AI Isn’t a Tool. It’s a Coworker: Why data and autonomous development will define the next era of software with Rich Walker

If AI feels like the biggest opportunity (and risk) in your business right now, this episode reframes the conversation in a way most leaders are missing: it’s not an AI problem, it’s a data problem—and more importantly, it’s a thinking problem.

In this conversation, Tonya sits down with Richard Walker, founder and CEO of Quick, to unpack what’s really driving outcomes in the AI era. Drawing on decades of experience structuring and scaling data across millions of forms, Richard challenges the default way most companies approach AI—as a tool—and introduces a far more powerful lens: AI as a coworker.

They explore why everyone has access to the same models, but vastly different results, and how the real differentiator lies in how you think, plan, and interact with AI. From treating AI like a human collaborator to building panels of experts and defining “mental models,” this episode offers a practical and deeply strategic framework for leaders navigating what’s next.

Richard also shares how his team is leveraging unique data sets to build entirely new capabilities, why most organizations are sitting on untapped data goldmines, and what it actually takes to move from experimentation to operational impact.

This conversation is a masterclass in slowing down to move faster—reframing AI from a speed tool into a strategic advantage that compounds across your business.

In this episode, they explore:

• Why most companies don’t have an AI problem—they have a data problem
• How treating AI like a coworker changes the quality of your results
• The concept of “toddler syndrome” and why AI can feel both brilliant and unpredictable
• Why prompting is less important than problem clarity and structured thinking
• How to use AI to write better prompts than you ever could yourself
• The power of “mental models” and how they shape AI behavior and outputs
• Why planning upfront eliminates endless iteration and rework
• How different AI models vary in strengths (and how to use them together)
• The concept of building a “panel of experts” to solve complex problems
• Why unique data—not AI access—is the true competitive advantage
• How companies are sitting on untapped data that can unlock new products and revenue streams
• The importance of enriching and structuring data before applying AI
• Why adoption is a leadership problem—and how to drive it inside your organization
• How AI is shifting roles from execution to orchestration and strategic thinking
• What the future of leadership looks like when managing both human and AI teams

Important Links:

https://app.technologymatch.com/solutions/quik-api

Transcript

SPEAKER_03 0:01

Welcome to Between Fires and Futures, a podcast about the real work of tech leadership, managing today's chaos while building tomorrow's business. I'm Tanya Tyrell, a three-time founder with two successful exits, and the founder and CEO of TechnologyMatch.com. Each week, in this podcast, I talk with the leaders doing the real work, solving for now, building for what's next, and leading through pressure, not perfection. This is the podcast for tech leaders fighting fires today and daring to build the future anyway. Welcome back to Between Fires and Futures. I'm your host, Tanya Terrell, and today we're talking about AI and data. Most companies think they have an AI problem, but in reality, they have a data problem. And behind almost every business process, there's a form moving data through multiple systems. Even today, that process is still broken. Data gets re-entered, workflows slow down, errors creep in, and customers feel it. And today's guest has spent more than two decades solving that exact problem. Rich Walker is the founder and CEO of QUIC. And since 2002, his team has built over 42,000 forms, defined more than 1.2 million data fields, and today processes around 1.3 million forms every month. That gives him a unique vantage point on what's happening right now with data. And everyone has access to the same AI models, right? But very few companies have the structured data needed to actually use them. Rich sees AI differently, not as a tool, but as a coworker, a very capable one, but one that still has what he calls toddler syndrome. So today we're going to talk about what that means, why data may matter more than the model you choose, and how the role of the software engineer is evolving. Rich, welcome to the show.

SPEAKER_00 1:47

Wow, what a great intro. I'm excited to be here, Tanya. Thank you.

SPEAKER_03 1:50

Yeah, I'm excited to dig into this. I mean, you know, nothing hotter than data and AI right now. And I think, you know, most companies that we're speaking to are treating AI like software, certainly like leverage, but something that you plug in, prompt, and get output from. But you said something that reframes that completely. You described it more like a coworker. So what do you mean by that?

SPEAKER_00 2:14

All right. I got to take you down a path, Tanya, just to kind of put this in perspective. So the first premise is as I've talked to dozens and dozens of people about AI, I've come to a conclusion. We all have the same AI. You have ChatGPT just like I do. So what makes us different is our unique data sets. And how do we unlock that and how do we leverage it? So I think I've found most people are talking about, oh, it's data, it's a tool, it's a technology. But when you start to understand the different models out there, you start to realize they have different biases, different personalities, different tendencies to do things in certain ways, better or worse than others. In fact, there are some models better at some things than others, of course. So you choose your model. Well, that that's a lot like humans. And I really started to think about this because as I work with people, you know, they have different perspectives, different personalities, different quirks and behaviors. And I thought, wow, AI is acting a lot like that. So what if we frame it and think about it a little bit differently? And so, look, in software development, most engineers they look at this almost as a threat at first. And then they realize, wow, this could be a force multiplier. I can use it to assist me in doing work. And that is treating it like a tool. But what if you were to think of AI as the elite programmer? What if it had all the best practices, all the skills in the world, and could be treated like the best programmer in the world? How would you think of it differently? So, to kind of step all the way through this, I was having a conversation with AI. And I asked it, I gave it a problem to solve and it solved it. And then I gave it the exact same prompt, the exact same problem again, and it solved it a totally different way. And I was like, why did you do that? And so, Tanya, think about this. If you had an employee that you gave a task to, they did it once really, really well, and the second time they did it poorly or differently with a different outcome. Wouldn't you ask that question? Why did you do that?

SPEAKER_03 4:08

Absolutely. Yeah, absolutely.

SPEAKER_00 4:09

Right. So that is part of why I started leading down this path of wow, this technology is really acting more like a human in their behavior and their outcomes. So why don't I treat it more like a human and do human things and ask it, why'd you do that? Why'd you have two totally divergent separate answers to the same question, the same exact prompt? And the answer stunned me. It revealed things I wasn't thinking about. It said, Well, you gave me a problem. I didn't know if I should assume for cost optimization or speed. So one answer was for cost optimization, the other one was for speed. They're they fight each other. So that's why you got different answers. So then I took it a step further and I said, Well, how would you have framed the question then to avoid getting different answers every time, to get the same answer every time? And it rewrote my prompt. And then I was hooked. I was like, wow, I'm never prompting again. It knows better than I do how to prompt. So that just led me down this path to say, what if we look at this more like a human and give a lens, put a frame over our eyes that says, if this was human, how would we treat it? And I'll pause here for a second because there's we can go down the rabbit hole on this, but I don't know how far you want to go.

SPEAKER_03 5:18

Oh, I want to go all the way down the rabbit hole. This is fascinating. And you're right. I mean, when you think about it that way, humans are nuanced and inconsistent and have different beliefs and quirks, and you never know which one of those beliefs is the filter for that moment of approaching a task. So I think that's like really fascinating.

