Why the 7th Biggest Player in a Huge Market Beats the Winner in a Small One

Sean Weisbrot: She brought her corporate nonsense and ruined the system we had. Took six months and wasted 20 or $30,000.

Rob May: You have to network in advance of job openings to identify and have great talent ready. Finally interviewed this guy, and he goes, "Oh, good question. How does your buyer buy?"

Sean Weisbrot: Claude has access to every word from every interview I've ever had.

Rob May: The best way to be successful as an entrepreneur is to-

Sean Weisbrot: What's the most valuable thing that you think about in terms of networking that founders should be thinking about that they may not be thinking about right now?

Rob May: I think you have to network in advance of job openings to identify and have great talent ready, because otherwise you're gonna, you know, you're gonna, you're gonna open a job listing, and you're gonna get a bunch of applications, and it's gonna take some time. And, and I think as an entrepreneur, and also I expect this of the executives that work for me, that they should constantly be cultivating talent, you know, a couple times a month, like, you know, going to events or taking people out to dinner or having coffees with talented up-and-comers so that when I ask you, "Hey, if I gave you, you know, the budget to hire two more people tomorrow, who would they be?" I think a good executive should know that off the top of their head, at least who they would go after and who they could call immediately that's sort of like, you know, clo-clo-close enough to them to know what they're doing and, and all that and, and could make a reasonably fast decision. And I, I... So I think your good executives network to, um, y-you know, to have the, the capability to bring in the best talent, and they're constantly thinking about, you, you know, when they meet people, making mental bookmarks of who's good at what and why and who they would hire if they had openings.

Sean Weisbrot: So that's amazing advice, and it's something I didn't really think about when I was starting my first tech company. Let's assume that you're not, like, a well-funded company, you know, whether you're a bootstrap company or you're a VC firm or, you know, you're a Series A company where you've got a bunch of money that you can start to look to hire people. What if you're much earlier in the game? How can you do that if you don't know when the money's gonna be available to hire that next person?

Rob May: Yeah. It's, it's a good question. I... You know, I think, um, one of the things you have to think about a-as startup founders too is, like, um, a lot of people think it's risky, and the best way to de-risk it is for them to see your progress. And so, you know, for example, if you reach out to somebody that you don't know or who maybe is a friend of a friend, and you just wanna kinda get to know them, and you're like, "Hey, I just wanna..." You know, maybe you talk to them about, like, "I'm building an advisory board, and I, I'm, I'm looking for contacts. I just wanna talk to you." And, and what you can see is most people who work at big companies will definitely be like, "Oh, you're three people starting a company. No way I'm coming to work for you or with you." But if they like your idea and they like you, there's a good chance they'll meet you every four months for coffee, right? And, and so if you can do that and nurture that relationship, and over a year and a half they see the process of you, uh, you know, making progress, raising a little money, making another hire, winning that first customer, it gives them a lot of that, that traction and momentum, slow as it may be in the early days, gives them some belief that you're onto something and, and can help. And so, you know, I, I think at that phase of the company, I think you're just casting a wide net and trying to figure out people that you think, uh, will share your vision and, um, and could be potential recruits.

Sean Weisbrot: And so when you're looking at these people to hire, what I've heard was, but I didn't do, uh, when I was running the company, was look for people... Uh, like, so I was looking at people that I thought would be a good fit, but what I've heard is a better way to do it is look for the person who's already done the thing that you wanna do. So if you need a marketing person who's taken a company from launch to, to a million ARR, you should find someone that's done that to help you to do it. Rather than look for someone that's done 10 million to 100 million when you're zero because they just may not know how to start something.

Rob May: Yeah. That's great advice. Uh, s- just sometimes you don't have access to those people or there's not enough in the market that you're in. So that, that is ideal, uh, if you can find those people. Um, but, you know, sometimes you can't, and you just have to give somebody who's really intelligent, like, a shot at it.

Sean Weisbrot: Yeah, I tried that as well, and that, like, it doesn't always work. Didn't go

Rob May: either way.

