Tyler (00:00.92)
We're live.

Rick (00:02.014)
wow, okay. hello. I saw no countdown again. This is bad. I feel like No, it didn't. well. Yeah. What's new with you?

Tyler (00:03.788)
It happened. Hi I'm leaving that in for sure. it didn't show you a countdown? Hmm. Riverside, get on your shit. what's going on? I'll answer. I'll answer. Here we go. I'm all all about health insurance right now. Or or rather all about finding health insurance agents for lessoning CRM as customers.

Rick (00:17.835)
No, no, I'll I'll go. No.

Rick (00:28.748)
That's so awesome.

Tyler (00:30.466)
Yeah. So yeah, we talked about this last episode. I I think I basically shared the the high level kind of we've we looked at all the industries, decided health insurance is a good one to go after, and we're starting to put a marketing campaign together. in that episode I said that we wanted to go to this conference of kind of independent health insurance agents and they ghosted me. Turns out their email just went to spam. They were not ghosting me. So I got in touch with them.

We had a great call. We are now a gold sponsor of Health of Palooza twenty twenty six. So we'll be going to the Twin Cities in the at the end of July to schmooze with some health insurance agents.

Rick (01:11.426)
That is awesome. I wanna come.

Tyler (01:13.838)
Do you do you do conferences? Well, I'm curious if you go to conferences in general, but especially like as a vendor versus an attendee.

Rick (01:22.07)
Leg up health, no. We've we've done a couple of like just trying to do it. but at at my at Winfall, we do it all the time. we're at particularly in I've gone to a couple, yeah. Yeah. It's we usually typically do like a sponsorship plus booth. and there is oftentimes speaking engagements tied to those sponsorships. and so some combination of like set sellers, customers facing people, customer service people, and marketing slash like

Tyler (01:30.198)
Okay. Do you personally go? Yeah.

Rick (01:52.056)
Thought leadership is represented at at all events. Yeah. Are you who's going for you?

Tyler (01:55.684)
Okay. Cool. Yeah, we're we're gonna go we're going with three people. I'm gonna be one, Eunice, the head of marketing, and then Ruth, who's a CRM coach, but they're kind of I would say like if anyone's going to step into a sales biz development type role, like schmoozing with people, very like charismatic, social outgoing type stuff of the CRM coaches, like Ruth is the one with that interest. So kind of bringing let's think of Ruth as a salesperson, Eunice as a marketing person and me as a

just there to I don't really know why I'm there, but to to buy into it I guess.

Rick (02:30.988)
Yeah, I'm trying to I'm I'm going to your team page right now to make sure I remember who Ruth is from a face perspective because I think I know who I was thinking of Ellie. I don't think I've met Ruth.

Tyler (02:41.563)
yeah, Ruth's one of the newer CRM coaches, but n new for us still means been here a few years. but yeah, it's gonna be it'll be a whole new thing. Like or we went to conferences in the mid 2010s, like 2016 ish. We were going to franchise conferences, try trying to do the same type of playbook, but I do think we have a much better sense of the business and where where our position is and all that.

Rick (02:44.75)
Yeah.

Tyler (03:10.189)
So, you know, it could still fail. Any all all marketing experiments I think are most likely to fail. But I I'm pretty excited that that we're we're giving this a much more legit shot than the ones we did a decade ago.

Rick (03:21.026)
This is awesome. I can't wait to see the results. and I wish I could come to the conference with you and help you sell.

Tyler (03:27.957)
Yeah. Late July. We got a ticket for you if you want it. We got eight tickets for with the sponsorship. It's in Minneapolis. Mm hmm. You're way too busy. You're not gonna come, but I would love it if you did. That'd be awesome. I'd I'd extend it a couple of days if you wanted to just hang out in Minneapolis too.

Rick (03:30.539)
Okay. Where is it?

Minneapolis.

Well let's actually consider that, 'cause that would be fun.

Rick (03:45.922)
Let's do it. Let's this is the reason to see see each other. Okay, let's play them around this. I think this is perfect.

