Latest / CTO Lunches / Ep2 - Rick Manelius on Forward Deployed Executives

Transcript

Ep2 - Rick Manelius on Forward Deployed Executives · CTO Lunches

• Rick: We have a lot to talk about, so I'll Kendall: That's right, that's right. That's right. Yep. Rishi: Yeah. Rick: try to be concise, which is not my strong suit, but I'll do my best. Kendall: You're not here you're not here to not talk about yourself, so please Rick: Yeah, yeah, yeah. Kendall: l loquacious eyes yourself. yeah. Rick: I love the lexicon we're going for today. Kendall: Yeah, yeah, it's good. Rishi: I'll be right back up to get a dictionary so can keep up with you guys. Rick: Ha Rishi: Kendall, what are you what are you drinking? Kendall: So good morning, Rishi. After shaming me last time we did this, I am drinking tea this morning. So I'm I've been caffeine free for a little over a year. So finding decaf teas are hard, especially loose leaf. But this is a an oolong. It's probably my seventh steep because I drink a lot of tea all morning long. But so there's there's not much left, but the the oolongs last longer than the the greens or the whites. But that's what I'm drinking. What are you drinking? Rishi: Well, for me I actually felt a little bit bad about the last time in trying to get out there. Kendall: good, good. Rishi: So I have both my coffee, and I have a final ward, cocktail right here, made with Friday deployments whiskey, and then the rest of the the good ingredients. You know, chartreuse always makes things go well, in my opinion. Kendall: Wait, it's the last word, right? I don't think I've ever heard it called the final word. Rishi: the final word it's final word with an A is a riff on the last word. so it's made with with whiskey and lemon juice instead of the gin and lime juice that's typical in a last word. Kendall: Okay, today I learned. Okay. yeah, Rishi: Yeah. Rick, what about you? What are you drinking? Kendall: Rick, what ya got? We heard you open a a can just a moment ago in in the microphone, so now I'm Rick: Yeah, so so Yeah, hopefully not a can of woopass, but yes, canana banana water, organic. Rishi: Yeah. Rick: Woodstock. I don't own any stock in this. it is expensive. I would not order an Amazon. It's more expensive than a beer, but my brother swears by this. everyone's into like these electrolytes and magnesium, but he works as a plasterer and he does like ten or twelve hour days and he's like, This stuff actually keeps him energized. So I was like, you know what, why not? I'm not a big banana fan. Tastes good. And it's also one of those like weird, odd things that, hey, why not? Let's give it a try. Kendall: Can't c couldn't you just Rick: Yeah. Kendall: just h hold with me for a second. If it's more expensive than beer, bananas are some of the cheapest fruit. They're like fifteen cents a pop, right? So I mean what can one banana cost? seven dollars? sorry, that's an Rick: Ha ha ha. Kendall: arrest development reference. but can't you eat a banana and then drink a glass of water for, I don't know, fifteen cents? Rick: You absolutely could, but what's the fun in that, man? What's the fun in that? Yeah, I there's th there's they're they're more it's more Kendall: Okay, I mean I don't care. it's but if you don't like bananas dr you know, sipping the go ahead. Rishi: I Rick: you know, you know how fast bananas go, man? It's hard to keep up with bananas. They they they go brown so quickly. This I can keep these cans in here forever as a prepper. I can keep my my banana water forever. And look, it's like the farmer's market. Exactly. Kendall: Fore forever banana water. Yep, yep. Rishi: I think it makes sense that it costs a lot, right? Like I can go downstairs, I can crack open a banana, I'm gonna get like three or four drops of water out of that banana. So I mean you're gonna have to go through whole bushels of it just to make one can. I think I think it makes sense. Kendall: You just think you're thinking about how Rick: Bush just Kendall: many bananas you have to juice to to make the Rick: Yeah. Yeah, Rishi: Yeah. Rick: I I I I can't wait to set up my own juicer. Like I wanna go like you know, like an orange juice, Rishi: I mean, you know. Rick: but now I have a banana press in the back just like to squeezing out those little bits of water. Kendall: really gotta put it in an industrial squeezer to get enough water out of the bananas. Rick: You go to Costco, you fill up the back of the house the car, you just you just like just get that one can as like it's absurd. It's bananas. Rishi: Right. It's a l it's an exotic luxury good. I think it makes sense. Awesome. Well Rick: Mm. Rishi: let's get let's get started. Hi everyone. and welcome to the CTLs podcast. I'm Reshee Malik here with Kendall Miller and Rick Manelius. Rick, why don't you start us off and introduce yourself? Rick: Yeah, so engineer by training, I've done startups now for eighteen years, tech startups. I've worn every hat. So started off more of the developer, then turned sort of C C O O, let me get the progression right, CTO, CPO, CEO. And then you know, these days the last year I consider myself more of a for deployed executive, which I know is very buzzwordy, but in this era of AI transformation, There's this growing need of people who are both technically capable of building stuff with AI, but also have been in board meetings, leadership, investment meetings, et cetera, and also have like operational expertise of like how to actually make these things happen. And so my last year has really been working with multiple companies, all the way enterprise, SMBs, traditional startups. and so that it's hard to kind of make one