul: Well, this is the thing, and you’re seeing it. We’re all learning to be very suspicious of AI as a coding tool, just as we would be suspicious of it as a contract-writing tool.

Rich: Okay, last question then.

Paul: Yeah.

Rich: Do you think AI will just continue to improve and that advanced, that sort of hands-on capability won’t be needed? An impossible question, but I’ll ask.

Paul: I’ll just go ahead and put it out there, right? So I think it will continue to improve and that there’s enormous value in the current wave of technologies. I don’t think we’re going to see as many exponential step changes in capability as we have. Like, I just don’t think like chat GPT-5 will be the miracle worker that—

Rich: You don’t ever have to worry, like—

Paul: Yeah.

Rich: That it’ll get it all right every time.

Paul: I just don’t buy it for a lot of reasons.

Rich: I agree.

Paul: As I learn more and more about it, I think we’re kind of, we may be hitting some very natural organic limits, just like we’re hitting with like Moore’s Law. Like, it’s just…

Rich: Yeah, yeah, yeah.

Paul: But, but what do I believe—I just don’t think that humans are, actually, all this thing can do is learn from human culture, and humans are terrible at finishing projects. [laughter] I wrote about this in the newsletter last week, and it was just sort of like, you know, I’m working on a project for a friend of ours, and I’m using it to understand how migration of old data is going to work in the new world. Like, I just want to get my head around it. It’s how I work. And I find myself constantly veering off to add more value and make this even better.

Rich: Yeah.

Paul: And I haven’t finished the core task.

Rich: Yeah, yeah.

Paul: Right? And I think everybody—

Rich: You keep going on little side projects?

Paul: Yeah. Because it’s. It’s infinite. I would like to make the interface a little better. I’d like to make it smarter.

Rich: Always.

Paul: I’d like to make the text a little cleaner.

Rich: Always.

Paul: Exactly, right? And so, like, I don’t—because there’s nobody, and this is a, it’s me dabbling. Like, it’s supposed to be—

Rich: You don’t have a deadline.

Paul: Nobody’s yelling at me. [laughter] Nobody’s waiting. Nobody’s like—

Rich: If you want, I’ll yell at you, Paul.

Paul: I mean, that’s no worries. But like, no, nobody’s like, “Hey, I really could use this by April 10th.” Right?

Rich: Yeah, yeah.

Paul: If I knew I had to get it done by April 10th…

Rich: Yeah.

Paul: So I think we’re always going to be negotiating with this and the systems, I said a thing in that newsletter and I turned to you and I was like, “I think we’re going to live by this.” AI is terrible at cutting scope.

Rich: Oh, yeah. It’ll always do more.

Paul: It just adds more. And so that’s what we’re going to be wrestling with. You can have infinite everything, but you actually need the product manager to come in and say, “Guys, it’s April.”

Rich: Yeah. I think what, the argument you’re making here, back to the question I asked, which is I think humans using their own judgment to sort of finish the job or step in, roll up their sleeves, and sort of button it up, I think is going to be very meaningful. I don’t know—

Paul: Right now, everybody’s dabbling.

Rich: It’s a dabbling world right now.

Paul: Yeah.

Rich: Yes, I totally agree. Especially in this, we’re speaking specifically about software, especially in software building.

Paul: Yeah.

Rich: Using AI to build software.

Paul: I’m very excited when we get our product out, because the truth is it doesn’t use AI to write code. It uses AI to organize.

Rich: Yes.

Paul: And so—

Rich: Sneak peek.

Paul: Yeah…

Rich: We’re going to be talking more—it’s getting closer and we’re going to be talking more about it.

Paul: But that’s also because we don’t trust it to finish code for enterprise and business clients.

Rich: That’s right. That’s right.

Paul: Like, that’s—I’m just letting everybody in. Like, it’s, I don’t believe this thing should write your end product that goes to your users in your organization or your customers.

Rich: Yeah, I think it’ll write some code in certain places.

Paul: Absolutely. Or you’ll write it to write this—ut there’s just new stuff is emerging. Anyway, here we are.

Rich: Stay tuned. We’ve got very exciting things to share soon.

Paul: And if you have test prompts that you like to use or things you like to put in the system?

Rich: Let us know.

Paul: Share them. Yeah, we’d love to see. I’ll read them in a poetic voice on the next episode.

Rich: All right, Paul, prompt and away.

Paul: There we go. Prompt and delivered.

Rich: Yeah.

Paul: Okay, so check us out at aboard.com. New website, new stuff coming real soon. Oh my God, we’re excited. I sound a little tired when I’m saying that, but that’s cause I’m so excited. Hello@aboard.com if you would like to communicate with us in any way. And we’re growing. It’s a good time to get in touch. The job board is a little bit out of date. I gotta go tidy that up. But nonetheless, if you’re interested in getting in touch, please do.

Rich: Have a great week.

Paul: Bye.

[outro music]