Tyler (03:48.952)
Yeah, okay. I mean great great. Okay, cool. so yeah, we can talk offline more about that. But yeah, I in addition to the excitement about like what impact this might have on the business, it's making me realize a lot of the work like being sixteen years into the business and having a good plan that has not changed a lot. Like we're not like we used to jump around back and forth between different strategies and now we're much more like stable. I didn't appreciate

Rick (04:16.206)
Mm.

Tyler (04:19.287)
The work itself changes when you're in that mode where you're just kind of iterating and maintaining and there's still interesting stuff to do, but one of the things that this is like I I just designed a bunch of banner, like we needed a tablecloth, we need a backdrop, we needed a retractable banner for our booth. So I designed those over the last, you know, week trying to get HIPAA compliant before this happens, because all our banners say we're HIPAA compliant and so we need to actually be. it's just like a a type of like urgency and

New projects that can actually be completed as opposed to this one decade long project that's just going on and on and on. It it's definitely a different type of excitement.

Rick (04:58.488)
That's awesome. I'm excited for you. And I'm gonna come.

Tyler (05:00.265)
Yeah. Thank you. Great. Great. I I'd love to see that. It is funny to me that we're we're selling to health insurance agents and then you and your business, which I work with, is at its core a health insurance brokerage. And and yet there's there's no connection between these two. Like it's just a total coincidence that that it worked out that way. yeah, I've got more to say on health insurance stuff, but what's going on with you?

Rick (05:19.223)
Nope.

Rick (05:29.076)
I the main thing that I'm the two the two threads I'm thinking about are both AI related. One is personal AI use cases, and the other one is more around my day job at Windfall around like how do we scale organizational context for AI use cases. I think it's much easier to unlock AI for personal workflows. It's much harder, maybe outside of coding. coding being the the one area where maybe this is more unlocked.

to unlock AI use cases systematically at an org level. and so that's what I'm thinking a lot about. my personal website's live and it's doing better. It's like already ranking better than my old website. So whatever Squarespace was doing was doing a disservice to me relative to what Vercel is doing. so yay, that's a win. I have not put pushed new content yet to the site since I put it live. So I'll be publishing my first articles most likely this weekend.

Tyler (06:16.343)
Nice. Yeah.

Rick (06:25.826)
So that'll be an interesting sort of exercise in whether or not this works. effectively we'll tell Claude to publish tech a text file, a markdown file.

Tyler (06:35.329)
Yeah. And and hope that it doesn't also edit all your other files without you asking.

Rick (06:39.49)
Yep. Yeah, exactly. Exactly. So I th I I think I trust it. but we'll see. and then so that's the kind of my personal journey. And then on the on the organizational side, I I I wanted to open this up for a little bit of a topic, see how you're doing it at Less Annoying CRM. But I think one of the challenges is that if you're an organization who took maybe for granted how important it is to document context for humans.

Tyler (06:45.133)
Mm-hmm.

Rick (07:09.024)
you're really behind the eight ball in terms or behind the behind the the pack in terms of you know being able to provide context to AI. Because if a human doesn't have the context in documentation, AI needs it even more explicit. and so anyway, I'm I'm I'm just realizing like the thing that was always important is now more important. and so the if you have a gap, you got to close it. And then second,

How do you keep this stuff up to date? Like let's say you close the gap. How do you keep context from decaying? it takes a whole like company working towards that as a primary goal, so that the AI has what it needs to to survive. and I'm just like struggling. I'm well, I'm just saying I'm struggling. I'm just like thinking a lot about how to how to operationalize that.

Tyler (07:39.308)
Yeah.

Tyler (07:56.632)
Yeah, it's a challenge. Yeah. I mean, this is why you said you should have been doing this the whole time anyway. I want to challenge that a bit, which is to say if it's if you're a small company, I don't know if that's true. A big company I I I've heard this pattern related to AI in multiple different contexts that a big company already had to operate like AI was there because like you don't know everybody, you don't

The the CEO can't just be like, I know them, they're trustworthy, or whatever. You just you have to document stuff, you have to communicate in different ways. Everything has to work like a bunch of separate machines that connect together through something like you can think of it as an API, let's call it. that's a bit too robotic of a way to think about all these humans working together. But at a small company, one of the advantages a small company has is I know everybody. I hold everything in my head. How do I know Michael's a good manager? Because I watch him.

