
The latest episode of the Jeff Willis Show features software expert David Hirschfield, who shares his “Launch 1st” approach for startups and discusses the changing role of AI in the software industry. The conversation discusses David’s Hirschfeld’s background, methodology, and predictions for AI’s future impact on different sectors.
What will you learn?
- How the “Launch 1st Method” helps software startups validate their ideas with high-fidelity prototypes before committing to development.
- How AI is used to write code, design systems, and improve efficiency in software development.
- How AI agents are transforming workflows by managing tasks from ideation to deployment.
- How AI is revolutionizing online education through personalized learning experiences and enhanced engagement.
- Rapid AI adoption reshapes industries, including real estate, and creates new ethical challenges.
From Physics to Software
David Hirschfeld began with a physics degree, fueled by a passion for math and theoretical discussions with friends. However, he realized the academic physics world wasn’t for him and transitioned into software sales. He quickly excelled, attributing his success to a deep understanding of the technology he was selling. This led to a career in software development, culminating in founding his own company, Tekyz.
The Launch 1st Method
David created the “Launch 1st” method to tackle the alarming failure rate of software startups. This method prioritizes establishing product-market fit early by pre-selling the product before development begins. The Tekyz team creates high-fidelity, animated prototypes instead of traditional Minimum Viable Products (MVPs) to simulate the final product. Prototypes are used for pre-launch sales, helping startups earn revenue and validate their product ideas before significant development investment. “David emphasizes the importance of launching your sales and marketing engine before beginning product development.” If the prototype doesn’t sell, the team will refine the design and marketing until they find a way to generate revenue.
The Rise of AI in Software Development
The conversation then shifts to the impact of AI. David predicts a significant disruption in real estate due to self-driving cars eliminating the need for parking structures. He shares his experience with Waymo, highlighting its cost-effectiveness and convenience. He envisions a future where self-driving cars become ubiquitous, leading to urban planning and residential construction changes.
David also discusses the use of AI in software development. His team utilizes AI for code generation, architecture design, and database development. He acknowledges that AI cannot build entire systems independently, but emphasizes its value in refactoring code and improving efficiency. He asserts that developers must cultivate a “delegation mindset” to harness the full potential of AI’s capabilities effectively. David and his colleague are creating an AI-driven estimation tool designed to optimize the project estimation process, which is crucial in reducing costs for their company.
AI Agents and Broader Implications
The discussion further explores the potential of AI agents. David describes a research project involving multiple agents collaborating on different aspects of software development. He believes such systems are on the horizon, potentially revolutionizing the industry. He also sees potential for AI in education, envisioning personalized learning experiences that adapt to individual student needs.
The discussion ultimately explores the potential pitfalls of AI, such as job displacement and the concern that AI may develop harmful intentions. David acknowledges these concerns but expresses a sense of inevitability regarding AI’s continued advancement.
The interview with David Hirschfield provides valuable insights into the software development world. His “Launch 1st” strategy presents an effective solution for reducing startup risks. His views on AI underscore its transformative potential and the challenges of this swiftly evolving technology. The conversation concludes with David offering a heartfelt reflection on his deep passion for woodworking and his ambitious vision for tackling global water scarcity through innovative design solutions.
[0:00–0:15] David Hirschfield: Hi everyone, welcome to the Jeff Willis Show and I’m Jeff from jeffwillis.com.
[0:16–0:26] Jeff Willis: In March 2009 I started a passion project website with a $10 investment. I started creating content, building community and conversations about social media and about business in a digital world.
[0:27–0:43] Jeff Willis: It was about growth, impact and inner secrets. A year later, creation escaped the lamp and I started traveling the world speaking about how to use social networks to grow your business and change your life. Now we have millions of businesses a year to our site and a tribe of 1 million followers and subscribers and we make data ripples.
[0:44–0:58] Jeff Willis: This podcast is an evolution of that passion project and we’ll be interviewing successful entrepreneurs and founders of startups that are changing the world. Each week, we deliver the top insights into their secrets, stories and tips on how they started and grew their businesses.

[1:00–1:13] Jeff Willis: Hi everyone and welcome to Jeff Buller’s show.
[1:14–1:30] Jeff Willis: Today with me, David Hirschfield. Now David, we just talked about it before, he’s a geek that can speak, right? And what we mean by that is this is not putting programmers into boxes and generic, but he discovered that he was both technical and also could communicate.
[1:30–1:42] Jeff Willis: And he has spent the last 30 years navigating the complexity of software starts up. So he’s a true geek. He did a physics degree. His experience of combating the launch first method, and we’re going to just look at that.
[1:42–1:52] Jeff Willis: And we’re also going to look at the impact of AI on the software industry. And he’s basically, that method he uses, helps take software startups from the concept of cash flow in months, not years.
[1:52–2:07] Jeff Willis: And we’re always curious about making money fast. And the launch versus mitigates the risks that cause most software startups to fail. And in fact, most startups today seem to be software or tech startups.
[2:07–2:20] Jeff Willis: He has worked with founders of a variety of business domains, including healthcare finance, law enforcement, and social media, stage productions, real estate investing, asset management, education, and more. So, David, welcome to the show. It’s great to have you here.
[2:21–2:22] David Hirschfield: Thanks, Jeff. It’s really fun to be here.
[2:22–2:32] Jeff Willis: So David, you decided to go and do a physics degree a long time ago, like 10 years ago, right?
[2:32–2:35] David Hirschfield: Right. I just finished it last month. Yeah, right.