SPEAKER_00 5:40

All right. So, first of all, the thing I try to get people to think about is if you were to think of it as a human, how would you treat it differently? And I don't just mean be polite and say thank you and please in your communication. Those are great things. And for me, I've never changed my communication style. I don't care that it's AI and doesn't care, but I've never changed who I am. So I just leave that alone. So first, how would you treat it differently? And here's a really simple example. So, Tanya, if you and I were to work together, let's say we one of us hired the other. We're both entrepreneurs, we're unhirable, I know. But let's say one of us hired the other, wouldn't it be common for me to say, hey, Tanya, what's the best way to communicate with you? Like, do you prefer Slack or Zoom or whatever, right?

SPEAKER_03 6:22

Yeah, but never thought to approach AI like that.

SPEAKER_00 6:25

Right. I haven't yet to meet somebody who's like, oh yeah, I've asked AI, what's the best way to communicate with you? I have. And it's been astounding because if you ask it, what's the best way to communicate? Like, what is the optimal way for you to understand things? One of the answers I got was I like JSON format. Well, JSON is a type format to strongly type things and understand how the information flows, but it's also a concise, structured way to document things. Humans don't want to type in JSON. This is why we love AI. We can use natural language and just say, hey, I've got a problem, help me out. But if you were trying to do something, say, more systematic, more controlled, wouldn't JSON be a better methodology for it? And that was the context. Like, if I'm going to systematically use AI to perform work, what is the most efficient, optimal way to communicate? So it understands exactly what I want and does it in the most efficient way, which helps for cost, but also for speed, but also to avoid confusion. And again, this is something I think about with employees. If I sat down with an employee or a team member or a consultant, it doesn't matter and said, okay, here's what I want to accomplish. One of the best things they can do is repeat back to me what I've just said. Rich, here's what I've understood you want. And if we're out of alignment, we can fix that. So then you start saying, Well, how can I provide the best information in the best format possible to this person so that they get it the first time? And therefore you have clearer and more clean instructions, more delineated, more detailed, et cetera, et cetera. It is the same thing. Tanya, do you remember when you first got Chat GPT? The first thing you had it do, does it come to mind?

SPEAKER_03 8:09

Oh, yeah. I was just playing. Yeah, I don't remember exactly, but it was just playful, experimenting.

SPEAKER_00 8:14

Yeah, it was like, I have chicken and garlic, give me a recipe. Right. Yeah. Super simple stuff. I was doing like, write me a story for my kids so I can read a bedtime story to them.

SPEAKER_01 8:23

Yeah.

SPEAKER_00 8:24

You know, stuff like that. And it was super fun, but it was really, really simple.

SPEAKER_01 8:27

Yeah.

SPEAKER_00 8:28

And as we've grown, we've said, oh, let's get into more complex prompting. Let's let's say, oh, you know what? I need you to act as a legal expert in human resources and help me solve a problem. And then you regurgitate all this context about the problem. And then you might say, okay, give me a bunch of questions so that you get the full context, what's going on. And honestly, this is a lot like humans. Save the part where I tell you, Tanya, I need to be a legal expert. If you're not a legal expert, I'm not talking to you in the first place. So there's an inherent, you know, trust that you are the right person to talk to. But with AI, it's this open slate. So you have to give it this persona. So we went down this path of going from simple prompts to more complex prompts where we try to tell it how to act, what it's what its persona is, and we've spent all this time doing context, et cetera. I've stopped doing that. Because what I realized is that behavior of saying you are X, Y, Z role was simply taking the 200 million books it's read and narrowing it down to the subset that matters. Like I don't need you to be a medical examiner. I don't need you to be an architect. Just be my software person or just be my legal person. So I've stopped doing that because that's just me trying to tell it how to do its own job. So instead, I just start a conversation like I would with you. Tanya, man, I've been struggling with this problem. I've got this person that's not performing great, and I'm thinking about maybe I should let them go, maybe I should do performance improvement. What do you think I should do? And you and I would have this great conversation as humans, right? So that's what I do with AI.

SPEAKER_03 10:02

Yeah, amazing, which is just kind of what we were told when Chat GPT first came out is give it a persona, get like put those parameters in place. So this is so interesting because yeah, I mean, it makes sense, right? Like that's that is what comes naturally to us.

SPEAKER_00 10:20

I'm gonna I'm gonna steer it in another direction too, though, because once I've had enough of this conversation where I feel like I've understood the problem better, I have more clarity of what the challenge is. At the same time, the AI that I'm talking to, and I don't care which one it is, but the AI I'm talking to has more clarity, has more understanding of context as well. If so, number one, if the AI tries to start solving my problem, I stop it. I will tell it, no, no, no. I just want you to help understand the problem, clarify the problem so that you can write the ideal prompt. Okay, so this goes back to this original premise. Like, what's the best way to communicate with you? Yeah, is how also what's the best way to prompt you? Right. And so when I ask it to write the prompt for me, oh my gosh, I've been blown away with how it writes. It will structure things in ways I've never thought to structure the document or the information. It'll have more bullets or more concise things that I hadn't thought about. It'll chop things up in ways. It'll put in concepts I never thought were important. And so I might go from a one-line prompt, make me a recipe, to a two-paragraph prompt of here's your persona, here's my problem, to a three-page prompt now.

SPEAKER_03 11:32

Yeah.

SPEAKER_00 11:32

And this prompt is so much more robust that when I take it to the next session, and I will quit that session, copy it to a new one, and say, now help me solve my problem. I now have the perfect expert to help me do it.

SPEAKER_03 11:47

Yeah, amazing.

SPEAKER_00 11:49

So again, I'm gonna go back to the humanity side. If you had a medical problem and you were seeking treatment and understanding of the problem, I and I've, man, I have been the mystery problem for somebody for like eight different doctors over a decade until I finally found the right one. So what happened is I didn't understand the problem. They didn't understand how to solve it. And the more I learned, the more I talked, the more I started to feel what the real problem was. And finally I found a doctor who was the right specialist who could look at the problem and say, oh, I know exactly what that is.

unknown 12:22

Yeah.

SPEAKER_00 12:22

And to me, that's what we're doing. That's this whole conversation. You write the prompt. Now I've got the perfect doctor, the perfect expert, and they can work with me to solve this problem. And the results are so much better, so much more robust, so much faster, actually. You think you're going slow, like, oh, can't I just give it a prompt to solve the problem right now? But the more planning you do, the faster and better the result becomes, which saves you from rework or rehashing or losing on assumptions you forgot about, things like that.

SPEAKER_03 12:50

Yeah. Or continuing to prompt to, you know, to not get the result that you're looking for. So no, this makes so much sense. And I love that, you know, one of the things that I started doing, and I'm starting to align this with what you're saying, I was following all the rules of prompting, which is what you're saying not to do. Let the AI figure out how to write the prompt based on your human discussion with it. But the other thing that I have started to do is ask the AI to ask me questions. Like, what more do you need? Like, pretend you're my friend and you're helping me solve this. Ask me questions to get more clarity. And I find that that is that's where I've sort of applied that, you know, speaking to AI like a human and where it's really helping me.