title for that and this one is the one that feels right and I think others are kind of landing on the same sort of archetype as a as a pattern. Rishi: Excellent. Yeah. I'd love love to dive into that. Go ahead, Kendall. Kendall: I really Well, I like the I really like the terminology, forward deployed executive. As as you're right, it's buzzwordy as heck. H star C K heck. Because but because like I think what makes an executive different is the capacity to kind of dive into any of the problems and solve them, business problems or otherwise. And the when you add the forward deployed onto it, what you're saying is, You know, this person can leverage the technology to solve the business problems. And I like there's so many business executives that can't leverage the technology to solve the Rick: Yeah. Kendall: business problem. And Rishi: Well so Rick: I and this results in the failure bo y yeah. Go ahead, Rashid. Rishi: so hang on, but before we dive in, I I was just saying, Kendall, I think it's a great time for you to introduce yourself as well. Kendall: yeah, sorry. I'm Kendall. you didn't Rishi: Okay. Kendall: actually say what you were up to these days, Rishi. Maybe that was intentional, but Okay, Rishi: I'll be coming back. We'll get there. Kendall: okay. I'm I'm Kendall Miller. I'm CEO of a company called Deployed. We're doing local AI things. I also run CTO lunches and do Friday deployment spirits with Rishi and a few other things, but I'll I'll leave it at that for now. Rishi? Rishi: Excellent. Thank you. Rishi Malik, tech executive, tech leader, you know, AI coder, an apparently confused individual. Cause when I hear forward deployed executive, I was imagining sending executives down behind enemy lines, putting them in real life mortal danger. And I was excited about getting a number of executives that I know out there in peril. Yeah. Yeah, Kendall: You have a short list of executives you'd like in those trenches as soon as possible. Rishi: so launch them out of a cannon behind enemy lines type of thing, you know. We'll have to work on that. But yeah, so you know, I I I've done lots of different things. Friday deployments is is a big focus, but these days I do a lot of data in I AI consulting along with security in the agentic and AI age consulting as well. but let's jump just just let's maybe start with kind of a recap of you know what a for deployed executive is, and then maybe Rick you can tell us about how how you started on that that path. Rick: Yeah, let's maybe start with that part and that kind of shows like the failure modes learned and kind of like why I think it requires sort of like all three of these facets. you know, I joined a company called Every as a an AI fellow about fourteen, fifteen months ago. and at that point we were still in the like Chat GPT projects era, right? You know, I just get in here, run through some basic beginning through advanced prompts, show how to use project folders, connectors, etc. Obviously, we are miles ahead of this as past the Cloud Code era, the open claw era. I mean, we just saw the Navier Stokes equations being solved. I mean, we're in a a different paradigm now. through that work, both with every and then as an independent consultant, you start to see the different failure modes. Obviously, there's no it's not a secret in the industry. Like eighty percent of these transformations fail. Either it could be like fully top down, like the executors like mandate, yes, we want to be AI for like everyone should use AI tools. Like so you get this mandate, this sort of like visionary decree, but right, without sort of like I would say the two other components. and this is similar to the emyth. The emith talked about, you know, of the entrepreneur, the technician, and the manager, right? So the leaders could have the vision, right? But without like making space and and to help actualize it. And without having the technical expertise, the people that could do it and do it well and actually see this actualize, you can't really sort of see this move forward. And so either bottom-up only or top-down only strategies always seem to like hit this bottleneck. And what I've sort of seen and experienced is this sort of forward-to-play executive kind of has lived all three. They're they're strong enough builders, usually former engineers, they feel very comfortable working beside the team with Cloud Code. They are very sort of entrepreneurial, visionary, innovative. They're they're willing to get creative, think outside the box, see where the the puck is going, be willing to get messy, you know, explore things that are aren't quite solidified. but they can also sort of bring it to a timeline and they can really bring it into like a like how do I make this space? How do I sell this vision on the team and and like take these wins and then reapply it, not to make more code, not to more make more things, but how can I reinvest that time operationally and get it to the right places to expand out. And so, I I think of it like again, those sort of three parts of the triangle. And when you can when you can live in all three, it's the most powerful, right? Because you can engage the experts at the company from those facets and be that kind of like Rosetta Stone connector that can see this improve. Kendall: The Rishi: So Kendall: so so one of the things I think is Rick: Yeah, please. Kendall: interesting about that is I mean this was just a a discussion we were having yesterday in the the the CTO Lunches Plus the the chief of staff AI chief of staff conversation that we have. one of the things that came up was That the way we make decisions has changed so much