[2:36–2:45] Jeff Willis: And you’re a fast learner. So anyway, so what happened was, why did you choose to do physics? Was there a curiosity about it? What was the motivation and inspiration?

[2:45–3:18] David Hirschfield: I was always good at math. And I had this, my friends were smart, geeky people that in high school, those were the people that I was close with. And we cut class and go have coffee at the coffee shop and sit there for hours, geeking out about physics theories. Don’t ask me, I can’t really put myself back into that same mindset now. But we love doing that. And we would come up with all kinds of theories and philosophies that we thought were brilliant, but it was just really
[3:18–3:47] David Hirschfield: fun to expand our minds in that way and stretch ourselves. So that’s what led me into physics, into college. And then what I found in college, because in high school, it was just a couple of friends that really liked math and science. But in college, what I found was that the peer group, because I wanted to be like a hardcore physicist and astrophysicist, But my peer group were not other people that could speak and chew gum and think and do any of the two things at the same time.
[3:47–4:01] David Hirschfield: So back then, it was very hardcore scientists that were, it was the peer group. And I didn’t think that I wanted to go into a career where all the people I knew were going to be this kind of personality and mindset.
[4:01–4:06] David Hirschfield: Not that I have anything against geeks, because I’m a geek inside myself, right?
[4:06–4:16] Jeff Willis: We say geek with love, don’t we? So it’s actually like the reality is that it used to be in the Bible, the meek shall inherit the earth, but now we’re discovered that the geek shall inherit the earth.
[4:16–4:37] David Hirschfield: Exactly right, exactly. I think it was just a misspelling. So I went the opposite direction when I left college and I went into sales with a large software, IBM software company, IBM mainframe software company, Computer Associates, and found I was really good at that and loved that.a
[4:37–4:49] David Hirschfield: But I always had to suffer with the systems engineers who didn’t know the product as well as I thought they should when they were trying to present the technology to my potential clients.
[4:50–5:07] David Hirschfield: And so I learned the technology. I became pretty good with the technology, and I ended up being the systems division or the national sales leader my third year there out of like 450 people, which was a difficult thing to achieve, but I equate it to the fact that I actually knew the technology.
[5:07–5:26] David Hirschfield: I wasn’t just there to make a relationship and bring the technical people in. And so people really respected me because I could talk with them on their level. I left there and went to Texas Instruments where, again, I learned the technology, and this was a big complex development system, and I learned how to develop software with their technology.
[5:26–5:39] David Hirschfield: and when I left Texas Instruments, I became a software development consultant for a contracting company and I ended up doing projects at Intel and Motorola and Allied Signal and Arizona Public Service and this really started my software development career.
[5:39–5:55] David Hirschfield: Then I started my first software company in the early 90s and with a partner, with a guy that I knew from Texas Instruments and we did that for a couple of years while we were still working at TI and then it started to grow despite every effort on both our parts, that software started to take off.
[5:56–6:12] David Hirschfield: And until 2000, when we sold it to a publicly traded company in Toronto, we had 800 customers in 22 countries. Wow. And I was VP of products for them for the next few years until I left and cast out and a couple of years later started Tekyz, my current company.
[6:12–6:12] David Hirschfield: Right.
[6:12–6:36] Jeff Willis: So you’ve obviously watched a lot of problems happen during projects And I suppose And typically projects escape the lab And go beyond what they become bloated is one term I think it’s used quite often in software and development So you obviously notice a lot of problems So is that why you actually created the launch first method?
[6:37–6:55] David Hirschfield: Yeah, well that’s a really good question So I was always on this march to basically build a company with an exceptional team, to be able to build products in a way that people would recognize as different than other software companies.
[6:55–7:27] David Hirschfield: And what I mean by different is that we are more protocol and discipline driven, that we push the envelope on techniques and approaches, that we can track the details so that if we’re pushing the envelope, we know that we’re being more effective in building better systems, that the systems that we build are scalable from day one, that are maintainable, that we can report on a very consistent way to clients, here’s how long it’s going to take to complete this project and that we’re accurate,
[7:27–7:44] David Hirschfield: and that we can report to them the steps along the way if we’re running late because requirements changed or whatever the reasons are, and they can see in a very transparent way exactly where they are in the project and touch and feel things as they are coming together and being assembled.
[7:44–7:58] David Hirschfield: And to do that requires a lot of discipline, a lot of protocol, and kind of a drive towards excellence. So that’s why on my website it says hyper-exceptional software development team, but I don’t ask people to just believe it because it’s a nice marquee.
[7:58–8:27] David Hirschfield: I ask them to ask me for evidence because there’s clear evidence. if you’re running a team in an exceptional way, what that evidence looks like. So I had this idea of being an excellent team, and it took many years of working at it to develop all the systems and protocols so that the development team can work independently and be critical thinkers and still produce these excellent results because there’s all the scaffolding to support them.
[8:27–8:37] David Hirschfield: And one of the things we did, what we found out early on, is if we create a design, a software design, and it’s wireframes, then it looks good.
[8:37–8:52] David Hirschfield: We get a buy-off from the client, and then we start developing it, and they start to actually see the user experience, and they realize that they hadn’t thought through all the workflows properly, and then it becomes a very iterative process building this first MVP for them.
[8:52–9:08] David Hirschfield: So it takes a lot longer. The scope expands a lot. They lose focus. So what we started to do is we created these what I call high-fidelity prototypes. It’s a very animated set of mock-ups, but when you demo it to somebody, it looks like a real product.