How do I know Robert's a good manager? Because I watch him. And I think in that context, a lot of policies or r rather documentation can be a liability for exactly the reason you said. You spend all this time writing stuff down, which does take meaningful time, and then no one knows how to find it is problem number one. And problem number two is by the time they find it, it's not correct anymore. Yeah. I have thoughts on this with AI, but like

Rick (09:12.919)
It's out of date, yeah.

Ki ki before we go to the AI piece, like can we talk about what you're just I think you're hitting on a bit so so I think you're making me feel better that you know a company with fewer than one hundred employees, any company out there probably should not have invested, over invested in documentation prior to AI. Like because of what you just described. Yeah.

Tyler (09:18.712)
Yeah.

Tyler (09:34.147)
That's my opinion, yeah. Yeah, just ask somebody, you need it you have a question? That's the person you talk to, go talk to them.

Rick (09:40.803)
That there it there there's an assumption there that there's low turnover though. So I think there's a variable where if like tribal knowledge is fleeting, then that is not true. so if you have a high turnover business where you can't count on the knowledge being in people's heads consistently, then document no matter your size. yeah.

Tyler (09:45.132)
Yeah, true.

Tyler (09:58.776)
Yeah. Probably high growth too. Even if you have good retention but you're adding a lot of new people, in general, the more people you're adding, it's like you you it's worth systematizing the process of bringing them on and getting them up to speed.

Rick (10:12.29)
Yeah, there's some sort of change quotient. Like it's like change of like people, change of like activities, like that kind of thing. Like number of activities, something something like that. but but and the pip you know, the number of people is another sort of variable. but like I think I buy that in general. Like you probably like there's there's a scale, right? and that that makes me feel better about maybe where where s an organization like Windfall is. what I'm what I'm wondering though is like when is it

Tyler (10:18.658)
Yeah.

Rick (10:41.442)
How do you justify the cost to close the gap for a purposes of AI? And maybe there is no justification. It just is what it is. You have to do it.

Tyler (10:52.023)
Yeah, I I I think if it's the one hesitation I have is is it actually gonna live up to the hype? And I don't mean the whole AI world in general. Like I'm I believe AI is a important tool. Like there's no question for coding and stuff, like you said. But with any individual project, we've thought about this internally at less annoying of like rather than thinking AI thinking of AI as this overarching, it can do everything, we have to automate everything away. It can do some things and it can't do others and

It's worth it if it delivers, right? And then it's not worth it if it doesn't. So I think that's the before investing heavily, I would just wanna build up some high confidence that AI can actually do the thing you want. So let me give you example here. AI for customer service for lesson wing serum. We're never getting rid of humans. It's a huge part of our value prop, but for a customer who just wants an answer right away, they don't want to get on the phone for 30 minutes. Instead of them having to search our help docs and read through stuff and

Find the right article, being able to talk to AI would be a big improvement for them. And that would help the business a lot. I think we could invest heavily in making this work, but I have major doubts. Or sorry, let me we could invest heavily in making the best AI support possible. Would it be good enough? I have major doubts about that. And so we're trying to validate that by we're we're testing out Finn, which unfortunately just got acquired by Salesforce. I don't if you saw that. Yeah, Intercom. They rebranded to Finn.

Rick (12:15.788)
I did not see that. That's intercom, right?

Tyler (12:20.813)
So there is no intercom anymore. And they just got acquired by Salesforce. Yeah. Salesforce, they also got Slack, which we use. Yeah. But they're supposedly like the state of the art support tool. So we're trialing them out just to see like, can this even work? And if it does, then we're gonna decide maybe we wouldn't use them. But like then how do we how do we invest in, okay, we need to write all this documentation. We need to get it up to, you know, we need to keep it updated and all that.

Rick (12:25.057)
What?

Rick (12:28.536)
Three point six billion dollars. my goodness. Good for them.

Tyler (12:50.125)
I don't know, what it like so sorry, I'm talking a lot here, but can it even do it? Is the first question I'd be asking.