Sean Weisbrot: Yeah. I... Like, I had hired a woman who was the QR manager. Sorry, the, the, the quality insurance manager. And she had 20 engineers working with her doing quality assurance testing on the software product, and we had two, and we wanted to hire her because we figured she would help us to build something that was scalable. And she brought her corporate nonsense and ruined the system we had. When we hired her with a specific task to build something different that... Because what we had was working, but it wasn't good enough. We wanted something- Yeah ... better that could then accommodate post, post-fund, uh, post-fundraising to hiring more people and, and have it, you know, more scalable. And she didn't do anything that we asked, and she took the way that the corporation did it, which is completely opposite of what we wanted, 'cause they were, like, using Excel and we were trying to get onto Jira. From Excel. And so she, like, took us backwards and, oh, it was a freaking mess, and it took six months and wasted 20 or $30,000. Um, it's just stuff like that as a, a first-time founder you don't know is gonna happen, and you can't afford to let happen. Do you have any sorts of things like that happen to the companies you, you're working with?

Rob May: Yeah. I think one of the things you have to learn, and that you have to remember, particularly when you hire executives, is these people are very senior. They know what to say in an interview to get hired. Uh, whether they're actually good or not is much more difficult to tease out. And, um, you know, a lot of people that have climbed really high in their career are just great imitators. They just take a playbook that they saw somewhere, they don't always fully understand it, they don't always know how to change it for changing circumstances, but they know how to implement it. And if they get lucky and they go implement it in a couple places where it works, then, you know, they think they're brilliant. Um, I'll give you an example. At, at one of my startups, um, I went through three heads of sales, and I was like, "Okay," I just like, I can't find the right person. 806 And everybody when you would ask them, you know, we, we had sort of like two or three salespeople, and I'm like, "Okay, I've gotta go to 10 salespeople, how would you parse out leads?" And they would typically give you an answer like, "Well, my last company blah, blah, blah, blah, blah," and this was how they wanted to do it And I finally interviewed this guy, and he goes, "Oh, good question. How does your buyer buy?" And I'm like, "What do you mean?" And he went between, he's like, "Well," we were selling an IT product, and he's like, "Well, uh, if it's a highly referenceable sale and your buyers are likely to know or talk to each other, then you wanna split the reps up by geo, because then when you call into Chicago and you're talking to this customer, you can be like, 'Oh, do you know Fred over at this other place? 810 Oh, you do. Yeah, he's... You go to the IT meetups. Yeah, he's a customer, right? You should ask him about us,' or whatever." He's like, um, he's like, "Or if you're, if the needs of the buyer are different for, like, SMB and enterprise," 'cause we had a horizontal product, he's like, "then you shift it out by, um, you know, by, by those groups so that you can have SMB reps who learn the value prop for that and the enterprise reps who, you know, learn the value prop for that, and you do that." And he's like, um, so, so that is maybe you do geo, right? And then, and then, and he walked me through, like, three or four cases of, like, oh, if your buyer buys this way, or like, like if, you know, if you have price-sensitive buyers and non-price-sensitive buyers, then you might, you might do it this way. Like, if other ways you would round robin them and whatever. But he walked me through, like, here's how you structure a sales org to match the things you've learned about how your customer buys, and I was like, "Nobody's ever told me this before." Yeah. And I just realized that he understood sales at a level that other people didn't, right? It's like, it's like you have people that are like, "I'm a channel salesperson, channel sales manager," or, "I'm a direct sales manager." 818 Not many people can tell you looking at your product and how your buyer buys, like, "Oh, you should do this one or that one," right? 819 People tend to either, like, they're, like, philosophically opposed to channel sales, or they love channel sales. 820 I mean, this hap- this happened with the company that bought my company. 821 They were very channel sales centric I didn't think it would work, work well for our product, and they tried to move us over, and after a year they had to switch back to direct. 822 Mm. But, like, their philosophy was channel. Right? And it... But it, it, it matches the buying behavior, and it matches the, the product characteristics, and that's how you have to think about it, and just very few people, regardless of seniority, have the kind of thinking pattern that allows them... 823 Like, they're not actually problem solvers. They're not actually able to match business tactics to the market situation that they find themselves in. And that, that's a really hard thing to find. 824

Sean Weisbrot: Is this person making content on YouTube? 'Cause I think that would be really valuable. 825 It's, it reminds me- ... of, like, Alex Hormozi almost. 826

Rob May: I don't know who that is, but, um, yeah, this guy's not making content on YouTube, 827