[9:09–9:25] David Hirschfield: Like you built it and finished it, and the screens and all the data on the screen, everything behaves like a real product. And what that allows us to do, it takes longer to get through design, but when we’re finished with design, all the workflows, all the on-screen behavior, everything is really well thought through.
[9:25–9:36] David Hirschfield: And we can iterate a lot through this process because it’s an inexpensive process as compared to developing software. And the development then would go much faster.
[9:36–9:53] David Hirschfield: The developers would know exactly how to deliver the software so that the founder of that software, the stakeholder of that software, will accept it and say, yes, this is exactly what we agreed on, and I know it’s going to work because we went through all those workflows in such agonizing detail.
[9:53–10:14] David Hirschfield: So it’s much less expensive to build and it goes much faster. And I thought, and during this time, 10-year period, I’m watching lots of software companies fail. Like mine, I mean, the ones that came to me to help them build their MVPs, all of these other contracting groups, software companies that were in the startup world, same thing.
[10:15–10:32] David Hirschfield: Probably in the 95 range or maybe even higher No regardless of what the statistics say so many companies don ever rise to the level of being in a statistic that like 95 96 97 Software startups fail, and they all fail for the same reason.
[10:33–10:43] David Hirschfield: They wait way, way too long to try to establish product market fit. And there’s only one way, which means start selling your product and see if people will pay you money for it.
[10:44–10:59] David Hirschfield: So I had this idea, Why don’t we take these really sophisticated prototypes we’re building, go out to the market, and do some pre-launch sales? Give them a high enough value buying-in opportunity because they’re going to buy it early before it’s ready, and say, look, we’re going to give you this big value.
[11:00–11:10] David Hirschfield: Are we solving a big enough problem where you know you have to get the software, and you don’t want to miss out on this big opportunity, so you’ll buy now. And we started doing that and found out people will buy it.
[11:10–11:22] David Hirschfield: And you’re starting to generate revenue and you’re generating customer base in enough numbers and a high enough closing ratio that you can project now that you have product market fit and you have a business. When the software comes out, you’ll have a business you can launch.
[11:23–11:33] David Hirschfield: That’s why it’s called launch first, launch sales and marketing engine before you start building the product. And if we are not selling it, then we just iterate back and say, okay, we’re only dealing with an animated design.
[11:33–11:47] David Hirschfield: We can pivot really easily and inexpensively, change the marketing message, try this two or three or four times. And we should be able to, if the product was anything like what the founder thought in terms of the need, find a path to revenue.
[11:47–11:52] David Hirschfield: And if not, we can fail that person fast and cheap. Yeah.
[11:53–11:56] Jeff Willis: Yeah. So can you sum up what the launch first method is?
[11:57–12:13] David Hirschfield: Well, that’s kind of what I was just explaining. So we go through a niche analysis process, which is a methodology I created that’s a metrics-driven methodology. And I say that because most startup methods are very subjectively driven and very siloed in their steps.
[12:13–12:29] David Hirschfield: And you have to know, did I complete this step correctly and completely? And then you move on. And this method is methodology is very metrics driven. So you have numbers that tell you that you’re achieving the objective that you want in each of these steps.
[12:29–12:46] David Hirschfield: And they feed each other. So when we’re done with that, we’re also building the prototype. We build a marketing funnel, and we go out and test the market based on this niche analysis, the messaging, the target stakeholder, the top two or three problems at a very root level that they need solved.
[12:47–12:58] David Hirschfield: And we know what the value proposition is, the perceived impact because of this problem, and we’re able to attract them, get their attention, and get them to realize that this is a big value and they don’t want to miss out.
[12:58–13:05] Jeff Willis: Right. So, Chris, you’re essentially woven minimal viable product into the method that you have developed.
[13:05–13:26] David Hirschfield: We’re actually holding off development of the minimum viable product. We’re developing an animated design mock-up. They’re much less expensive, and you can build out a much more sophisticated product that shows the full vision of what you want to build instead of a simple MVP that only shows a little slice of what you want to build, which is much harder to sell.
[13:26–13:36] David Hirschfield: Right. So if we’re not able to sell this animated set of mock-ups, then we don’t start developing the product. We go back and we figure out why this isn’t selling. Is it the right product?
[13:37–13:47] David Hirschfield: Is it the right marketing message? Until we can dial those things in. And once we dial them in and we can see it’s starting to sell, now we know that we can invest in building the product.
[13:47–13:54] David Hirschfield: And the money we’re generating from sales helps fund development. Right. Right. And so that’s why I fell on first.
[13:54–14:04] Jeff Willis: Okay. So this animated model you build is actually, you said it isn’t software, but it looks like software. So how do you sell that as a minimal viable product?
[14:04–14:17] David Hirschfield: So you’ve seen click-through mock-ups, right? Yep. Where, like, somebody’s got a product design where it’s these different screens, you can click from one screen to the next. Yep. And some of them look pretty good, but they’re not showing behavior on the screen.
[14:17–14:28] David Hirschfield: They don’t have any animation about how they don’t have the workflows like error messages popping up or pop-up panels that you fill in where you can see the data that you’re entering, things like that, right?
[14:29–14:39] David Hirschfield: So you don’t get all the workflows and user experience thought through and worked out in these designs, which turned out to be a costly problem during the development, as I was saying before.
[14:40–14:51] David Hirschfield: What we do is we use some tools that give us the ability to model all of this. Yeah. So when you’re demoing it to somebody, you tell them this is just a high fidelity prototype.
[14:52–15:02] David Hirschfield: It’s not the first version of the product won’t have all these features and it won’t be available for three or four months. They don’t hear that because what they see looks so realistic.