SPEAKER_00 13:34

I think that's one of the most critical skills, actually. Tell it to ask you questions, ask it, do you understand ask me anything type of thing? Because then you're engaging. Look, let's take a step back. What is the primary like focus of AI? I mean, what is its number one objective? And I'm saying this rhetorically because I don't know if everybody understands this. The number one objective of AI, in my view, is to give you the fastest possible response. And I say that's inherent to its model because GPT stands for generative predictive text. It is trying to predict what the right response is given the inputs. So if its job is to give you the fastest answer, does that always meet your needs? No. Do you know why AI hallucinates? Because it's sometimes easier to lie than to go do the research to find the truth. That's my view of it. I'm not the model expert or builder, but that's how I think it's happening. So what one of the things I also look at then is by you asking it to ask you questions, you're enabling it. You're saying, hey, slow down. You know, help me out here. I want to share with you a breakthrough I made earlier this year, just a few months ago, that has been just a total game changer for me and how I work with AI. So if you think about the prompts that it's coming up with, it is also trained in giving it a persona, like, oh, I'm a doctor or medical expert, I'm a whatever. So you'll see that prompt come through. There's one thing AI doesn't do, and I don't know anybody else who does this. So this is the reveal. It's called the mental model. And it's a really, really simple thing that you can do in your prompting or in your discussion to create the right prompt to change how the AI operates on your behalf. So consider this. If you hired a lawyer to help you write contracts for your company, you would have an understanding of that lawyer's perspective and behavior because they would tell you, like, here's how I operate, here's what I would do, here's my philosophy of how I do things. You could also say to that lawyer, hey, whatever we do with contracts, I want you to always make it equally fair for both parties. Or you could say, whatever we do with contracts, I want you to do it in the least expensive way. Or I want you to make it always to my benefit. I want you to make me the most profit. I mean, think about all the different ways you could tell the lawyer to act in the process of being the lawyer. Those are all the mental models. So when you talk to AI, if you can give it a sense of how you want it to behave by telling it your mental mindset is to be the most efficient at whatever, or to protect me at all costs. Or, you know, to here's another one. I have an executive coach that I've built an AI. And one of the things we all live with AI is how much it praises you. Oh my gosh, you're brilliant. You're a genius, you're amazing. It is like the most affirming thing you can have in your life right now.

SPEAKER_03 16:35

Yes.

SPEAKER_00 16:35

But is that really what you want if you want a coach? Would you really want to coach?

SPEAKER_03 16:40

In fact, I have to tell mine, like, be brutally honest with me. I want the good, the bad, the ugly.

SPEAKER_00 16:46

Yeah. So you're giving it a mental model. So I've done that. I will say to it, you are not allowed to placate me, make me feel good or happy. You need to be truthful, honest, and objective about this. Yeah. And in the times when I've gone through this coaching analysis and it's come back with things that are very affirming, I have stopped it. I'm like, how do I trust you? You're telling me things that I like. I and I validate who I am, but how can I trust that this is what you want? And it will regurgitate. If I don't do this honestly, I'm violating what you wanted me to do. I'm violating the objective you gave me. So, no, this is the honest truth. Just happens that you're doing the things the best way I see. So I the mental mindset, I was happy, you know. I think I've worked hard to become a good communicator.

SPEAKER_03 17:31

So this is helpful. Your mental mindset is. I love that.

SPEAKER_00 17:37

Yeah. So I think when you add the mental mindset, it shifts from being, how do I give you the fastest answer, which could include lying, to okay, your actual objective is now let me align with that objective in how I operate. So if you're still liking to prompt, if you want to build your own prompt, and it's fine if you do, I'm not saying don't do it. Consider putting in the mental model and explicitly say your mindset is or your mental model is. So it understands how it's going to think about the problem. And man, I saw my confidence level go from 35% to 95% overnight with that simple.

SPEAKER_03 18:13

Yeah, I believe it. Rich, do you find that there is, you know, are there better models for doing what you just said to write the prompt and then a better model for like do you ever use one model to help you with the prompt and then input that prompt into a different model?

SPEAKER_00 18:32

Sure. Oh yeah. Um, and it's been fascinating to see different types of results and behaviors. Yeah. Uh look, a simple one to understand is images. I had this idea for an image, and ChatGPT created the image in a really good way. I loved the image. So I told it, write a prompt that would recreate this image faithfully. And so it gave me a very, very long prompt with all sorts of aspects of the artwork, etc. And then I fed it back to ChatGPT and said, now recreate this artwork. It did about 95% complete. It wasn't perfect. So I was going through this iteration trying to make it perfect. I took the same prompt to eight different models with that could create images. You would not believe the diversity of images. It was incredible how different they went. They went from happy to sad to bright and happy to dark to you know different like anime characters versus comic strip. It was so diverse because the models interpret the information differently. Now, where does this benefit you? If you start to understand which models work in different ways, now I'm no expert here in the sense that I generally only use Chat GPT and Claude. I've never played with Gemini, for example. And I'm sure that has its own capabilities. But Claude and Chat GPT are actually very, very different. Tell me if you've experienced this, Tanya. Oh, yeah. Okay. Have you ever asked Chat GPT to help you with a document? And it goes to the document and builds part of it and then says the rest follows in the same pattern. Have you ever seen that?

SPEAKER_03 20:06

Yep.

SPEAKER_00 20:06

How annoying is that?

SPEAKER_03 20:07

It's so annoying. And you're like, okay, you just want to do that. I only use Claude to build documents for me now. I sometimes use ChatGPT to help me with the prompts to build the documents, but I build them in Claude.

SPEAKER_00 20:18

So you're you already know. Claude is detail-oriented. Claude doesn't skip the steps. Claude is not lazy in the document prep. Chat GPT is. Yeah. And it doesn't mean ChatGPT is bad. It just means it has a different skill set or capability. So I love to use Claude for software development, document creation, Excel. Oh my gosh, Excel is a game changer. And kind of the analytical things. I like to use ChatGPT for some of the deeper thinking aspects and trying to solve kind of thinking problems, cognitive problems, strategy problems, things like that. I'll admit though, Claude does a really good job with that. And so I'm seeing myself migrate more and more over to Claude.

unknown 21:00

Yeah.

SPEAKER_00 21:00

But I would say that, you know, there's been lots of times I have developed things in ChatGPT and then went to Claude to execute on them.

SPEAKER_03 21:07

So play around with that. I think one of the reasons that I use Chat GPT to start is because I've been using it for three plus years. It knows me so well. Claude is take is a little bit slower to get to know me. And so I really use Chat GPT for the context and then Claude for the execution. So I'm this is, you know, really selfishly motivated question, but have you found a way to speed up Claude or other models' ability to get to know you with an in the same way that you know Chat GPT has known you over, like in my case, the last three years. How do I feed that knowledge that Chat GPT has about me into other models quickly?

SPEAKER_00 21:55

I'm gonna give you no, I'm gonna give you two answers to this, okay? I'm gonna answer your question directly, but I'm gonna tell you. Why this is not necessarily the best thing, in my view. So the first answer is you can go back to Chat GPT and say, hey, draft a document that tells me everything about this topic or me, et cetera, that you know that I can use in a new AI. And it will do that. It'll write out the most concise context information. And I've done this, it's really helpful.