Rick (12:56.994)
Yeah, so like I think the number one use case I see for this internally is getting people the information they need when they need it internally. Like so mostly like I think particularly in a high growth business with that's trying to like conquer the world, there is so much information that that exists and that needs to there there's some that needs to be known by everyone, and then there's some that information that needs to be known by other people.

And th knowing what that is and getting the right answer when you need it internally is like the primary use case that I'm thinking of. everything else, like assuming you can do that, it unlocks all sorts of other use cases. Like, so let's say that it can answer a human's question you know in the right context at the right time with the right amount of information. Well, then you can automate the action that the human is is.

going to take with that information eventually. And so that's kind of the yeah.

Tyler (13:54.786)
Yeah. You could say to the AI, Go launch a new marketing campaign and it could be like, Well, at first I need to understand what's going on at the company and it c yeah, yeah.

Rick (14:02.708)
Exactly. So so like so so you know, kind of the the the the the tip of the spear to me is any human could get an accurate answer about the business, business context, from the AI. And then, you know, everything from there is like, okay, whatever the human was gonna do, the most simple stuff can be automated by the AI with a button, versus versus or or a prompt versus the human doing it. And then, you know, it it kind of builds from there.

Tyler (14:30.209)
And so these questions, I'm imagining two things come to mind. Maybe there's more. Like one is just internal policies and procedures, like a wiki basically. And the other is what would normally be in a project management system? Like what are we actually working on? what other types of things does when you say internal questions and answers, like is that is that the bulk of it or is there other stuff it needs to know about?

Rick (14:54.762)
I think I think I think there's like, you know, p policies and procedures, but there's also like, okay, I want to generate more opportunity. I'm a sales rep. I want to do my job. Let's just say let's let's just take any person. I want to do my job better. I like I'm a sales rep. I need to generate four opportunities per week. Let's just make this up in order to hit quota. AI, what should I do?

Tyler (15:24.179)
That's an internal question. I get

Rick (15:25.846)
It is an internal question. It's it's a what's my job and how do I accomplish it?

Tyler (15:30.755)
Okay, I get that in order to answer that question, the AI needs some context about what's going on at the company, but the question the AI asks to the repository of knowledge is more specific than how do I get four opportunities, right?

Rick (15:43.041)
Yeah, well it the the natural question is, okay, what's your target vertical? What what pod are you on? Like like what what accounts do you own? It starts going into okay, well, like what's the big what's the path to getting those four opportunities? which is w in in in go to market sales driven language, this is called the pathway to quota, which is generally a well understood playbook that is documented that you want sales reps to execute. so it would start then it would start.

you know, prescribing, well, you know, you should reach out to these types of companies. Would you like me to go find some companies? Like that that you start to get down there. So first question could be what companies should I go after? And it could say these types of companies. Well, what would be better is like here are companies you should go.

Tyler (16:25.475)
But sorry, what company should I go after is not a question querying internal knowledge base. The

Rick (16:30.614)
It is. It's a C it's our querying our CR CRM.

Tyler (16:33.963)
No no no, sorry. The it it needs to ask a different question to the CRM, which is what companies do we currently serve, what team are you on, whatever. The question what companies should I now go after is derived from the answers of those other questions. Do you know what I mean?

Rick (16:48.318)
that's assuming it's not already documented. There is assume generally in a go-to-market playbook, like what you're doing with insurance agents, you would document, we're going after insurance agents. and you know, the question is like, how do I how do I get more insurance agents? and there's some sort of strat like doc playbook that says this is the play this is the general strategy.

Tyler (16:55.875)
Mm-hmm.

Tyler (16:59.884)
Okay.

Tyler (17:08.429)
So what's the AI doing? If someone already wrote up a document saying here's the playbook and the sales rep says, Tell me what the playbook is, why didn't the sales rep just read the playbook?

Rick (17:16.472)
Sales reps don't read, man.

Tyler (17:18.541)
But the the AI is something else they have to read. I'm I'm I'm confused about what the AI is doing here to provide value.

Rick (17:24.82)
teaching the it's basically acting as a trainer.