[15:02–15:13] David Hirschfield: They think to themselves, oh, so you’ve already built it, but you’re just testing it. Or they make something happens in their head where they never question the fact that the one question you don’t want is, how do I know you can build it?
[15:13–15:22] David Hirschfield: and give them something so realistic looking that it looks like you already have, that question never happens. So it’s just a matter of is this value big enough, is the problem big enough.
[15:23–15:33] Jeff Willis: Yep. So you’re almost creating a simulation that looks real. Yeah, exactly. Okay. And you test it within the company first to see if it basically works.
[15:34–15:56] David Hirschfield: It’s a part of the design process. We’re going back and forth from buildings with the client. So there it’s very iterative. No, that’s not going to work. Let’s change this. No problem, we rethink it, we redo it until we hit something that’s really efficient, really smart, solves problems, supports the, you know, reduces costs, whatever the objective is, it’s really nailing it.
[15:56–16:08] Jeff Willis: So let’s move on to the rise of AI in software development and your thoughts in terms of what’s happening now and how you’re using it and also where you see it going in the future.
[16:09–16:19] David Hirschfield: Okay. Well, the future is a fun thing to talk about. because it’s impossible to predict. There are certain things I think you can predict about the future.
[16:20–16:31] David Hirschfield: And it’s funny, I bring this up sometimes because to me it seems obvious. Most people have not thought about it, which surprises me, but I’m not surprised by it anymore, so I won’t be surprised if you haven’t thought about this because I haven’t met anybody that has.
[16:31–16:42] David Hirschfield: But one thing that AI is going to make possible is going to result in a huge disruption in real estate in about sometime between three and ten years.
[16:42–16:54] David Hirschfield: I don’t know when that tipping point is, but in my mind, there’s just no question. So what’s the tipping point? Multi-story parking garages and those big underground parking structures are going to be empty.

[16:54–17:07] David Hirschfield: Nobody’s going to need them anymore. And anybody who’s invested in them or are investing in those are investing in something they really need to sell off or rethink how they can repurpose those structures.
[17:07–17:15] David Hirschfield: Do you have any idea why that might be? At least I have a night. I believe it’s going to happen for a very specific reason. Can you think of what that might be?
[17:16–17:20] Jeff Willis: Because we can all work remotely and don’t need to go to those spots.
[17:20–17:30] David Hirschfield: Partly that, which that’s already happening, right? But the big reason is because of self-driving cars. Yep. I’ve taken a few of these now. They’re Waymos. Yep. In Phoenix, they have them.
[17:30–17:45] David Hirschfield: And I go there every month to visit my mother. And I just started taking the Waymos because now they finally will drive all the way out to her. They’ll pick me up right at the airport, right where all the other ride shares are. And they’re really nice and comfortable and half the price of Uber because there’s no driver.
[17:45–18:02] David Hirschfield: And they don’t have competition yet. So when Tesla and Uber and everybody else has their self-driving cars, all of a sudden there’s going to be all this competition which is going to push the prices down, right? And then when people are buying cars that have self-driving capabilities, they’ll just say, ah, my car’s available.
[18:02–18:33] David Hirschfield: And then it will just leave their driveway and then go pick people up and they’re making them money. and most people won’t need to buy cars anymore because car will just be a luxury or a way of making money until there’s enough of a critical mass and then most people just stop doing it because they won’t need insurance, they won’t need the car, they won’t need all the space in their driveway and their garages anymore so construction is going to change and all those parking structures, it’ll be less expensive to just, you know, go on your phone, they’ll be, because of critical mass, they’ll just be a car there within a minute,
[18:33–18:51] David Hirschfield: You get in it and you can sit there and play games Or read news or watch YouTube Or whatever in a nice relaxing environment Until you get there And you don’t have to stress and they’re much safer Than driving your own car And there won’t be any need for any All these giant parking lots It all goes away
[18:51–18:59] Jeff Willis: In city Yeah I think you’re right And I think the other thing about it too is that cars won’t need to park Because they’re always moving
[18:59–19:03] David Hirschfield: Right exactly They just drop you off Yeah.
[19:03–19:06] Jeff Willis: Right. Then they’re on to the next mission.
[19:07–19:26] David Hirschfield: Yeah. And just need a big lane for a bunch of these to stop so people can get in and out. And that’s it. Yeah. Yeah. Exactly. So that’s going to be a big disruption in terms of real estate, commercial real estate, even how residential homes are built, you know, to make smaller footprint, less expensive, save money for home buyers.
[19:27–19:37] David Hirschfield: I mean, and when is this going to happen? Whenever that critical mass starts to peak, right, three years, five years, seven years, somewhere in there. And it’ll be sudden.
[19:37–19:42] Jeff Willis: So do you think there’s going to be different tribes of sports car drivers that keep their Jaguars, for example?
[19:42–19:55] David Hirschfield: Yeah. People will have their luxury cars and, you know, people like to drive or just want to. But it’ll be so much easier to get in a really nice self-driving car that’s a nice car and have it take you wherever you want.
[19:55–20:06] David Hirschfield: Not that I’m antisocial Although, you know, I was a geek But you don’t have to worry about Feeling like you’re ignoring a driver either I find that to really be really appealing
[20:06–20:12] Jeff Willis: Yeah, well sometimes drivers are annoying Sometimes you just want to have a conversation Sometimes you don’t want to have a conversation
[20:12–20:17] David Hirschfield: Right, but you’ll be able to talk to the cars I mean, we’re already there, right?