Sean Weisbrot: no. So Alex Hormozi is a guy who has, uh... He's k- kind of, like, developed a portfolio that's like private equity, and he started off with a gym company he built and sold, uh, like 50% of it or 60% of it for, like, $50 million. 828 And then he started, like, consulting and advising other companies and investing in other companies and, and now his portfolio I think does, like, 250 million a year in, in revenue combined, and he has several million followers and he is... 829 Basically his goal is, "When I make this content, it's to help you reach $3 million a year, a year in revenue so that you're at a size that I'm gonna be interested in wanting to invest at." 830

Rob May: Yeah. 831

Sean Weisbrot: So basically the content is free. H- I want you to learn. 832 So he does a lot of stuff, um, and he, he had a, a book launch that they did, like, 100 million in sales in the week- over the weekend. 833 Wow. Over the first weekend. Yeah. Um, definitely helped by the fact that, you know, he has millions of subscribers. 834 But he doesn't... He's not doing it for the money. 835 He's doing it because he knows if he gives it to you for free, you're not gonna value it. 836

Rob May: Yeah. 837

Sean Weisbrot: So but yeah, really, really smart guy. I love watching his content. 838 He does a lot of videos where he's talking, uh, with entrepreneurs and, like, helping to find, like, one thing that's preventing their business from, from, like, growing. 839 Um, and, like, sometimes that one fix could add a million or two million a year in revenue to, to it. 840

Rob May: Yeah. There's a perfect example, right? One of the things I've learned having built companies that sold to, like, sales and marketing people and companies that sold to IT people is, like, your comment on, we can't give things away for free for people don't value it. 841 Um, very true in the marketing and sales world, very not true in the IT world. 842 Like, when you're selling to IT people, the best thing to do is actually give your core value prop away for free, open source it, then they... 'Cause they- they're so reactive to, like, "You're trying to scam me. You're trying to sell me something I don't wanna be sold to." 843 I- tech people don't like that. But if you can give them for something for free, and then they come and they complain about it, and they go, "Oh, it doesn't have good analytics," whatever. 844 You go, "Oh, that's 50 bucks a month," right? Or whatever. 845 So those, like, freemium and open source models work really well there, but, like, you would never do that for marketing and sales tools because if you give people away for free, they, they... 846 it's very similar. They're like, "Well, this is crap," right? 847 I, I even do that, like, when I, when I put on events and conferences. 848 You don't charge the tech people Um, generally, but like sales and marketing stuff you do, because then they'll, they'll buy the ticket. 849 They'll feel like it's valuable, they'll show up. 850 Um, so it's, it's, you know, it's, again, it's one of those different mindsets that you have to learn for your different markets. 851

Sean Weisbrot: For sure. So I wanna shift the conversation to inference costs, because I've been seeing a lot of people talk on LinkedIn about how, like, investors are basically funding token spend. 852

Rob May: Yeah. 853

Sean Weisbrot: And so if a company is trying to raise money from investors, and one of their largest spends are token usage, how can companies kind of cut that cost without losing the value that it's bringing to them in order to allow the company to run? 854

Rob May: It's a great question, and there's-- we're seeing more and more companies, because we, we play tangentially in this space and see a lot more companies that, uh, actually have that as a line item, right? 855 It's like trying to reduce LLM COGS for the year. 856 And, um, what's interesting about it, there's a lot of ways to do it, right? 857 There's simple ways like moving hoster- hosting providers, right? Depending on the models that you use, um, and what you're doing, uh, you know, moving to a different, uh, either, you know, frontier lab for your use cases, different types of models within a frontier lab. 858 Um, if you're hosting your own models, moving between fireworks and together and whatever they're optimized for, like, those can do things. 859 Um, you could opti- also optimize your model usage. This is what we help with. 860 So sometimes, excuse me, sometimes figuring out that, well, you don't need Opus Four six to summarize an email. 861 So if you can build out your infrastructure so that the simple tasks go to small language models or open source language models, maybe that you self-host. 862 Um, and then your, your frontier tasks go to frontier models. 863 So the complex stuff, the orchestration stuff, go to these more complicated models. That's, um, that's possible. 864 Uh, and then, and then sometimes if you just rewrite your prompts, you can be a lot more efficient with token usage. 865 Uh, a lot of prompts, there's a tool, I'm, I'm blanking on the name of it, but when you, when you do a lot of automated coding, like with Cloud Code, it passes a lot of context on every prompt, and this tool just strips out all like the s- terminal system gibberish and passes the prompt on, and it's like, you know, it saves you twenty percent on token costs, um, just to do that. 866 So there's a lot of... The- there's a lot of ways to do it. 867 Um, I think it's gonna get worse before it gets better because my guess is as OpenAI and Anthropic IPO, Wall Street's gonna scrutinize their unit economics, and it's probably gonna ask them to raise prices, push them to raise prices, uh, on inference. 868 But I, I think you're starting to see some counter trends that we just saw an article in late February from AT&T, where AT&T was doing eight billion tokens per month through their core AI systems, and it was costing them too much money, and they rewrote their entire architecture to take advantage of small task-specific language models and, um, were able to cut those costs by ninety percent. 869 So I, uh, yeah, I, I think it's gonna be-- I think it's an emerging issue. 870 It's not a problem for a lot of people yet. 871 But depending on the size of your company, somewhere when you hit that sort of like twenty to fifty thousand dollars per month in inference spend, and you realize it's growing at ten, twelve, twenty percent per year, whatever it is, that's when you start to go, "Oh, we should do something about this." 872