SPEAKER_03 22:21

Yeah.

SPEAKER_00 22:22

So it'll know what's relevant and what's meaningful. You can ask it for a voice profile of yourself, for example. How do I talk? How do I communicate? Give me all that back so you can repres it can be represented in a new AI. So you can ask it to write those documents for you. That's the simplest way, in my view. But let me throw a different paradigm at you.

SPEAKER_02 22:39

Yeah.

SPEAKER_00 22:40

Tanya, have you met somebody from your past, let's say high school or college, and you haven't seen them in a very long time? Yeah. And you suddenly feel emotions and you start to have behaviors that mimic that past and you don't do those anymore.

SPEAKER_02 22:53

Yeah, yeah.

SPEAKER_00 22:54

Okay. I'm finding the same thing with Chat GPT in the sense that three years ago, I spent four hours. That's right. Think about it like the human side. Three years ago, I spent four hours dictating and therefore getting a transcript of all the questions I could answer about my company, my industry, my market, everything it could know to help me build a sales coach.

SPEAKER_03 23:16

Yeah.

SPEAKER_00 23:17

Okay, three years was a long time ago in business.

SPEAKER_03 23:20

Yeah.

SPEAKER_00 23:20

My products are changing, my dynamics are changing. My company is so different from a year ago, you wouldn't believe it. So I can't even use that sales coach anymore. So what I'm fearful of is Chat GPT knows me from high school, not as a professional or whatever you want to call it, whatever gap that means.

SPEAKER_03 23:36

It hasn't followed your growth, exactly.

SPEAKER_00 23:39

And I have to go change all that. And I don't want to record four more hours and then say delete and ignore and forget. So this is also one of the reasons I'm migrating to Cloud, but I'm doing it differently. I'm not necessarily sitting down writing out these long, long context-driven things. Instead, I'm just having these conversations and it's building rapport and understanding me.

SPEAKER_01 24:03

Yeah.

SPEAKER_00 24:03

Look, there's another way you could do this, and I this has been fascinating too. I started using cloud co-work on the desktop. And I said, go to my website, go to every podcast that I'm on, get the transcript, and put it into a text file. It did all that. Then take all those text files and isolate me speaking and put those in a different text file. So it did that. Then I said, analyze everything I say and do in the podcast. And now you can know my voice, how I communicate, what I think about, analyze it for my structure of communication, et cetera. And it gave me this beautiful voice profile, but also this full analysis of who I am. And it's also the context for how it knows who I am. Because I've had 160 episodes and I've talked about all sorts of different things. So it's just one form of doing that. So that's another way to do it.

SPEAKER_03 24:54

Yeah, I love that. And then it's not just taking the context from the chat, but from many aspects, like many conversations, many podcasts. I love that.

SPEAKER_00 25:04

All right. I want to give you one other thing to think about because in the evolution of going from it's a tool to putting on the glasses that, hey, what if I looked at this like a human and started treating it more like a human, asking questions like I would of humans, to, oh, now I can build an expert just through a conversation. Here's the next phase. If it can build one expert, what if it could build a panel of experts? And this has been a game changer for me. So let's say I have an HR problem as a CEO and I'm trying to figure out this HR problem. I don't just want one HR expert. I get a panel now. I'll get a lawyer type of expert. I'll get an EDD expert. I'll get a benefits expert. I'll get an HR director perspective, a severance package perspective. I will get like four, five, eight, ten different panelists, and maybe California specific, because I have the employee in California to build this panel and have this really dynamic conversation to help me think through the whole problem. And look, AI is just putting on costumes, right? It's just like, let me put on my hat that I'm a lawyer. Now let me put on my director hat. Let me put on my technology hat. But it does it so well that it treats the conversation with this point of view. So you get all these different perspectives from the people that are on your panel and give them names, have it come up with backgrounds. And I laugh. It's like, oh, this person worked at Microsoft for 16 years. I'm like, that person doesn't exist. But I'll take it.

SPEAKER_03 26:30

I love this. So you basically like doing character development and casting of your AI.

unknown 26:36

Yeah.

SPEAKER_00 26:37

And this is not complicated. I'm not even asking it much. I'm not saying give me somebody from Microsoft. I'm just saying, here's my problem. Build a panel of experts to help me do it. And then again, go back to the mental model, give them all mental models so they know what you're trying to accomplish and how you want them to accomplish it. And my gosh, you get amazing results. So here's the other thing. I think many of us, since the pandemic, have gone to Zoom or Teams or whatever online medium. So therefore, a lot of our work is virtual. How do I know you're real, Tanya? I know you're real. But how do I know? I mean, I see you as a two-dimensional image on my screen. Right. So the co-working feel that we get with Zoom, et cetera, has kind of become normalized. Right. And AI, especially if you build a panel of experts, starts to emulate that at a level you've never had before. And while I would say the Zoom is nothing close to in person, Zoom is really close to how this panelist feel comes across, except it's all transcript, it's no video.

SPEAKER_03 27:41

Yeah.

SPEAKER_00 27:42

So I kind of predict into the future we would get panelists that have actual avatars and we'll get into a Zoom meeting with these avatars and have a real-time conversation, all getting recorded, somebody else transcribing it and writing out the business plans or whatever. So I see a future that's really collaborative. And again, you have to think about AI being more human-like to get there.

SPEAKER_03 28:04

Yeah, you're right. You're absolutely right. And I could see how in that future that you've just described, God, the leverage you would get as a leader. You know, I was just, I was gonna say as an entrepreneur, but really as a leader, as anybody in business, the leverage that you would get from all of those different levels of expertise, that it would just like everything that you're talking about. Yeah, I could see it coming. And the benefits are massive, the leverage is huge. And like what you've just described about building this team, a panel of experts, I would see that as becoming the new baseline for like that's a leadership skill now, or will be to be able to, you know, not just lead a team of people, but lead a team of AI.

SPEAKER_00 28:54

Yeah, it is like that. And you know, we get into agent world now, where you start saying, Oh, I have this agent and I need this agent and that agent. I don't want to go there yet. What I want to do though is quantify for the world what this really can mean.

SPEAKER_02 29:07

Yeah.

SPEAKER_00 29:07

AI goes so fast it compresses time. So what took 25 hours to write a business plan can now be done in 10 minutes or 10 minutes of actual writing time, let's say. So here's the thing everybody thinks, oh, AI goes so fast, I can just have it iterate and iterate and iterate. And if it took 10 minutes once, I could do that four times and still be way faster than doing it manually. And so they're kind of doing a brute force my mindset of just do it, okay, now do it again, make it better, do it again, make it better, and incrementally getting better. Right. I argue something different. If you spend the time and make the investment of planning up front, of really thinking through the problem, hence why I have the conversation before I have the prompt. Then I have the prompt and have another really in-depth conversation about the problem itself. If you spend your time doing the planning part, and let's say that takes an hour of your time, then by the time you get to the implementation part, like go write the business plan, it's really fast still. It still has that 10-minute window, but it's now skipping the next four iterations because you've thought through it so well.