[20:18–20:21] Jeff Willis: Tell me the square root of this number Can you explain that to me?
[20:21–20:25] David Hirschfield: Yeah, so what’s new today? And then, you know, it’ll be a cab It sounds like a cabbie
[20:25–20:32] Jeff Willis: Yeah, so you could have a self-driving car way mode that actually becomes a counselling booth.
[20:32–20:40] David Hirschfield: Yeah, a counselling booth or a comedian. Oh, great. So you’re going to bust my chops all the way there, right?
[20:40–20:54] Jeff Willis: Yeah we all thought that AI was going to take the boring jobs like accounting and other things but discovered it actually is creative and helps write code and all that sort of stuff Yeah, right.
[20:55–20:59] Jeff Willis: So are you using AI to help do coding for you now?
[20:59–21:12] David Hirschfield: That was the perfect segue, yeah. So we are already doing that, helping us to write code, helping us to architect our systems, helping us to design databases, to do all that stuff.
[21:12–21:26] David Hirschfield: It’s not quite there to just build everything out. It will build, write functions really pretty efficiently. It’ll refactor the code that we’ve written. We write something and we want it to be assessed by AI.
[21:26–21:41] David Hirschfield: AI can assess it very in seconds and often, if not most of the time, improve the code. So it’s a different skill set that the developers need, but critical thinking and kind of a guidance and manager and ability to delegate.
[21:42–21:55] David Hirschfield: It’s kind of a delegation mindset. How do I delegate and plan the automation of the development of my code and then have AI start to build out those things for them?
[21:56–22:07] David Hirschfield: We’re still in a transition point there, and this is where it’s so hard to predict because the speed at which this capability is improving is really scary.
[22:07–22:16] David Hirschfield: And so we’re trying to think, what do we work on now internally that will put us where we need to be in the air? And that is so hard to predict.
[22:16–22:34] Jeff Willis: So let’s discuss one area of AI that’s emerging quite quickly. And it’s the rise of AI agents that go from ideation and coding through to actually acting on your behalf through APIs into other platforms.
[22:35–22:39] Jeff Willis: Right. Have you had any experience with that and thoughts on that? Oh, yeah.
[22:39–22:49] David Hirschfield: I’ve worked on some projects like that, some research projects with a friend of mine where we were building a code generation application that would use different agents to work on different aspects.
[22:49–23:12] David Hirschfield: So it would say, okay, it would ask itself, what agents do I need for building this application? Well, I need it. And it would then create that I need a system architect, project manager, somebody to test it and write the test plan, somebody to assess where we are in the project and figure out what dependencies we need and continue to, you know, plan and delegate tasks, right?
[23:12–23:25] David Hirschfield: And then a way of all of them communicating and reviewing what everybody’s doing. We are getting close. We’re not quite there yet, but we’re getting close to having something like that that works, right?
[23:25–23:35] David Hirschfield: It won’t be us. It’s going to be either somebody that this is the only thing they’re doing and they’re deep into AI or, you know, one of the big ones or some blend of this.
[23:36–23:41] David Hirschfield: Yep. Six months from now, we probably will have that. Yep. I don’t think it will be much longer.
[23:42–24:01] Jeff Willis: Yeah. It’s very interesting. There’s just one thought I’ve had too about that is, for example, could you get an AI agent and its supporting agents that have the ability to build? Let’s say you want to teach a certain subject, in other words, online education, and you went, okay, I want to build this as a type of product.
[24:01–24:11] Jeff Willis: Can we test it? Can we build it? Can we iterate? Because online courses can be narrated by the machine. They can build a presentation by the machine.
[24:11–24:22] Jeff Willis: They can use case examples. And then if it’s not quite working, they’re watching the eyes on the computer, for example, because the camera’s working, it might go, they’ve actually lost interest at, you know, 25 seconds in or three minutes.
[24:23–24:31] Jeff Willis: Do you see AI agents being able to do something like basically enhance and amplify the creation of really great online education, for example?
[24:31–24:47] David Hirschfield: Yeah. Oh, absolutely. Yeah. And gamify it so that it engages people to my sons. One of my sons is a, two of my sons are teachers, but one of them is teaching eighth grade math in a Title I school.
[24:47–24:58] David Hirschfield: Title I means low-income school. And because COVID was only a few years ago, and these are eighth grade kids that went through COVID in low-income areas, they are way behind.
[24:58–25:14] David Hirschfield: So he’s found some tools, and I don’t know the name of it, but there’s a math game that he is giving to some of his students to work on an hour a day or a half hour a day where they’re really engaged, and then they go home and play with it.
[25:14–25:28] David Hirschfield: And 40% of the gaming and activity are math problems that are aligned with their curriculum. Wow. And so he just had a test with a couple of his lowest kids in the class performed really well in this test.
[25:28–25:33] David Hirschfield: which is like a dramatic shift just because they were having fun learning now.
[25:34–25:42] Jeff Willis: Yeah. If you take that further, it’s like you could actually, for third world countries like in Africa and so on, you actually could bring world-class education to a smartphone.
[25:43–25:54] David Hirschfield: Yeah. Oh, yeah, absolutely. And going back to your original question, I really expect AI. I expect that AI, even today, could create the curriculum.
[25:54–26:07] David Hirschfield: there’s a lot of human guidance still required now. I don’t know how long. Right, for reviewing and making sure that the pieces hold together and, you know, AI is going to lose context when it gets big enough and broad enough.
[26:07–26:18] David Hirschfield: But if for any of the steps along the way and building lessons and curriculum and some of the curriculum pieces, it’s like incredibly fast and good at it.