because of how fast AI makes us. So historically, we need to, you know, products talking to a customer or engineer comes up with an idea, brings that to product, product discusses that with their leaders or other people, confirms that the idea maybe makes sense, then sits down and figures out, okay, where do I slit this into the roadmap so that we can get it done in a reasonable time? How high of a priority is it? and then, you know, it and given the priority, it might be there's two other 200 other things that are that are warring for our time and attention and the engineering's time. But when In the in the age of AI, sometimes it's I'm having a discussion with Rick, I pull up Claude two minutes into that meeting, I ask it to go implement the thing, and 20 minutes into the conversation, I have a prototype to kick the tires with and go, maybe this is a good idea or maybe this is a bad idea. I mean, like literally, you might be able to create three versions of the thing faster than you can schedule Rishi: Yeah. Kendall: a meeting to discuss should we build the thing or not. And then when you get to the conversation around should we build the thing, you go, hey. Number one, I already built it. Number two, I've already been through three iterations. And number three, here's how I can already prove it has value. Instead of should we do this? And let's make a decision because engineering costs us so much and is so time expensive, financially expensive, etc. And now, as that cost has gone almost to zero, decision making is literally, well, why don't I try three times? And then I'll decide if it's a good idea. Like, let's actually bring it into the world. And then decide if we want it. And that's kind of wild. Like it's cheaper to bring it into the world and then decide is this worthwhile than it is to try to come up with it in a nebulous sp and like I just think that paradigm shift Rishi: So Kendall: is a really big adjustment. Rishi: I is this something then where we have to and maybe this is Rick tell me if I'm wrong, but like part of the the concept of this forward deployed executive, that definition of the output metrics that you need to know if which one of the three builds you just did was the was was successful. Rick: Hundred percent. And like you with that experience as being a startup founder, you you you're get used to part p pattern matching, right? You know, part of like Techstar's mentor is like like how do I sort of that word taste is overused, right? But the idea of like as it's coming back, is this good? Will this would this actually be used? Would a customer actually use this? And I like to invert this because I think a lot of people, Because it used to be so expensive, like we we we don't like to kill our darlings to use Shakespeare quote. Like when we POC something, there's this immediate need to like and now now we can just get into production. It's like, well, actually, this is where you need some seasoned person, an expert that is gonna sort of kill the bad ideas faster. Because like what Kendall's saying, like some of these ideas in your head that they seem better and bigger than they are, but then when you actually POC it, you may find out it's actually really bad one. In fact, most of them are still like not worth carrying through. So if you can come at it with the right rubric, which is We will only ship it when the POC is an absolute hell yes, as opposed to like it's good enough, it's you know cheap enough, so let's just deploy everything. that it it sort of changes around what your standards are. And it's very, very important to have the right kill criteria and rubrics in place such that this this newfound superpower that you have of building things and prototyping faster, which is by the way, amazing, right? Because you're getting fast feedback, you're you're learning faster. Before, if I gave a task over to an engineering team, it may take four weeks to get the the thing done. And then only then can the learning happen. Only then can we taste, touch, you know, do the things and determine like was it working? And now, like you said, you can do it in twenty minutes. So if done well, the speed of learning in our iteration cycles and and knowledge gains is massive. And so but you have to contain it within the right rubric. because Kendall, we also had a an AMA recently with Ori, who has this new AI native not a new one, but an assembled software development lifecycle. And like as long he has lots of these frameworks and rubrics in place which give him the guardrails to take this newfound superpower and be able to bring it into a way that change management and culture can absorb it at the right impedance, right? As opposed to the chaos of all this newfound building power. Rishi: The AMA with Kendall: It's Rishi: Ori Ori Steel was was excellent. So yeah, I I definitely see that. I almost wonder though, and you know, Kendall, if you want to take this a different direction, you know, jump in, feel free. I I I wonder part of my theory is with the the lower cost of build, yes, you can build it three times, you gotta pick which of the three is the best one. but Rick, it almost sounds you're saying a little bit more of like what I would call product or business or market segment vision. Like cool, we built it, but you know what? This actually isn't a good idea. We need that kill criteria to say we should not have built this feature or this this aspect. And I'm wondering if because maintenance costs are so low that like you don't actually have to do that. And we end up with a lot of apps that do kind of everything. The segmentation that exists that you know has blurred