Sean Weisbrot: Yeah. So I, I have, uh, Cursor I pay twenty dollars a month for. 873 But this month I spent a hundred and thirty. Yeah. 874 I just-- I went over because I, I decided to build a lead magnet and an e-book that I wanna sell, that, you know, as an upsell. 875 And I-- at first, I used the model... So b- let me just take a step back. 876 I've created a system within my website, and I've created my entire website inside of Cursor and Claude. 877 Um, before I used WordPress for like twelve or thirteen years. 878 But- Yeah ... what I decided to do was, every time I have a new interview go live on YouTube, it creates a new blog post. 879 And so it gives us its own page, pulls in the video embed, pulls in the title and, and the description, and the chapters, and, and everything. 880 But also, I then use Gemini to create a transcript, and I pull the text of the transcript into the page. 881 And in doing so, Claude has access to every word from every interview I've ever had. Mm. 882 And so I was able to say, "Hey, look, Claude, I've done three hundred interviews now. Let's make an e-book about three hundred lessons that I've learned from talking to three hundred founders." 883 And Claude's like, "Great." And it spent like ten minutes, and it, it created an e-book for me. 884 And I didn't really look into the details of it yet. 885 Obviously, I'm not stupid enough to know that I can just publish it. 886 I have like a lot to work, a lot of work to do inside. 887 But I was like, "Hey, uh, I probably need like a lead magnet for this, something that will get people to sign up for the newsletter that later on I can then push them to this other thing. 888 So like, let's also make a shorter one. 889 Let's make it six lessons." And I go, "Pull out everything related to networking from the original book, and let's make it this other thing," and it makes this other thing. And so I go through and I, I start editing it, and I realize I have to rewrite it all. And of course, I, I'm probably more proficient than a lot of other people. 890 I can actually go into the, the file inside of the repository, inside of Cursor, and make changes to a lot of the things without having to use the prompt, so I'm able to save on inference costs there. 891 Um, but what took a lot of time and energy from the AI was I originally was gonna release it as a, a Word doc. 892 And then I was like, "Ah, this thing is ugly." 893 Like, I can't- I, I don't feel good even though it's free, releasing it without any designs or anything like that. 894 Like, there's just more I have to do. So I sent the doc back to Claude and I said, "Hey, turn this into an HTML file with CSS, and let's turn this into a PDF that we can generate and have the brand, the colors embedded into this entire thing, and let's make some designs on some of these things." 895 And so the-- a lot of my cost this month was on all of this because it-- I'm using, uh, Sonnet 4.6. 896 Maybe I'm using the wrong... May-you know, maybe I should be using 4.5 or 4.0, I don't know. 897 Um, but a lot of my energy has gone into that and, and so, um, it's produced incredible, wonderful results, but it has to, like, create the, the script, and they're-- part of the script, like, wants to generate the PDF every time, so I have to stop it so it doesn't generate a PDF and waste more tokens. 898 So like- Yeah ... I'm thinking about how to keep the token cost down. 899 Um, so I guess in very basically, what do you think the model I should be using for something like that? 900