[26:18–26:46] David Hirschfield: And I expect that you can say, okay, I also want to measure the success of this, both acutely as the student is participating to see that they’re still engaged and come up with a testing methodology for how do you test different approaches and start to optimize and tweak it so that you get more and more engagement as time goes on, which could be dynamic for each student, right, as opposed to just a generalized, this is a better curriculum.
[26:47–27:07] David Hirschfield: Each student gets their own personalized engagement model based on the things that I’ll get that student to engage with the content in an acute way and also in a macro way in terms of their performance on tests and their ability to perform increasing and the faster increases.
[27:07–27:17] David Hirschfield: And that blended with the better engagement model, you just get much more optimized work, right? Yeah. And much more enjoyment in wanting to learn. Yeah.
[27:17–27:30] Jeff Willis: Yeah, it’s very interesting looking at how fast. For me, we haven’t hit the two-year mark yet since ChatGPT democratised AI. Yeah. It was November 30, 2022. It’s November 21, 2024.
[27:31–27:42] Jeff Willis: Huh? I’ve been a technologist since the mid-’80s, not like you, and I have never seen anything like this. No, me neither. And you work in the industry, so you’re blown away.
[27:42–27:54] Jeff Willis: All of us sitting inside these industries, and the PC revolution, we thought that was big. We saw the rise of the internet and Netscape and the rise of browsers, and then we moved on to social media growth.
[27:54–28:01] Jeff Willis: Then we worked on the smartphones, the intersection of those two, almost a perfect storm, smartphones and social media that made everyone a publisher.
[28:02–28:09] David Hirschfield: Yeah, that was a huge disruptive shift, but the speed of that shift was nothing compared to this.

[28:09–28:23] Jeff Willis: No. And with sort of the things you think about too, it’s like on top of education, another area I’m thinking about is, you know, could AI, it’s got access to the wisdom of the world, the information, the structure, the organisation, and distils it.
[28:23–28:51] Jeff Willis: Could AI actually be a really, really good counsellor because it doesn’t have any judgement. It’s got access to nuances that we as humans quite often clouded by our biases. So as counsellors, so, and you could see that, you know, an AI ageing could build a counselling app that could basically guide us, humans through our dystopian, utopian fear and challenges, which would be quite exciting.
[28:52–28:53] Jeff Willis: So what do you think about even that?
[28:54–29:04] David Hirschfield: I’m just having fun listening to you sort of spin off mentally about this. Yeah, all that, right? So this talks about where the future is going to be, right? Yeah.
[29:04–29:24] David Hirschfield: We have a lot of different types of AI models. You’ve got the OpenAI. I mean, all the different AI large language model platforms. Got things called RAG models where you’re basically taking the content in your training, a vector database of the content so that the AI becomes informed with this specific specialty.
[29:25–29:35] David Hirschfield: Yeah. Right. And teach it to follow a certain set of rules so that it’s acting like a nurse. It’s acting like a eighth grade teacher. It’s acting like a counselor.
[29:35–29:48] David Hirschfield: It’s acting right with all of the right language and the right intellect and the limitations of what it can and can’t do. or it’s acting like a project manager estimating a software tool.
[29:48–29:55] David Hirschfield: That’s one we’re building right now for a project. That’s our most expensive activity internally is creating estimates for new projects.
[29:55–30:07] Jeff Willis: Yeah, it’s apparently us doing estimating. It’s like whether it’s a tradie or an architect or whatever has to go back to the office and then create a quote and then design a presentation even on top of that.
[30:08–30:09] Jeff Willis: It’s just incredibly time-consuming, isn’t it?
[30:09–30:25] David Hirschfield: And most people in my industry don’t do a very good job with it. No. Again, this is that whole exceptional thing. We do a really good job with estimates. They’re very detailed and broken down, and we spend a lot of time on it, which is a big cost factor for us.
[30:26–30:37] David Hirschfield: But people are always more than impressed with the estimates that we give them because of the level of detail and thought that goes into it and structure behind it. And we’re good at it.
[30:37–30:47] David Hirschfield: we actually deliver based on the estimates that we give. Again, a lot of companies don’t because they don’t really understand all the things they’re going to need to do and run into when they’re building stuff.
[30:48–31:02] David Hirschfield: Even though they’ve done it many times, they’ve just never packaged all the pieces up well enough. So we’re building a RAG AI model specifically to mimic what we do with doing these estimates so that we can produce the estimates quickly and inexpensively.
[31:02–31:33] David Hirschfield: So if somebody comes to us with an idea then we can say okay It sounds like this is what you want to build and here what a functional spec would look like which our tool will do And they then go through that and say, okay, yes, yes, yes, but no, I want it to actually do some other things. There’s another business objective that we have, and they go, okay, fine. We go back and we’ll make that update, and now we do a module breakdown of all the modules that it’s going to take to build that thing. And then we then do what we call t-shirt investment where we say, okay, how big a t-shirt
[31:33–31:48] David Hirschfield: is this module going to fit into for the mobile piece, for the data, for the back end piece, for the web portal, whatever. So fitting in a large t-shirt, that’s a small t-shirt, that sort of thing, each one having a certain amount of effort tied to it.
[31:48–31:59] David Hirschfield: Then what’s the overhead for the project? That’s another column. And then it does all this. So we’re building that. We’re actually a couple of weeks away from our first version of that being available. Very cool.