lines of kitchen sink apps that do kind of everything you you need. Kendall: Well, so I actually address that first, Rick. I have another direction that I want to take this that's related, but like you have ended up playing a lot of times the product role, particularly in early startups when you're the CEO and you have to decide which direction this is going. You've pivoted companies, you've done all kinds of things. Rick: Yeah. Kendall: So so in that I'm curious related to Rishi's comment, like I've always thought the most important thing a product person does is say no. And so I'm curious how this changes that. Like we can say yes to everything, basically to Rishi's point, maintaining it, im iterating on it, et cetera. We can we can almost make kitchen sink everything, but maybe that's the worst possible sni know. I I don't know. I'm just curious like how Rick: Yeah. Kendall: does this affect that? Rick: So I I went through a a very strong like learning curve on the sort of sales go to market side with a a company called Jellyfish, Jellyfish with two J's, working with like Jen Abel and and other folks there, which is Like look, it's not a matter of like can you ship like a hundred features in this app, right? Because ninety of them may not be used at all or n and maybe five are actually the differentiator. In fact, maybe the kitchen sink approach has actually distracted the core product and you're actually reducing its usefulness and utility for the your end user, right? And so I like to, you know, have a stronger validation loop with like we can build this feature, but like how are we going to quickly validate that users are gonna use it, that it's actually moving our company forward in doing so. So But I love this, right? Because you could rapid like, as opposed to showing up to a customer discovery conversation with a question, you come up with a prototype. So the advantage of being able to build this fast is you're not just talking theoretically like, hey, it's live in this PR, we're ready to go. You know, is this meeting your need and gauge their actual experience testing it in real time, which is a huge difference versus talking about theoretical questions about the problem space. You can actually see their experience live with these prototypes that you could stand up quickly. Kendall: Well, and so that's the direction that I wanted to go with it is what I heard many years ago and and it still lives in my head that the the primary difference between an executive leader and you know middle management or other people in the organization is the executive has to live with incredible volatility and incredible uncertainty. They have to make decisions with very, very little data, right? That you're all the time making a decision with ten percent of the data. And and that's it's one of the hardest things about being. An executive is like, yeah, I would love to tell you confidently which direction is the right way to go, but we don't actually have a clue and we have to make a decision right now. So I'm gonna say this is the direction we're going and we're gonna go try it out. And I think what's interesting about this shift, and and this is related to the title, Forward Deployed Executive, right? Is if you have the ability to think of something. Make that thing exist in the world, iterate on that thing, in the time it used to take to brainstorm as a single person, is this a good enough idea to even have a meeting about it? All of a sudden Rick: Yeah. Kendall: you're making decisions with substantially more information. I mean, at least in the software space, right? Like there's there's an awful lot of people for whom AI has not yet, you know, whoever's making your banana water probably does not get to just create a product, fiddle with it, Rishi: Yeah. Kendall: try 10 more ti I mean, think about how much squeezing they're having to do to get, you know, that much Rick: I mean Kendall: liquid for for version two, right? I mean, frozen Rishi: It's lot of squeezing. Kendall: bananas, like we're we still need 500 of them just to make one cup. but Rick: that's the worst. Kendall: I mean i am I right there? I think that that's one of the things that's interesting about this is it's giving you more information is what I'm hearing. Rick: Absolutely. I mean again, AI used correctly is, you know, in DevOps they used to optimize for the in the three ways with Gene Kim. It's like, you know, accelerate, you know, the first time through, how what's your cycle time? Because then you get this sort of continuous learning and innovation, right? This sort of like faster learning cycle with safety and security and guardrails, right? The you're not like abandoning that principle. You're able to do it, you know, with the right process and pipeline. And this This is where I think the executive slash operations side of me comes in, which is created the right rubrics. So like I have these scoring rubrics, you know, similar to like the rice matrix in the product space, which is like as these ideas flow through Actually test like what's the effort, what's the value, is it recurring, is it repeatable? Like it you know, can I move it to deterministic code? And when you can sort of like do this and then quickly without a meeting, like validate it, you know, rank it against a thing, and then say, does it match your vi values? We have a context file of like our values, our roadmap, our rubrics. again, similar to what Ori is doing on the software side, but even for non-engineering teams, like what's the product, what's the business rubrics for these evaluations? All of a sudden you can go back to the boss being like, hey, POC