Rob May: Yeah. I mean, at the, at the lower end, sometimes it's harder to figure out if you don't have enough scale on your inference to measure stuff. 901 But the, um, you know, sort of what we do for enterprises is, uh, we just have a drop-in where whatever you're using to call your, you know, Claude or OpenAI or whatever, um, you call us instead. 902 We still call your model, but we siphon off some of that traffic, and we run it through a bunch of other models. 903 We add various configurations, like changing the temperature, which is a variable of how the model runs, um, adding some thinking algorithms, whi-which the industry calls, like, test time compute tactics on top of it. 904 So this might be like adding chain of thought or something like that. 905 And then we look at the net system impact of that, because sometimes when you try to use, like, one of these reasoning models where you add chain of thought or, or, or one of these tactics, it'll actually use more tokens. 906 About twenty percent of the time, it'll be more expensive. 907 Um, and, uh, anyway, so we sort of, you know, cha... You know, tw-tweak all those knobs. 908 I think, um, I think what's gonna happen to the industry is I think you're gonna start to see people think about tasks more than models. 909 And you're gonna-- when you have tasks that are done, you're gonna have agentic workflows, and the agents can call tasks, and those tasks will hit varying levels of sophisticated task models that just say, like, "This model is really good at making e-book designs." 910 Um, because if you think about it, there's a base level of knowledge that you have to have. 911 So, so compare it to humans. There's a base level of education you have to understand to even be able to interpret a command. 912 But beyond that, like, you know, we, we talk about these smaller models, but, like, why do you need a person that's got seven PhDs when you're like, "I need you to do an e-book design"? 913 You're like, "Your astrophysics PhD is just wasted knowledge that I'm paying for." 914 And the models are the same way. Like, you have this model where you're like, "Oh, it can solve You know, it can develop new cancer drugs. You're like, "That's amazing." I just need it to reconcile my bank statements, right? Like, so you're gonna pay for these big models that run slow that have all this knowledge in them, and what you really need to figure out is, what's the base level of knowledge I need to get this task done? 915 Um, and, uh, and I, I shouldn't pay for more than that, right? 916 Um, you don't need to pay a rocket scientist's salary for somebody who's doing a bookkeeping task. 917 And I think you're gonna see the industry move to this. 918 I think the r- the only... W-we're seeing the early stages of it. 919 The only reason it's not bigger yet, well, there's two. 920 I- the, the industry has been, first of all, people getting started with AI, and they're trying to figure out what's capable, what it can do. 921 And when you're getting started, you don't wanna try to optimize your models 'cause you don't really know what you're gonna be doing, right? 922 Um, so you don't wanna be like, "I'm gonna use these eight different models out of the gate when I design my system." 923 Like, no, no, you start on, start on one big frontier model, prove your use case. 924 Once you do that, you're usually fine with that for a while, and then y- you start to see your inference spend go up, and then that's when you start to look for optimization tools. 925 Um, it's similar to how you, you design a code base, right? 926 Most companies start with a monolithic code base, everything in one repo and one giant thing, and then they start to parse off the pieces that need to be optimized and make them microservices, and I mean, it's just, that's, that's how businesses are gonna go with AI as well. 927

Sean Weisbrot: I, I think that's a lot more complicated than a lot of people would expect to hear. 928 I, I think, I, the, I'm just thinking, like, you said the word microservice. 929 I'm like, I didn't hear about microservices until my CTO was like, "Yeah, our, our stuff is way too monolithic, and, like, it's getting much bigger. Like, we need to break things off into microservice." 930 I'm like, "I don't know what you're talking about. 931 Just do it." 932

Rob May: And that's what needs to happen to the AI tools, right? 933 Which is the AI tools just need to take care of this for you. 934 They need to select the models automatically. They need to route this for you automatically. 935 You know, there's, I, I think you're gonna see... Y-you know, we have a more enterprisey tool, but I think you're gonna see tools like this for consumer usage and everything else that are just like, um, you know, we, we, we optimize your, your coding token usage. 936 Um, I mean, if you can say, you know, there's millions of people that code with these tools. 937 I mean, if you can save everybody, you know, ten bucks a month on that, it's a lot of money on the table. 938

Sean Weisbrot: Ten thousand dollars a month. 939

Rob May: No, e- t-ten bucks a month per person, right? Okay, okay. 940 And then there's, you know, you can, you can, you can do that for millions of people. 941

Sean Weisbrot: Okay. Well, 'cause you had said, you, you were talking before, like, if someone was spending fifty thousand a month, so I, I thought when you- 942

Rob May: Well, but, but a lot of individual developers who are spending hundreds of dollars a month would rather spend tens of dollars per month, right? 943 It's, um- Sure. Yeah. 944