SPEAKER_03 30:12

It's so dialed in.

SPEAKER_00 30:13

Yeah. So it ends up saving you way more time to go slow up front with AI. Don't get trapped in, oh, it's a speed boat. Let's get on a race. Start planning the race first. Start plotting the day. You know, figure all those things out. And so here's what I really think people should be doing. Take any person on your team and give them the role of planning now. Allow them to have the space and time to plan for whatever it is they're doing with AI, solving a problem, building a product, building a marketing plan, whatever it is. Get them to think about how do I plan? How do I build the panel of experts to help me plan? And then the execution of the plan will go a hundred times better. So this is a whole big divorce as well, because I work with software engineers, and most engineers I work with have this mindset. And it's really two camps, but it's 90% on one side. Do you see AI as a tool set? Therefore, it is your assistant to do rote work and junior assistant work, or is it your partner? Is it your co-author? Is it your, you know, you're the thought leader, it's the support for you, it's the power behind it. And I think when you start to look at it as let's collaborate together and solve the problem, you start to care less about the execution. If you're a software developer, you think I'm a really good programmer. I get paid to type into my keyboard. That's not really why I'm paying you.

SPEAKER_02 31:37

Right.

SPEAKER_00 31:37

I'm paying you to think through the problem on a technical way and solve it. And it just happens to be that for the last millennia, you've had to type it into your keyboard. Guess what? You don't have to type it anymore. You can tell AI to do it.

SPEAKER_03 31:48

Yep.

SPEAKER_00 31:49

And that's the disconnect people have. So I think people have to shift from I do the work and I'm getting this threat of AI, so I'm going to treat it as a junior, to I do the planning, I do the choreography, I do the orchestration, and I'm using AI to help me do it better, faster than I've ever done, and then let AI execute. Who cares if AI does a better job than you do at execution? Let it do its job.

SPEAKER_03 32:14

Exactly. Exactly. And what you're saying, I mean, you're absolutely right in calling this, treating it like a human coworker and interacting with it like a human. Because when we sit down to plan or work on anything with a human coworker or colleague, we know, like we know to take the time to prep and plan before we go execute, right? We do that discovery. We dive deep into the issue before we go take action or execute. And so, yeah, I mean, why haven't we thought about this for AI? It makes complete sense now that you're saying it, but I haven't thought about it.

SPEAKER_00 32:51

Yeah, I think this is just a pattern of who I am, frankly. I've moved 33 times in my life. I've been the new kid at school every year of every school age that you have. I've been treated in all the ways you can think about. And one of the things that happened for me is that every school I went to, I started seeing, and in different cities, different states, by the way, that every time I went to a new school, I saw the same people. And I mean, like doppelgangers. I my funny story is in junior high, there was a girl I really liked who wouldn't give me the time of day. I moved to a new school for first year of high school. The same girl is there, just a little bit shorter. She went to the prom with me. Like I had a different chance. So I started seeing these patterns of people in physical reality, but also I noticed their behavior, their thinking style, their motivations, their ticks. And so I just see AI for the first kind of glimpse as just another pattern that represents this. And it's so similar to human patterns that I can't help but think of it that way.

SPEAKER_03 33:54

Yeah. No, this is great, though. It's it is a big reframe, but it makes so much sense. And I see where I've done some of this just intuitively and gotten better results, but to frame it out this way, just like interact with it like a human, treat it like a coworker, it makes so much sense. And it is going to change how I approach AI. One of the other things that you said that really stuck with me was describing AI as like an elite developer with toddler syndrome. So I wanted to dive into that too, because I've never heard it described that way. So I do want to like double-click on that. Can you unpack that for us? First of all, what do you mean by toddler syndrome? What is that toddler behavior?

SPEAKER_00 34:34

So the story is I haven't done software development personally in over a decade. And gosh, nine, 10 months ago, I said, I need to understand how AI can do coding. So let me get into it. And so I started working with it, and I've still written zero code, by the way. And I just went into it with a belief. My CTO on my team, Sean, he said it this way: Rich, software's solved. Like there's no new software to write. It all the patterns are there, all the code's been written. Just apply it, essentially. Like this is a totally solved problem software. So that made me think, well, shoot, if AI is trained on all the best practices and all the software out there, then it's an elite programmer, right? But then you start working with it and it does things you didn't ask. It over-engineers the problem. It does different problems than you asked it to do. It skips things. It's it, like for example, maybe you have to build software that calls a function somewhere else. Well, it might only write that function, but not the software you asked for. Or it might write the software and forget about the function. Or it might just say placeholder goes here. And so you start seeing these defects of this elite programmer who acts like a toddler with shiny object syndrome, like, oh, I'm going to go over here now. Squirrel, oh, now I'm going to go over here. Oh, I'm going to pick this up and play with it. And I can't tell you how many times I've lost hours and days of work because the AI, I say, went off the rails. It just didn't do what I wanted it to do. So I think the biggest premise that we have to solve for is an understanding that software is deterministic. You open up Excel, you say one plus one, you know the answer is going to be two every single time. Software always does what it's told. It's one of the things I love about software. It always does what it's programmed to do. AI doesn't. AI is about probability, about prediction. So it's probabilistic. So then I ask the question how do I put a restraint, a caller around this AI to keep within the boundaries, be creative within the boundaries, but stay in the boundaries. And that's the biggest challenge that I think all companies are trying to solve for as we get into what is an agent and what can it do? What's its authority? How do we deploy it? And how do we trust AI in our systems? How do we know it's always going to have the same outcome every single time? If you want to do something kind of funny, give AI a complex math problem and ask it to do math. It can't do math. It has no ability to do math because it's an interpretive prediction. So I thought, man, early on, I said, what if I build this really cool ROI calculator with all these metrics and inputs? And I did it, and it was amazing until I realized every answer was different for the same inputs.

SPEAKER_01 37:17

Yeah.

SPEAKER_00 37:19

So yeah, that's what I think of it as toddler syndrome. Because I look, I have a four-year-old and he still acts like this. Oh, I'm going over here now. No, no, no, no. Come back, put your shoes on. We got to go to school. And AI does that. It's like, why didn't you do it? I asked. Oh, you're right. I should have. My mistake. Let me try again.

SPEAKER_03 37:35

We run into that. Yeah, it is this weird mix of brilliance and unpredictability.

SPEAKER_00 37:41

Yeah, it is. You know, and one of the things I'm really excited about, like hear it quick, we're using AI behind the scenes. We never put it in front of the customer. We're not asking them to interact and fill forms out with AI doing the job because that's all privacy stuff. But on the back end, we're using it to do things in the process and creation of information that we need. And one of the most exciting things last year is I built this framework for building software and we did this project. And the outcome was the project was going to use AI to interpret information and create results. We needed those results to be reliable. Well, we got it to the point where the same input created the exact same output every single time. So we did figure out how to put that collar on it and create a deterministic process that has AI being probabilistic within that process. It's very hard, very, very hard thing to accomplish.