works, and by the way, I've already scored against our things, and by the way, it does this thing. like it becomes a no-brainer, right? And and you said before, Kendall, that part of a CEO's job is like working with like incompleted information and fast and vol, maybe it's Rishi, it's fast and volatile, which All the more important in AI where this information is generating faster. So if you don't have these rubric guardrails in place to like help catch and like score things along the way, then it feels like complete chaos. And then you s have people panic and halt because they're like, there's so much chaos going in the system. I don't know where all this is. It's even it's even harder to get my hands around. But if you can kind of add these sort of CI CD pipelines, but in an AI sense, you can say, actually we've we've tested these things with these rubrics, here's what's coming through. Our product people are picking things that generally match our roadmap and generally match these repeatable processes and all of a sudden you can have that confidence because you baked that into your process. Kendall: So h how much do you trust Rishi: Hmm. Kendall: the AI to make those decisions that like you write the rubric? Part of it part of the problem with trusting the AI is I have to have confidence I wrote a coherent rubric, I have to have confidence I wrote a sufficient rubric, I have to have confidence the AI can make what's the word? you know, common sense decisions, right? Because when I write a rubric Rishi: Always risky. Kendall: Well, Rick: Ha ha. Kendall: always that that's what I'm saying. When I write a rubric, my default is to assume that the person reading the rubric can read between some of the lines. And the AI Rick: Yes. Kendall: can't re read between a single fucking line, right? Like it doesn't have a clue. And so how if if you're automating I I I believe that there's no way to do this without automating. I I believe that to manage AI, we need to leverage AI, but how are you building Rick: Yes. Kendall: systems That you trust enough to actually give that to it and is it just you always pay top dollar for fable because it has the best common sense or So go go for it. Rick: Yeah. No. No. It's it's not about that. I mean we we actually had one customer where sonnet was the better model, right? Because it was less creative than Opa like Opus wants to be the almost like the more creative architect where Sonnet's like I just want to the job, right? And we had this rubric where I can't go into too many details, NDAs and stuff, but basically evaluating these hundred page like compliance reports. the first reviewer would leave all these inline comments, and then a senior reviewer would go and like correct those comments and then like so there's like a twelve to twenty hour process of like just commenting this very large compliance legal document. And because we have this this golden data set of like hundreds, if not thousands of versions of all these inline comments, we could score like how well did the AI agent like full find the area, what was their summary, and how much did it compare against another another human? And we had the humans review the AI's comments. So like we went through like multiple layers of evals, like you know, using another model to benchmark the other agent and then using the humans to do the final check. And when you get to some number like 95 to 97%, you feel pretty good. And if you have like a a a probability score and some other stuff, you're probably gonna start like zeroing in on those needles in the haystack, the ones that are that are suspicious, but trusting the the the lower hanging fruit ones because you rank those risk low or medium and only can focus on the high item ones. By the way, that level of confidence is probably better than most humans. Like if you gave ten humans the same document and had them all score that like what they identified and what they summarized it would would probably fluctuate more than 10%. So you might have a 90% consistency among humans. So as long as kind of you're within the tolerance level of what human variation might be on a reviewer by reviewer basis, you would feel confident that you've done the evals once and kind of like you'd have test coverage, you'd have test eval coverage, like once a week or maybe like a nightly build, the model changes because Opus does change. You sometimes compare to the API, sometimes you can't. And so you would wanna say like here's our golden set, and every time a model change or every night we want to rerun it against these things and just verify this. It costs some money, yes, but if something now lags or like trips you know off of a threshold, now you have an alert, now you kinda look in a little bit deeper. and then you should always still do human spot check. Any QA system, I I worked at a plastic factory when I was fourteen, fifteen, like you still have manual oversight, right? You still have to like take the QA's work and look at one percent of them and then check it again, right? And so you You can trust it to a point, but you still should have human in the loop, but then it's at scale, right? You're then you're checking over a thousand runs as opposed to like per run, right? So you're still getting that leverage of using AI in bulk and then sort of spot checking to make sure nothing's drifting weird. Rishi: So you said the E word, which I I love. Eval is, you know, fantastic for everyone to be implementing now just from a pure code perspective. But I, you know, I think the concept will generalize, and we'll actually talk about that in a future episode some more. but