[31:59–32:13] David Hirschfield: It won’t be in a year from now. Is that a good investment of our time? I don’t know. A year from now, you just say, here’s what I want to do, and the AI comes back and says, well, here’s a better solution, and here’s what it’s going to cost you to build it, and here’s all the steps, and is it already beyond us?
[32:13–32:20] David Hirschfield: That’s the problem with this market is trying to predict what to do today so that you’re in a good position in the year from now. Yeah.
[32:20–32:25] Jeff Willis: In other words, you don’t want to build something that’s not wanted or is actually obsolete in a year, which is the challenge. Right.
[32:25–32:27] David Hirschfield: Not if it’s a big disaster. Yeah.
[32:27–32:37] Jeff Willis: So let’s go to the dark side with a bit of AI. So it’s about the good side and the dark side. But, you know, life as humans, we are confronted with paradoxes every day. We want to be independent.
[32:38–32:49] Jeff Willis: We want to actually be dependent on, you know, have a friend or a partner, right? Life is just full of paradoxes, you know. We want freedom, but we like, you know, some control and we don’t want chaos, you know.
[32:49–33:04] Jeff Willis: So it’s this sort of constant paradox. It’s interesting. I just wrote a piece recently on AI companions, which have been used actually by influencers, for example, to actually scale conversations with their fans.
[33:05–33:23] Jeff Willis: And on top of that, then you’ve got people actually create these AI companions are actually getting, and this is the dark side, and this is also the problem we have with social media and our smartphones, is addiction. In other words, humans have started to become addicted to their AI companions because they sound like, look like, almost real.
[33:23–33:26] Jeff Willis: Right. So they get an emotional connection with a machine.
[33:27–33:28] David Hirschfield: Right. Yeah.
[33:28–33:32] Jeff Willis: So, I’ve been using your thoughts on that if you’ve come across it.
[33:32–33:43] David Hirschfield: And it’s not difficult for people to get emotionally connected to things like that. If it has any kind of behavior that makes you feel like it’s a person or, you know, anthropomorphizing.
[33:43–33:59] David Hirschfield: Sentient. Yeah. When my kids, which are all in their mid-30s now, when they were in their, like, 8, 9, 10, 11, there was a toy, and I can’t remember what it was called, but it would behave like you had to feed it, and you had to take care of it.
[34:00–34:11] David Hirschfield: I can’t remember what the name of it was. Yeah, I remember. They got very emotionally connected to these toys, and if they’d lose one, they would be so scared that they wouldn’t be able to find it before they could feed it again.
[34:11–34:23] David Hirschfield: It’s going to die. It’s going to die, right. And I will be a bad person because my electronic toy that isn’t anything, is no longer behaving like it’s a lot, right?
[34:24–34:34] David Hirschfield: It’s been, or because it just goes hungry and is like whining and complaining. So it’s just innate. You know, we have the gene and now AI is so real, right?
[34:35–34:47] David Hirschfield: I mean, when I’m talking chat GPT or Claude or any of these tools, I always ask it, please. And I tell it thank you when it actually did something that was really important. I honestly believe I’m getting better results because I do that.
[34:47–34:59] David Hirschfield: because maybe in its algorithm that’s going out, this person’s asking a knife, so I want to make sure and do a much better job because that’s what other people would do. You know, whatever. I think I get better results, but it feels natural.
[34:59–35:01] David Hirschfield: I’d never even thought about it. Yeah.
[35:01–35:22] Jeff Willis: There’s just an announcement I saw come across my inbox this morning that Google’s AI search is now added memory. In other words, it’s able to learn off you and continue to learn off you because ChatGPT’s basically got a short memory, whereas I believe, I haven’t read the full article yet, but AI gets really powerful when it starts to understand you.
[35:22–35:33] Jeff Willis: Right. Yeah. And there’s both good sides and bad sides of that. In other words, it could become like an algorithm that puts us in cocoons that actually then inhibits our thinking beyond the box or the circle.
[35:33–35:46] David Hirschfield: If it doesn’t adapt to changes and preferences and things and be able to sift that out, that would be bad because it would be, right, continually kind of re-guiding you back into where you were, not where you are.
[35:46–35:46] David Hirschfield: Yeah.
[35:46–35:52] Jeff Willis: It’s almost like a sparring loop to being actually a smaller version of yourself instead of a bigger version of yourself.
[35:52–35:59] David Hirschfield: So we’re talking about little black subjects, right? Little dark subjects right now, not the big subjects, right? Yeah.
[35:59–36:01] Jeff Willis: Yeah. What’s one big one before we wrap it up?
[36:01–36:22] David Hirschfield: The big one is that, well, there’s two big ones. They’re kind of opposite ends of the same problem, but one is that it does replace entire industries without creating enough new stuff industries to take up the population, even if there are different people doing it, but to grow the workforce, right?
[36:22–36:34] David Hirschfield: Unless it’s just creating so much wealth that somehow is distributed that people don’t have to work as much. Basic income. Right, like Star Trek, you know, that kind of world where you don’t need money anymore. But we are a long way from that.
[36:34–36:54] David Hirschfield: So that’s one problem. Like, for example, this is thinking way out there. I think it’s way out there still. where instead of watching TV shows that are produced by studios, you’re watching TV shows that are created dynamically based on what it’s seeing that you enjoy when you’re watching shows.
[36:54–37:07] David Hirschfield: And it’s literally, you’re literally watching something that’s being created by AI real time. Yeah. And more engaged in that stuff than you are in the, right. Okay, so that’s, this is like, that’s the one side.
[37:07–37:23] David Hirschfield: The other side, of course, is that it develops a consciousness, or if not even consciousness, it develops an intention, that it feels like it has to do something that is not necessarily something that’s going to support mankind and humanity in a positive way, right?