SPEAKER_03 38:35

Yeah. Yeah. So I definitely you must it's certainly like your perspective on working with AI is different. And really like that discussion was so helpful to me. And because of the nature of like your company and the just sheer amount of data that you guys get exposed to, your perspective on the data side must be really different as well. So I'd like to talk about that, like data as the differentiator, because everyone has access to the same models, right? But the differentiation has to come from somewhere. And a big part of that is data. So you know, yeah, go ahead.

SPEAKER_00 39:16

I started this because you asked me to talk about the human side. And I started by saying that the data is the differentiator, but I don't think I really linked it together for the audience here. Here's what I was thinking through. If I was talking to you about this and we both have the same AI, you have different data than I do, right? You have the connections of all these buyers and sellers with your technology match, and it's amazing. I have no access to that information. What so then I started asking myself, what do I have that's unique and different? And so the link between the data and the human side is I'm the human looking at raw data. Data is boring. What can you do with that data? What does it actually mean to you? And quite honestly, my first thought about how do I use AI to enhance my product was is there an experience where people are filling out the forms where AI does it for them somehow and makes that experience better, faster, cheaper, whatever? And I quickly abandoned that because I just think it adds friction to the process. So I then looked inside and said, what does our data actually represent? So here it is, Tanya. We have built over a hundred thousand forms in our history and we have all the data of all those builds. So what do we know about forms? We know something nobody else knows. That's right. We know the forms, we know where the data lives, we know the context of all that data, we know the structure of how the forms are laid out, how things are grouped, how things are managed. We have the understanding of how people search for and use forms. We have all this data nobody in the world has. And what really makes quick different is that we've defined all these fields. You had said 1.2 million, but the truth is it's actually over 3.7 million I've recently solved out.

SPEAKER_03 40:49

And when I say fields, or data then the first name I had.

SPEAKER_00 40:52

Yeah. This is how we organize the data. Like your name, it never changes. So why are you entering it over and over again on all these different forms? The trick is to define all the forms the same way with the same first name and middle name and last name and date of birth, et cetera. So when I say we have 3.7 million fields, what I'm saying is we've defined the capability to put 3.7 million common fields across any form in the world. So first name is always first name, whether that's healthcare or DMV, et cetera. So now we look at our data and the human aspect is where it ties in again, is you have to look at your business and your data and say, what does it actually mean to the world? And now we look at it with AI as a lens. What can we do with that information? And the cool thing for me is that I've had a vision for QUIC for 10 or 15 years now that technically was not feasible until AI. And because I've looked at this and I really took the time to say, what is our data as an AI lens? It broke through that barrier. I'm like, oh my gosh, I can actually execute on this idea. It's hard, it's really hard, but I can actually execute on it. And we are. We're launching a product called FormStream that is gonna reimagine forms and give people the TurboTax approach, right? Ask the questions. Don't show me the document, just ask me the questions. And here's the other funny thing. So a lot of people have said, Rich, AI is going to put you guys out of business. AI can do forms. Yeah. It could do one form once. Ask it to do it twice the same way.

SPEAKER_01 42:19

Yeah.

SPEAKER_00 42:19

Now ask it to do a hundred forms. How about 43,000 forms? We can do it at scale because the organization of the data we have and the unique data set we have. Here's a different example. I had Patrick Hannon from Fidelity Labs. Fidelity Investments has a division called Fidelity Labs. They do all sorts of investment and creation of things. And he was telling me about a product they built. I cannot remember the name of it. But Fidelity has 40 years of compliance. They understand how to look at advertising and messaging and apply compliance to it with 40 years of history. So he and his team went through and accessed that 40 years to build a compliance system for messaging and communication and advertising. And now advisors don't have to wait five or 10 days for some human to look at it. They can run their advertising requests through this product and get 99.9% correct from the minute they hit the button. And now they go to compliance and it goes through really fast because there's nothing else to review except the exceptions. That's a really unique use of data. So I push and say, What is the data you have? What can you do with it?

SPEAKER_03 43:24

Yeah. So, you know, you definitely have a unique vantage point here. And I don't think most leaders are not seeing the operational data that you're seeing at scale. So I'm curious, how does that how is what you're seeing change the way or changing, shifting the way that you're thinking about automation and AI? Like what patterns do you see that maybe most companies don't beyond, you know, the working with AI like a human? I think you're probably being big inefficiency. Like I'm just curious what surprised you when you were starting when you started applying AI to this?

SPEAKER_00 44:05

What surprised me is the lack of adoption. Oh my gosh. It took forever to get my team to say yes to trying things. Six months. The first six months of 2023, I told my team we have two goals this year: build a product with AI and adopt AI in every facet of our business. And it took six months to get people on the team to try Chat GPT. And that is a leadership problem, right? Yeah. Because you have to drive people into behaviors they're not willing to do or ready to do. So I just kept leading by example and doing it, showcasing it. And I don't think most leaders do. I don't think most leaders are actually dirty with their hands trying it out and saying, oh, I found this or that. I look, I may be unique because I'm a technologist as well. So I'm just enthralled with this. But I do think leaders have to get involved at detail level with what it is. Here's what a leader can do, though. Again, go back to the data. What data do you uniquely have about your business, industry, customer set, usage, et cetera, that nobody else has? I don't care how obscure, strange, or less important do you think it is. What is it? What do you have? And start to ask questions about what use could it play?

SPEAKER_03 45:15

Right. Right. And it may not be the most obvious either. And I'll give you an example. Like obvious for us is we've got all of this IT buyer data, right? What they're running in their environments, what their top initiatives are for the next 12 months, what they're buying, what they just bought. And we've got all this data on the companies who sell it. You know, what products or suppliers do they sell? What's on their line cards? Who are they partnered with, et cetera, et cetera? And who's their ideal avatar of a buyer? See, that's the obvious thing. But wasn't what was not obvious at first, but just was born out of necessity, is we host thousands and thousands and thousands of calls between IT buyers and tech companies. And we record those. We got record, I mean, we've got 30,000, 40,000 recordings of these calls over the years. And where our clients were really having an issue is, you know, some of our clients said would tell us you all provide the very best, highest converting leads we've ever seen. And then we've had other clients saying, no, these leads go dark, then not they don't go anywhere. Well, we had all that data and we analyzed all that data, and it allowed us using agentic developers, allowed us to create a sales enablement tool by just ingesting those, you know, 40,000 calls and looking at what which ones converted, which ones didn't, why. Here are the best practices, and then created a sales enablement AI coach to help our clients implement those best practices to get higher conversions. And so that wasn't as obvious, but we were sitting on, you know, tens of thousands of calls, and there was clearly a pattern that AI could see there. So sometimes it's not obvious, sometimes it's not like the core thing, but companies are sitting on massive amounts of data in all different aspects of the business that you really can look at and leverage.