when I think then about that concept and I think about, you know, what you're then building with AI kind of across the board, from that forward deployed executive aspect, I'm almost wondering if The future of being an executive then is you're writing evals for your area. Like CTO, sure, it's all code and what the LMs are doing it. But if you're CMO, you have your evals around how well your marketing is working. You know, did you ship this thing? Did you know top of funnel improve? You know, whatever your your key metrics Rick: Absolutely. Rishi: are there, right? So that i is is the future of being executive just running a team of agents. Kind of at any scale. Maybe it's ten, maybe it's a hundred, maybe it's a million. Rick: Yeah, it it's th th I think there's there's a lot of focus on humans putting guardrails on agents. I love the the reverse. Like we need to have agents put evals around humans. one of my clients is very much in the compliance space Rishi: that's interesting. Rick: and like look, everyone hates AI writing now. Like literally people have visceral reactions to it. And so we actually have a three agent setup where we run it through a thing called humanizer, we run it through their brand guide because sometimes market people pick wrong colors, wrong fonts, wrong whatever. And we have both agents and determinism, like code deterministic code. So like we can we have a a markdown sorry, a PDF to markdown extraction. We have a font extraction thing. So it's like, are you using approved fonts? Are you using approved like font sizes? Are you doing all this stuff? So like Everything in the marketing collaborator goes out the door now actually uses AI and code to better ensure that the quality levels of like what's being produced is matched. And so that's an area in which AI is actually helping consistency. You're not getting more indeterministic behavior, actually helping put guardrails on humans, ensure they can still be creative, but they still like meet standards that the company has set forward. And so that is something I'm seeing a lot in these companies, especially ones that don't trust AI yet. Perfect. Let's use AI to show you we can make the humans better and and catch more error rates and and so forth, build the trust there, and then we can extend to other areas. Kendall: So I don't Rishi: I I l I love that hot take. AI should be putting guardrails around humans. Like that that just Kendall: Well and Rishi: like flips everything open AI and dropping it, everyone is talking about Rick: Well Rishi: on its head. Now I absolutely love Kendall: Wait, wait, wait. Rishi: it. Kendall: AI Rick: She. Kendall: putting guardrails on humans is the story of the Terminator. Like, that's AI Rishi: Ha ha ha Kendall: coming back to guardrail by killing the humans. I mean, am I wrong? Rick: So well hold on. Hold hold hold hold No, there's you know in extreme ownership they talk about, you there's freedom and constraint, right? And they they use this whole like military operation of like when we standardize around this process We actually improved speed and improved quality and did all this other stuff because we don't have all this variation. And in one of the companies I'm I'm I'm consulted with, they have fifty different NDA templates across the different like business units. And like how different can an NDA you know get, right? You know, most of it's just a template with maybe some variation, right? So if you could standardize and say, look, we just don't do this. We actually have a check where every time something gets uploaded to SharePoint, it compares against our standard NDA, you know, things. you still have the flexibility, right? But it's within the the speed optimization of consistency across the brand and across the company. So I guess what I'm trying to say is like, no, I'm not talking about Terminator style thing, you know, like the Amazon warehouse stuff where they're tracking like every second how many bathroom breaks you take, that's that's a level I think that's none of us would work for. That's dystopian. But if you could say, like, look, we can make sure that the the stuff that goes out the door for marketing isn't AI sloped and we're catching and correcting and coaching people, like, hey, this thing can't go out the door because Rishi: Yeah. Rick: blah, blah, and blah. Like this was just and it's like when that guardrail's there, it's kinda like you can't submit a PR if it doesn't pass the green board, right? It's like, man, yeah, just this just Kendall: Yeah, it's just AI linting. Rishi: Yeah. It's a l yeah. Rick: J just this one PR, man. It's like yeah, it's like no man, that's Kendall: Yeah. Rick: breaking my test. Like you can't do it. Kendall: Yeah. It's so Rishi: It's more fun to think about it, dystopian wise, Kendall: I do I do like Rick: Yeah. Kendall: the dystopian nature of it. okay, I want to change the direction a little bit. Rick, you wrote a book called Save Your Startup. The Things Rick: Mm-hmm. Yep. Thank you. Kendall: That AI's doing and the way that it's changing executives in the you know, the way you're thinking about the forward deployed executive. What would what's in this book That you would change based on how the entire world has shifted in the last few years in our industry. Entire world in our industry. Rick: Yeah. Kendall: You heard it here. Rick: Yeah, I mean I would say maybe seventy percent of the of the book actually is more valuable because I actually was able to turn them these I I made it very practical with playbooks because as a Techstars mentor, you know, you're working people through heuristics like here's how you might you