Tyler (17:30.465)
Okay, that feels a lot more specific to me than like I need a way to get all the knowledge at the company to the right person at the right time.

Rick (17:36.472)
That's interesting. So okay, so that's that's interesting. So that's a use case for more specifically, which is like how do we train people?

Tyler (17:45.217)
Yeah, you just build an agent for like here's how you train a salesperson to do their job.

Rick (17:50.411)
Okay, that's that's a really good use case on top of that necessary context. Another context piece of context could be like help a sales rep find more accounts to prospect.

Tyler (18:01.729)
Okay. Yeah.

Rick (18:02.486)
Like that's another more more granular use case. but you could say

Tyler (18:05.825)
Okay, this this feels very different from what I was imagining when you first said it. I thought you were gonna say like, I'm starting a marketing campaign, I need to know what features the product team's working on to do this, or I need to know are there VPs in a different department that has some strategic priority that conflicts with what I'm working on. That's what I was imagining when you started talking about this. This feels a lot more tactical.

Rick (18:09.943)
Okay.

Rick (18:24.94)
Yeah, that's not the kind of stu Yeah, I would say this is much more oriented around accomplishing goals versus like understanding goals.

Tyler (18:37.091)
Okay. That each one of these just sounds like a c you spend a f a few days making a custom agent that knows about that specific thing. No?

Rick (18:47.106)
You're making it sound way easier.

Tyler (18:48.993)
Well sorry, I I've never even made an I say make an agent, I've never done it. You tell me you have done it before though. Is it easy?

Rick (18:52.024)
Yeah.

Rick (18:55.508)
It it it is easy on simple information. So like if the in a simple one person company when there's like one product, yes. It when nuance comes in, it's like, okay, well that's actually not useful because you're not giving me the full answer. So I think it gets harder with like, okay, but but what about this? What about this situation? If this, if that. like queer like that that's that's the yeah. Yeah.

Tyler (19:18.851)
And that's where it gets into maybe this won't work. I'm not saying give up on it, stop talking about it. Yeah. Like I this is h exactly how I feel about the customer service use case, which is it's really easy to get a proof of concept that can answer some of the questions right and all this. But then because we have a good help site and good search on the help site, and we've already solved that problem pre AI, the questions we actually get asked to our support to our inbox are the hard ones. And by hard, I mean they're special. They require real thought.

And it's so hard to get the AI, even if you're like, this is the exact answer to this exact question, it can't find it.

Rick (19:53.967)
Okay. Thank you. You just unlocked it for me. There is unique versus like re frequent types of questions. And I I think like the unique questions we want that's gonna be captured over here. And then if it comes up more than once, we start documenting and capture the answer. But the frequent ones, we know what they are, and we wanna basically eliminate a human having to answer those.

Tyler (20:02.413)
Mm, yeah, yeah.

Rick (20:21.206)
Incons and a and and in yeah, potentially inconsistently. So yeah.

Tyler (20:21.687)
Yeah, so

Tyler (20:25.067)
Yeah. So ignore AI for a second. Just like a workflow we're imagining is person has question, person asks question, look up is this question in our database yet? If yes, give answer. If no, log it and then say, Okay, you need to go talk to a human to get that answer. But then the next time someone asks the question, it like pings somebody and says, Hey, this is the second time we got this question. That type of thing. Yeah.

Rick (20:47.49)
Yeah, exactly. Yeah, like that kind of workflow. And then I guess the the the thing that the reason to invest in this is one, I think just that use case alone is ROI on this. But but if you build this right, now the AI has the context to do more capable things. Does that make sense?

Tyler (20:58.403)
Mm-hmm.

Tyler (21:04.449)
Yeah. I mean it kinda again, I made the analogy earlier of like every team at the company is like communicating via an API. I mean, it's sort of like each one of these things is its own little MCP server. Have you used MC I assume you've used MCP servers with your AI stuff? Like what's the difference between an API and an MCP server? And I think that's either a very simple or very complicated question, but

Rick (21:12.323)
Yes.