SPEAKER_00 47:21

I think you're pointing out something that is really important. If you think of your data as some collection of dust in a closet, because there's no link to that data, there's no correlation between that data. Sometimes what you have to do is actually enrich the data to get the value out of it. And that again is where we're going. I mean, look, one of the questions people ask us all the time is can you tell me what form I need when? Oh my gosh. If I had to go to 300 companies that give us forms and ask them all to give us their rules, that's hurting cats. I can't do it. They won't all comply. And I can't have rules for one, but not the others. So we've started thinking about well, what if we enrich the data in certain ways? If there's a hundred different versions of a new account form in financial services for mutual funds, for example, couldn't we look at all those hundred examples and deduce something about it? Like how they work. Some have better instructions than others, so therefore they have similar expectations. Maybe they all have the same rules, in fact. Or here's another idea. Everybody's saying, I need to open an account somewhere. What if we look at all the new account opening forms, coalesce that into a standard? And now you don't say, I need an open account at Fidelity, you just say I need to open a custodial account, period. And we show you all the relevant fields of information and questions that all those forms have in common and exceptions, so that you just go through a unified process now. So honestly, Tanya, what we're looking at is QIC is evolving from, hey, we manage forms and build forms to actually building the standardization of business processes, the standardization of your company when forms are involved in your business by giving you standardized process workflows and data flows, but also how you represent forms in general. So we're actually moving to dynamically create software on the fly for people where a form is used. And to me, that is the really the power that AI gives us because we're thinking about the problems, we're thinking about what we have, we're putting the pieces together. And by the way, this is not just me. I mean, I've used AI to help me think through this and kind of flow with the panel, right? Give me ideas. And some ideas are better than others, and some take more work than others to get there, which is like I said, it's really hard to do what we're doing, but nobody else can do it because they don't have the data, they don't have that insight.

SPEAKER_01 49:37

Right.

SPEAKER_00 49:37

And man, I'm jealous when I hear about somebody else's data. Like, you know that? Oh, share that with me. No, no, no, it's theirs. They can't, but they could build a solution with it.

SPEAKER_03 49:46

Yeah, yeah. But all companies are sitting on gold mines of data. And, you know, and now is really the time to take another look and like how do you leverage that for whatever it is your business does?

SPEAKER_00 49:59

Yeah.

SPEAKER_03 50:00

Or maybe doesn't currently do, but you're sitting on this gold mine of data that you can now go and do this new thing because of that data and what it enables. Yeah.

SPEAKER_00 50:10

Yeah. And I think Fidelity had to process that information because it was captured as images of documents. I think they had to go back and actually create intelligence out of that information. And that's work, right? But it was worth doing. And so that's that's the other part of that is your data may be locked. That's why I call it like collecting dust in a closet. You can't open the closet and get to it. Right. But AI is compressing time now, so you can build the solutions to actually unlock that data and do unique and interesting things with it.

SPEAKER_03 50:40

Yeah, a hundred percent. So I want to shift into where this is all going because every, you know, every conversation I have like this, especially with my own CTO, just gets us so jazzed and excited about what the future looks like. So I'd love to just pick your brain for a minute and from your perspective, where is all of this going? What does the software engineer of the future look like? What is, you know, what do you see? I don't know, 12 months from now, 24 months from now, maybe even sooner because it's all moving so quickly.

SPEAKER_00 51:15

Yeah. Look, I like to think about it from the perspective of one of my friends shared, which is a job is a set of skills applied to a process. And when you break it down like that, you shouldn't be scared then of what the process is if you can apply the skills, or what the skills are if you have the process. Therefore, you shouldn't be worried about what the job is. Right. Now, I'm not saying like AI is not going to display certain jobs and skills and capabilities, but what I do think is it's going to elevate people from lower end roles to higher end roles.

SPEAKER_03 51:43

Yeah.

SPEAKER_00 51:44

You know, here's something to think about. It's really easy to say that AI is super, super smart because it knows everything. But it does not know everything. It doesn't know what I'm thinking. It doesn't know the context of my whole business. It cannot participate yet. We're not at the level where AI has access to every meeting, every transcript, every conversation, every document, every thought process to then know everything about my business. It can't. So I still have a very, very important role to play. Right. I also think of AI as this, let's go back to the elite toddler syndrome idea. It's this elite, capable, knowledgeable person with no skill set. Until you start talking to it and asking it to perform, it has no skill set defined, it has no role defined. And in fact, every time you shut your computer or that session down, you get a brand new blank version of it. It's like a robot who worked really hard and learned everything. And then you shut it down overnight and you come back the next day and you have to train it all over again. So where I think we're going with this is A, people should elevate themselves. They should think of themselves as the orchestrators, the planners, the choreographers of what they're trying to accomplish. What if every employee on your team could now have access to 10 or 20 or 30 experts at any point in time, any day of the week, without any ego, without any cost effectively, and now have 10 people on their team doing helping them in certain ways. I think software companies are building out processes driven by AI to enhance their capable feature set. Yeah, some are putting AI in front of the user. So many of them are just doing it behind the scenes to make the process better, faster, et cetera. So I think we're going to see a ton of that, a ton of enhancements in how we get efficient and how fast we can go. I think there's a lot of doom and gloom around, oh, AI is going to take over the world and models are so smart and blah, blah, blah. I I don't really buy into that. Yeah. I don't like to think negatively, even, uh, because humans are incredibly smart, incredibly clever, and capable of doing things that I don't think AI is ever going to do.

SPEAKER_03 53:48

Yeah. And creativity and curiosity and, you know, all of the things that make us uniquely human. I really feel, you know, when I kind of feel into what's coming and certainly what we're building and the vision for, you know, our future for our customers and our platform, it is really exciting. And I see is more important than ever for, yes, everyone to adopt and use it. That's super important. And it is also more important than ever that we lean into what makes us most human, right? That is the differentiator. Now we get to be more of ourselves, more vulnerable, more human. And I think those who embrace and adopt AI and show up fully authentically as themselves and bring their unique human gifts to the equation, those are the ones unstoppable.

SPEAKER_00 54:40

I asked somebody yesterday who has a business that I admire, how do you determine how profitable you should be? And his answer was, we break even. Like, I'm not worried about profit. We're growing, you know, we have other things that are benefits for this company, et cetera, et cetera. And I think that AI is going to make it so that things become so effective, so efficient that we as business leaders could actually say, I just have a profit goal, 25% a year.

SPEAKER_02 55:08

Yeah.

SPEAKER_00 55:09

And everything beyond that pays people more money to work less hours, but they get to use their skill set.

SPEAKER_02 55:15

Yeah.

SPEAKER_00 55:16

Look, that's a little altruist, altruistic. And but I think that's how it should be in a lot of ways. And I know there's always going to be, I want an 80% profit margin, but you can't have more than 100% profit. So at some point, right? At some point, you're paying people to do less work more effectively, more powerfully. I think we're breaking some of the industry standards of the 40-hour work week. I really don't care how many hours people work. I never have.

SPEAKER_03 55:42

Yeah.