know prioritize a roadmap, here's how you might think of like there's all these like sort of like frameworks that you're sort of merging together and making it your own. And to skillify those is actually valuable. In fact, I use them on a daily basis. My prioritization framework, I use it all the time. When I check my week, like what's my North Star, what's my next action, what's my next week, I actually use my own skills because that that's useful. where I think it's no longer relevant is that full collapse of build cycles, right? You it's like you used to you know, there was very much some some a focus on like how do you like break apart POCs, MVPs and and and you know full builds. and again, we've talked about this at length in the CTO group. it used to be maybe one product person feeding a team of engineers, now it could be like, you six product people feeding one engineer. Like it's completely inverted like the balance of ideas to delivery. And I think there's still kind of a a rework of like what the the shape looks like. Also, things are less less durable. I think we're just kind of in a fash fashion fast fashion era, which is like it's not so much to like grow and we can't be keep building this monolith and it's gonna take you you know 10 years and you know you're gonna build it in different sequences gonna add new features. It's like staying relevant requires an incredible adaptability. It's it's actually okay, if not better, to keep things in sort of MVP stage, get people to a win because the the context of the world is changing every three to six months and sometimes even more aggressively. So whereas before, I would be less inclined to to tolerate that level of change and pivoting I think now it's like the bias towards pivoting is actually so much stronger. because the paradigm can change underneath you and it's like, dude, you just can't have a a business model that doesn't fit the current paradigm anymore. It's it it it could come out underneath you and you have to be willing to like whole hog change on a dime, which is terrifying to certain people. But we've lived through this. COVID was another example with my my company contact mapping. We were a personal CRM, which was all about like in person coffee shop stuff and meeting people for network marketers or real estate agents or solopreneurs. And the week or two before the shutdown, I kept saying to the team, it's like, guys, we built an app for in-person stuff. And like it's we see what's happening in Italy and China, like it's going to happen to us and we're gonna have no this app isn't gonna work. Like we're gonna have to figure out a new way to like train or help people in the same problem. But know following up with the problem, not the solution, but our our current solution isn't gonna work. And there's like, no, it's not gonna happen, not gonna happen, not gonna happen. And then the day the shutdown happened, it was like, shit, here we are. Like we have to completely rethink how we're doing things. And we don't have choice. Like the the bridge bridges behind us are burned. And I think right now, like you have to assume the bridges are being burned behind you every three months. Like what is happening, which is terrifying for some people, but you just have to just accept that it's Here's the sort of the hot take. And I've told this to one of the companies that I work with. Like, their fear is that it's too risky to adopt AI. My statement is if you don't, like your irrelevance is good, like you're it's more risky that you fail that you don't keep up with it all and you're irrelevant. And like that is going to happen on a two to five year time horizon. So like it's Rishi: Yeah. Rick: and now don't be risky. Don't like YOLO, don't like, you know, have a data breach, don't do this crazy stuff. But like if you're not at least at the middle tier, like, you know. It yeah, fast enough where you're not gonna get like left behind, but slow enough that you're not taking a a risk. And not a lot of people are not comfortable with finding that that sweet spot. They're still kind of like, I can just catch up, and it's like maybe. I don't I don't think people Kendall: Hm. Rick: realize like how how much the paradigm is changing underneath them and they don't see it yet, but one day they'll you're gonna have a full lockdown and they're gonna be like, shit, I'm not in the game anymore. And that's gonna be the biggest risk. Rishi: So so you're Kendall: Hm. Rishi: you're saying kind of going forward, at least for the foreseeable future at the moment, your hot take is that business models for most companies, whatever the subset that is, are gonna be reinvented or gonna have to be reinvented every three to six months. It's like COVID every every six months from a impact perspective. Rick: Yes. And and people adapted, right? We went from restaurants to figure out like food delivery and like some thrived, some didn't, right? But but and and and the ones that thrived did really well. Some of them actually may have blown up in popularity because they were they people discovered a great restaurant that was just down the road and now it's the only game in town and and that The winners could also then become the losers in the next cycle, right? Something like I'm I'm on top of the heap right now, and now all of a sudden three months later something else changes and now you're back at the bottom. And like that's that yo-yoing is not comfortable for a lot of people. A lot of people are used to like careers of four to five years before they change companies and or they you know, maybe ten years in a given role. I also think people are need to