Tyler (21:30.828)
It's that the the AI now has the context of what can I do? I mean, you could do that with API docs too. I'm explaining this poorly, but like you're saying these different agents, I'm gonna use the term agent, maybe that's not the right term. These different agents exist for all these different use cases of the company, and then all of those feed into a centralized system that's like, here are all of the functions you can call. And then you you have you ask that kind of master AI to do something, and it's using all of these different agents that you've already built. Yeah.

Rick (21:59.189)
Exactly. Yeah. So like if we build this for humans, we then have an asset for AI. And just building it for humans I think is worth the investment. That's my that's my logic. Yeah.

Tyler (22:08.023)
Yeah, yeah. And what I like about this, I'm still highly skeptical that that final vision ever comes to pass with our current technology. But what I like about this is you're building these what you do now is build a specific agent to do a specific thing that's pretty well within the scope of what AI can do right now. and i if all you do yeah, you have upside, and even if it that never comes to pass, you you're it's still worth it.

Rick (22:28.032)
And you have upside. Yeah.

Rick (22:33.496)
Thank you. You've you've helped me like clarify why I I was having trouble articulating why, despite me not being bullish on the upside right now, it's worth the investment now. And it's like, well, one, you want to be able to seize the upside and we can't seize the upside if you don't do this. And two, just doing this if it works, probably is worth it.

Tyler (22:44.547)
Mm-hmm.

Tyler (22:53.089)
Yeah. Now there's a the a thing you started this conversation with that we haven't fully addressed is building it is one thing, keeping it up to date in a fast changing company is the harder thing, I think.

Rick (23:03.128)
That and that's that's where it's like, okay, this isn't a one-time cost. This is now a like this is now a recurring overhead expense and it has to be top down too. Like that's the other problem here, is it can't be bottom up because if it's bottom up, it's not reli it's not controlled, it's not reliable. So it's a top down initiative and and then it's like, okay, well, who's doing this work?

Tyler (23:26.519)
Yeah, by top down you mean the source of truth that's coming from a human has to come from a human who really knows what they're doing. Yeah.

Rick (23:32.952)
Correct. I it's or it's like it's you know, y the the most simple version of this is authority of org chart, like top of the org chart down, but what I'm really talking about is like knowledge authority down.

Tyler (23:47.49)
Yeah. It's like you have to take your best people and make their job kind of soul sucking, which is your your job is training in AI now. And then you also have to build the tooling on top of it. This is in in the customer service side of things, a thing I've thought about a lot is in that world, you probably do something like the customer's chatting with AI, it l they have the ability to say that did or didn't answer my question. If it didn't, that has to go to a human. The human has to

Enter something so that you know, I don't know enough about the the underlying infrastructure, but they enter something that either the data was out of date and you have to go correct it, or this is a specific use case that the AI had never been asked, so it hallucinated something, so you have to go enter that. But like it's basically a human's job to review all the failed QAs and then do some kind of like content creation data entry, and then it just gets better and better. The the better and better part is where I become skeptical. These these AI tools.

Do not hold very much in their head at once. And maybe there's all kinds of sophisticated tricks to synthesize the data and combine it together or whatever. Or like certainly if you could train your own model, like if you had all of anthropic at your disposal to make a AI just do what you need, that would be doable. But do you and I have the ability to actually cur like enter all this information and have it trained and actually have it understand it and hold it in in context at once? I don't know.

Rick (25:18.648)
There are tools out here who do that do this. So one tool that I am I am exploring is gitguru.com. and it is a it is basically positioning itself as a it's called the like knowledge layer between all of the AI functionality. So like sits between like Claude and Confluence and all the other tools. and so anyway, I can report back on this after I've done some more thought on it as a future topic, but like

Tyler (25:38.307)
Mm-hmm.

Rick (25:45.026)
They their value proposition is one, we're gonna make it easy for you to maintain this.

Tyler (25:49.368)
Yeah. I I would love to hear more about this. Finn, which I said we're trying, also has this claim, but we haven't gotten deep enough into understand But let me say the other side of it, which is we use Resolve twenty four seven for drafting AI responses to our support emails and then a human reviews them. and we like Resolve, but like and Front has their own version of this as well. What they are all missing, in my experience, is like it it does the draft wrong. There's just no way to tell it. Like, here's what the d here's what it should be.