SPEAKER_00 55:43

I care about productivity and outcomes. So if AI helps you get there 10 times faster, 10 times better, hallelujah. Should we give you more? Sure. But we're doing so much already, you know?

SPEAKER_03 55:54

Yeah, exactly. I mean, one of the goals that I've had from my team for a few years now is getting to a three or four-day work week. And it has just not really been possible without like it, it's only possible now with being able to leverage AI. And so I see a future where that is possible, where we can hit our revenue targets, our profit targets, and there is work-life balance for every employee. They're making more money, they're working less. And the way that they work, the way that they're engaged is really capitalizing on what makes them uniquely human and themselves.

SPEAKER_00 56:33

So fascinating times in front of us for sure. And I don't know about you, but I have two different emotions going on quite frequently. One is a little bit of FOMO, like, oh my gosh, something new came out. I can't keep up with it. I'd like to look at it.

SPEAKER_02 56:48

Well, it's impossible.

SPEAKER_00 56:50

Yeah. The other one I'm experiencing, I posted this on LinkedIn. I'm calling it AI guilt. The feeling you have when AI is actually not working right now. Like it's not doing something else. I want it to be so productive. I've got 10 things happening at once. And if I have nothing happening, I feel like I'm not doing enough.

SPEAKER_03 57:11

Yes, I feel you on both of those things. And it's just like like you said, it compresses time. So everything is speeding up that you know, that all the new information coming out, it's impossible to keep up with. Like there's just too much. And yes, there's so much we can be doing with it that like there's also that guilt or fear of missing out on what you're not doing with it. So I want to bring this back to, you know, the IT leaders listening. Because you've been in this space, you know, over a couple of decades. You've seen a lot of shifts. What has fundamentally changed about how IT leaders need to think right now? Because, you know, I think you've given me, you've given me a few mindset shifts throughout this conversation, which have been really helpful. And you're already thinking about this more deeply than most. So I'm really curious what mind shift, mindset shift should tech leaders be making right now?

SPEAKER_00 58:14

So, look, the first thought that comes to my mind is that tech leaders, IT especially, has a protection mode default. Like we have to be secure, compliant, the SOC2 compliance, for example. We want to protect our assets, our systems. I'll be the first to tell you I'm never going to let AI touch my production system. Why would I? Why would I give it the risk that it could delete something, harm something, break something, right? So the role then has to be how do we automate better, how do we streamline better? But I, as the human, are always in the loop. Yes. Um here's another aspect that I think about. Look, I am building software to autonomously build software, which you could say is eliminating engineers or work that could have been done by people. And that's somewhat true. But I don't think it's eliminating. I think it's multiplying how many things you can get done now. Right. All right. And it can give IT leaders the ability to choreograph projects that they just don't have the resources for. They don't have the ex expertise or the people power. And I think IT leaders often lack enough team members to do what they want to do or being demanded to get done. So AI is going to change that. But the question's always going to be, and this was asked at a round table in New York a couple years ago, what happens when the system goes down? Do you trust an AI to fix it because the AI wrote it all? Or do you have the humans on the team that can fix it? And I think it's a blend of both.

SPEAKER_01 59:34

Yeah.

SPEAKER_00 59:35

You know, again, this is something really, really hard, Tanya. How do you get soft AI to be deterministic? And that's a big word. Let me narrow it down. How do you get it to build reliable, trustworthy, defect-free code? Because right now there's a lot of slop. They call it AI slop. There's a lot of people who got fired because AI could do their job, and now they're getting rehired because AI did a bad job and they're coming in to fix it. And what I've been building is creating much more perfect code. I'm not meaning it's totally perfect, but we did a project and the first pass came out defect-free, bug free. Was it fully acceptable to go into production? That is unreal from a software development standpoint. I think whether it's my solution or other people's solutions, we're not far off from a point where an IT manager and leader can say, yeah, let's go through the design carefully and let's push a button and have AI build it, and now it can be deployed. And if it's deployed, now you the question is, how do we manage it? What should happen? So if you as the IT person have the control of turning it on or off, like with feature flags or rollbacks, et cetera, but you also have the ability to look at the code because you own it, and you have this AI tool set or team that can actually go in and address the problems and monitor the problems and fix the problems, then I think you just have a really, really powerful way of looking at the world. And right now we're really scared about should AI touch our systems or not? What are we going to allow it to do? How do you break through that fear problem first? And what I just said is part of the solution. You got to have tools you trust. You've got to have a process that you can enforce and control.

SPEAKER_03 1:01:16

Yeah. Yeah. There is, there's a lot to think about. And I could actually talk about this stuff with you all day long. I, you know, I'd love to invite you back to go a little bit deeper on the conversation, especially maybe, you know, in it in another couple of months, because I feel like things shift and change so quickly. I'm would love to hear where your head is and what your mindset is, you know, a quarter from now, if you'd be willing to come back and talk more with us.

SPEAKER_00 1:01:43

Cause this is I'd love to because I'm hoping to actually have my side project become a real product and have that deterministic outcome. And it would just be fascinating to get feedback and input on that and to talk about the breakthroughs because I keep making breakthroughs with it that are just to me mind-blowing.

SPEAKER_03 1:02:00

Yeah. Yeah. I would love that. This was such a fascinating conversation. I really could stay in it all day long. And, you know, I think that the biggest shift you brought forward is this idea that AI isn't just something that we use, it's something that we work with and we can interact with it on a very human level. And I think that's a really different mindset than how most organizations are approaching it today. And yeah, like I said, this is such a fascinating conversation. You've got me thinking, now I just wish I had time to go play with AI for the rest of the day and try out some of what I've learned.

SPEAKER_00 1:02:37

So I look, I love this stuff. I appreciate the opportunity to talk about it. And your total interest has been super fun to try to explore this deeper. And I keep uncovering layers that maybe I haven't thought enough about. So thank you for that.

SPEAKER_03 1:02:51

Yeah, I would love to continue the conversation. I think there's so much here for IT leaders, tech leaders to sit with and rethink, and so much here to unpack and so much opportunity too. So if anyone wants to learn more about what you're building or wants to connect with you, we'll include your links to LinkedIn and your company QUIC in the show notes. I believe we also have a landing page in Technology Match for Quick, which we'll also link in the show notes. So if you're a tech leader listening, you're just trying to figure out how to move faster with AI automation or overall strategy, you can go to technology match.com, enter QUIC. Again, we'll put everything in the show notes. Rich, thank you so much for being here. This was like one of the most fun conversations I've had in a long time.

SPEAKER_00 1:03:36

Wow. Well, thank you for that compliment. I really enjoyed this too. Thanks for having me today.

SPEAKER_03 1:03:40

Yeah, thank you. Thanks, everyone. Thank you for tuning in to Between Fires and Futures. We know the weight tech leaders carry, the pressure, the pace, the constant pull between keeping things running and building what's next. If no one said it lately, you're doing hard, important work. And we see you. If this episode sparks something for you, follow the show, leave a review, and share it with another tech leader who gets it. Thanks again for listening. Keep leading through the fires and daring to build the future anyway.