figure out the changing roles rapidly. This this sort of CX role I'm was in before is I gonna I think it's gonna be the norm. Like product people are gonna do more engineering, engineers are gonna like become more product or design because they can, and we're gonna see this more amorphous shape shifting among roles almost inevitably. Kendall: Yeah. Yeah. I agree with that. Rishi: Yeah, I it's it's funny, like funny's not the right word. It's interesting because that it that's that is a hot take and yet like I'm kind of just sitting here nodding along. Like, yeah, I think I think I think you're right. I mean, I've talked to a number of companies recently who have raised significant VC money. They had a great business model and in the last like thirty six months they're like, you know what? Our product can be replaced by a small number of tokens. And they have to reinvent themselves. So I I think you've spotted on there. I actually think You know, maybe my hot take on this is a little bit tangential, but maybe it's the death of VC. Or, you know, VC has to do the same thing, right? That business model of cool, we're gonna find we're gonna pick one winner out of a hundred, but they're gonna be massively outsized, then they may have to shift because you're not gonna be able to grow a company big enough that fast, for most kind of funds to to to win Rick: You're see Rishi: there. Rick: you're s you're seeing it VCs are in the business of taking risks and bets, but when the context like they'll usually develop a thesis, generally. Some VCs won't. They'll just yellow money into whatever, but they generally have a thesis, like in the next three to five years. Rishi: Wait, if you know those VCs that are just yellowing money, you connect them with me. Rick: I well absolutely I'll send you a DM. you know, but you know, I I've heard many people say, like, we have to sit in our hands. There's a thousand applicants of an AI this and like we like I can't pick a winner. Like I I wanna like I want to get the category winner, and there's a hundred people with the same idea. They're all swarming at the AI for X plays, right? And and so like a lot of VCs are sitting in their hands because like I can't pick winner so I'm just gonna go throw it at anthropic. So like there's like concentration of bets at the already winning scaling things and less bets which is killing the ecosystem and of sorts because you know you're not having this sort of growth funnel of like new seed stage things that are gonna bubble up. Or that path is like now known as like your best bet is like build something with a flash and then get acquired by the big guys. And like the the the the space is very much changing and it's just hard to stay afloat of that. Kendall: Yeah. It's I I completely agree with that. I think everything's changed, from investment to executives to day to day developers. and and I for the time's sake, I'm gonna wrap it up a little bit here, but I think I think even the consultant for a senior person like you The role has changed. It used to be, I'm gonna come in and I'm gonna help you know where to go. And today it's I'm gonna come in, I'm gonna advise you on where to go, and I'm gonna build some of it as I'm going, because I can solve your problem faster than telling you how to solve your own problem because I can literally go build it. I can literally fix the problem Rishi: Mm-hmm. Kendall: while you're talking. That engineer over there who told me it's gonna take six weeks, I'm gonna show him it's gonna take twenty minutes, and I'm Rick: Yes. Kendall: just gonna go do it. Rick: And I've done that. there's a fire crawl Kendall: Yeah. Rick: release a library on Friday. It was a it was a extract any file into Markdown and by Tuesday add a prototype which is now the the client loved so much and it went from eight hours of human work that took it took eight hours to run this whole document, you know, by hand, to AI can kind of do it with ninety-five percent accuracy in like ten minutes. to I wrote a deterministic code extraction and comparison thing. It wrote that same document in 500 milliseconds with like 99. And I was able to do it in three business or two business days, right, in a weekend. and by the way with me, it was clawed, right? So like the ability to go from idea to like a wow factor, like holy crap. that's possible and I want that and like let's get this now in our team has collapsed. And so having the f the ability to see that, the ability to do it, and then the ability to like help map how how we're gonna roll this in the team, the culture, the training, the the things, is that critical tr triangle, I think, to to succeed here. Kendall: Yeah. Yeah, well and and good good luck out there, Rick. it's a it's a fascinating, messy world. And just just so that everybody who's listening to this knows, the entire world feels left behind by AI. Everybody's sitting around thinking, Holy shit, how do I keep up? And we're probably ahead of that curve by a substantial margin. And in the CTO Lunches Plus group. Rick is making everybody else feel that way. So you just so Rishi: Yeah. Kendall: you know, Rick is on the bleeding edge of the cutting edge of the edge of whatever the hell that is. So don't listen to this and panic. he's making everyone around him panic all the time. That's part of his charm. but we appreciate you coming and sharing with us, Rick. Rishi: Thank you very much for the time. Rick: I appreciate you having me. Rishi: This was an excellent conversation. j just as Kendall said, every time I chat with you, I learn something new. and I very much appreciate that about you. So thank you for coming on.