Sean Weisbrot: Yeah. 945 'Cause a lot of people that are using these tools, even though they're not really talked about, they're people that wanna start a business, and they don't have funding, and they may not have any, you know, savings to work from. 946 And so they might just be spending from their credit card, but they've never known money before, and so they're scared, right? 947 The, I think the thing that's really great about these kinds of tools is that they The word democratize is probably overused, but they make it something that, you know, a guy in the Philippines can compete with a guy in Silicon Valley because AI allows the cost to be low enough that you can just do it. 948

Rob May: Yeah. And they also open up capabilities to people... Like, I don't know your technical background, but, like, I was an engineer for a while, and then I, you know, got into the management side of things and got to the point where I couldn't write code because there would be pieces I would get hung up on and couldn't solve, and, uh, a lot of times, like, IT pieces, right? 949 I don't have the right Python package installed or something like that. 950 And these tools hit a threshold where they can solve that for you now, right? 951 Which means I can write code again, um, I can build apps again because I know enough about it to do it, and the pieces I don't know, the tools can fill in for me. 952 Uh, so I, so I also think it's opening up a new... You know, and obviously there's higher level tools like Lovable and Vercel that'll let, let you build a, whole apps without knowing anything. 953 Um, but yeah, it's, it's a very cool time. 954

Sean Weisbrot: So I started off on Lovable, and then, like, a month or two after that, I moved to Cursor, and I still use Lovable, like, if I wanna just start a website, and then I'll just move it to Cursor and keep building it, 'cause Lovable can do it a lot faster than Cursor, and it c- it has the Veet package, like, just different packages, so I don't wanna bore people with the technical details. 955 But I don't have a technical background. 956 I've been fortunate that I learn whatever I need to learn pretty fast, and usually I'm self-taught. 957 Um, although I did learn a lot from my team for that last tech company. 958 And I think it's because of that experience with that company, working with the front end team and the back end team and the architecture, you know, my CTO, and the, the product manager and the quality assurance team, that I'm able to use these tools much more proficiently than anyone else, even though I never used them in the past. 959 Um, so yeah, I, I'm thankful for that knowledge and experience. 960 So, uh, yeah, I, I know a lot of people don't know how to vibe code to production because they just don't understand the architecture, and it's something that I've fortunately been, uh, spent many, many, many hours hitting my head against a wall as to why we can't do something right now. 961

Rob May: Yeah. 962

Sean Weisbrot: Or, or why it's gonna take a lot longer than I thought. Yeah. 963 But, but now AI makes it so that you can. 964 And what would I-- what my team hated was that I would change my mind sometimes and be like, "Oh, okay, but like, uh, what if we did it like that instead?" 965 And with AI, you don't have to worry about design, you don't have to worry about feature sets, 'cause you just go, "Hey, look, actually, that thing didn't work. 966 Let's revert that." And you can just do it locally, so you don't have to push something to, uh, uh, like a test, um, a test branch, right? So you could do everything locally, and then when you're ready, you can just push it And so you may do may have a bunch of changes happen, but you can test the thing you wanna change before you push it. 967 So like managing branches and managing features and making sure you don't push broken things to production are, are a lot of things I think a lot of founders aren't aware of. 968 So yeah, thankful for those experiences. 969

Rob May: Yeah. That's great. 970

Sean Weisbrot: So what's the most important thing that you've learned so far? 971 It seems like you've had a lot of experiences. 972

Rob May: Um, wow. Well, I think the best way to be successful as an entrepreneur is to cozy up to big markets. 973 And, um, and sometimes, you know, you don't know if a market's gonna be big or not because it's nascent, but you have to believe that it's gonna be really big. 974 Um, or you have to be doing something in a market that is really big, uh, you know, payments, insurance. 975 You know, I, I, I think inference is gonna be a really big market. 976 The reason you wanna do that is because when you're in a big market, you know, you can make a lot of mistakes and still come out on head. 977 You could be the seventh biggest player in the industry and still do really well, you know. 978 Um, whereas in, in, in, you know, in markets that are hundreds of millions of dollars or maybe a billion dollars, like a lot of times it's like one company takes most of it and everybody else is kind of struggles or, um, uh, you know, or, or, or things like that. 979 And so I, I think these really, really big markets are things I would encourage people to, to work on for that reason. 980

Sean Weisbrot: Thanks for watching. If you liked this insight, I've handpicked another video for you right here on the screen. 981 For more actionable strategies that get you real results, hit Subscribe. 982