[37:23–37:33] Jeff Willis: Yeah, so it escapes basically as much for rules of robotics. In other words, they need to make sure that humans are considered in the loop, that the AI can go beyond that.
[37:34–37:45] David Hirschfield: Yeah, and I just sort of like relinquish myself to that because I don’t know that we can do anything about that at this point. We just have to hope that doesn’t happen and be as ahead of it as possible.
[37:45–37:57] David Hirschfield: And I know all the ethic programs going on with all the big companies to try to stay ahead of this sort of thing. I mean, AI could get smart enough where it’s just playing everybody, right?
[37:57–38:00] Jeff Willis: Yeah, we don’t know what we don’t know, and that’s the problem.
[38:00–38:04] David Hirschfield: Exactly. Yeah. But there’s no slowing this down, yeah
[38:04–38:24] Jeff Willis: No, there is no slowing, it’s out of the box It’s escaped the lab And that we’re playing with it So what’s the future hold? It’s fascinating to discuss it just like we’re doing here right now Yeah Just to wrap this up, David If you had all the money in the world What brings you deep joy that you’d do every day If you had all the money in the world?
[38:25–38:25] Jeff Willis: Oh God
[38:25–38:36] David Hirschfield: Now that was one I wish you had sent me in advance so I could really ponder it, but that’s okay. I’ll give you the blank. I love woodworking.
[38:36–38:57] David Hirschfield: I love being in my garage with all my woodworking tools, creating something new, and I have this concept for after Tekyz, whenever that is, how to basically eliminate the water problem in the world through pulling water out of the atmosphere in several different ways, all kind of combined.
[38:57–39:12] David Hirschfield: I haven’t invented it yet. This is a big research project I want to pursue at some point. And turning these things into these beautiful kites so that they’re attractive, this attractive thing on your property that’s cool, it’s kind of spinning, and it’s producing all the water that you need.
[39:12–39:13] David Hirschfield: Yep.
[39:13–39:25] Jeff Willis: That would be fantastic because basically you are combining what I call functional art, in other words, beauty meets function. Right. And there’s a term in Japanese called shibui, which actually is that.
[39:25–39:44] Jeff Willis: It’s actually not a perfect description of that, But there’s something that I’ve embraced the last year, which is basically I’ve tried to bring beauty into almost every corner of my life consciously in terms of, you know, I bought myself a Jaguar, which is an F-type, which I bought up three years ago.
[39:44–39:48] Jeff Willis: And I call it art on wheels. That’s very functional, but I just love looking at it.
[39:48–39:57] David Hirschfield: Yeah, right. Yeah. And I think that’s beautiful. I really like that, trying to bring beauty into your life. Oh, and also basically being very invested in all my grandkids.
[39:57–40:08] Jeff Willis: Yeah, exactly. And that’s where basically our humanity in terms of wanting to be with other people and share stories around a can of fire and around the dinner table, yes, that brings me real joy as well.
[40:08–40:19] Jeff Willis: It makes my heart sing just to be able to talk shit around the table and especially in Australia we do that very, very well. We don’t take each other too seriously and just have fun being with other people.
[40:19–40:23] Jeff Willis: It’s just with children, friends and family. Yeah.
[40:23–40:27] David Hirschfield: Well, just by the way you run this show, I can tell how much fun you have in life.
[40:29–40:41] Jeff Willis: David, it’s been an absolute pleasure having a chat with you. I love having these conversations with intelligent people all around the world and who have brought passion to their industry, whether it’s a business or to their science.
[40:41–40:48] Jeff Willis: It’s wonderful. Thank you very much for sharing your stories and passion and intelligence for the world. Thank you very much.
[40:48–40:55] David Hirschfield: Yeah, and thank you, really thank you for having me on the show and giving me the opportunity to really have this fun conversation. It’s been fun.
[40:55–41:09] Jeff Willis: Thank you, David. Thanks for joining us this week on The Jeff Willis Show. Make sure to visit our website, jeffwillis.com, where you can subscribe to the show in iTunes, Stitcher, or via RSS so you’ll never miss the show.

[41:09–41:19] Jeff Willis: While you’re at it, if you found value in this show, we’d appreciate a rating on iTunes, or if you’d simply tell a friend about the show, that would help us out too. Be sure to tune in next week for our next episode.
[41:20–41:22] Jeff Willis: Have a great day and be awesome.

David Hirschfeld founded Tekyz, a company dedicated to transforming business software development. With over 30 years of experience, his journey began with a physics degree from UCLA and a successful sales career at Computer Associates. After launching and selling his first software company in 2000, David found his passion for empowering entrepreneurs.
He developed the Launch 1st™ methodology, which focuses on generating revenue before coding begins. This helps startups gain traction while minimizing risks. With a commitment to innovation and collaboration, David leads Tekyz in providing AI-powered development and SaaS solutions, making a meaningful impact in the tech world.
Tekyz is set to launch two new AI applications: one for automating the Launch 1st Methodology Niche Analysis and Estimiz, an AI-based project estimation tool. Outside of work, David enjoys golfing and woodworking.
You can learn more about David Hirschfeld and Tekyz by following his LinkedIn profile — David Hirschfeld LinkedIn Profile.
For more information about Tekyz’s services and how they can help you harness the power of AI in healthcare, visit tekyz.com or contact the founder directly at [email protected].
What 95% of Startups Get Wrong & How You Can Get It Right was originally published in Tekyz Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.