You can correct it and send the correct email, but it doesn't learn. So try out get guru dot com. I'll try out Finn. I wanna compare notes on what what does the actual process look like for improving the mistakes?

Rick (26:32.088)
I like that. what what what Guru is selling me on is that they have an interface for observing where people are prompting it and where what what the answers they're getting and rating whether or not the answer was correct.

Tyler (26:46.989)
Great. And then how does it yeah. That's it sounds like exactly what you need. I would love to know what they're they're probably not going to tell you, but like then what happens on the back end to is it just, and then we go update the knowledge base article that it was based on, which to me feels like not good enough. I feel like the yeah, I don't even know what I want here, but

Rick (26:48.034)
Which is like I'm very excited about.

Rick (27:05.42)
Yeah. I I'll I I I know what you're saying. Thank you for going down this rabbit hole with me. It's really helpful.

Tyler (27:07.713)
Yeah. Okay.

Mm-hmm. Yeah, I'm I'm interested to hear how that goes. speaking of AI, I'm gonna go out of order here. What do you do if AI price is 10x? Like what changes about windfall and leg up health and lessening serum? Because it seems like these prices might 10x or are 10xing already. And what I mean by that is we can't be on the subscription. We have to be on the API pricing, which is ten times as much.

Rick (27:33.39)
I someone's

Rick (27:38.53)
Yeah, yeah. This is where I like I don't know enough technically about this yet, but like someone's gonna solve this with a local version. that so so to me, I don't know. What what are your thoughts here?

Tyler (27:48.556)
No, I think that's almost certain so I think in the short term, probably we pay a little more and use it a little less. And but yeah, in the long term, it seems like there's so there's local is one thing, but then there's like all these open source models, which for whatever reason they all seem to come out of China, that I have not used, but what everyone says is they're like six to twelve months behind the frontier models. can you accept six to ten six to twelve months let's say it's twelve months. Can you accept

Think think back to what AI was a year ago. If you could pay the current price that you're paying now to get that, to get year old AI, or you could pay 10x what you're paying now to get state of the art current AI, what do you do? And I think the answer is you mostly go twelve months ago and then occasionally you pay the premium for like really high value stuff. Does that sound right?

Rick (28:40.494)
I think that's right. Yeah. I think n not all tasks are created equal. and I think most of the stuff I I'm doing like don't need better models.

Tyler (28:43.373)
Mm-hmm.

Tyler (28:48.439)
I'm so bad at intuiting when I need a good model and when I don't. Like I see all these people online talking about I spit I throw this to Sonnet, but this needs to be done with you know with Opus and all and I I just can't tell which is which. Agreed. Yeah, which which has to mean you start with the expensive model and it decides for you, or something like that. Yeah.

Rick (29:01.506)
I don't think I don't feel like this is my problem to solve. I feel like that's the service provider's problem.

Rick (29:12.492)
Yep, otherwise people don't use it.

Tyler (29:15.051)
Anyway, okay. That that didn't end up being that meaty of a conversation because I think we just both agree on this. But the interesting takeaway for for me from that is Anthropic and OpenAI have terrible business models because the moment they start charging us, we're just leaving, right?

Rick (29:32.994)
You mean charging us more than they're charging us? Like 'cause I'm still p I'm we're paying a hundred and twenty five dollars a month to anthropic, but yeah, you you think you think that's highly subsidized?

Tyler (29:34.755)
Charging it.

yeah, but that's nothing compared to what their costs are. Yes, caveat. A thing that I find annoying about all the discourse about this is everyone is talking like the token the API pricing, the token pricing is the true cost. We don't know what the true cost to them is. We don't know what the the cost to provide the service is. We believe it's more than what you and I are paying. The subscription costs are less than the true costs and they're being subsidized. The token pricing is more.

than the true costs, but we don't know where in between those two endpoints it actually lands. So I don't know how like what realistically could they charge and still make a profit. I don't know. But I don't think they're making a profit off you and me.

That's my understanding.

Rick (30:22.926)
Well, that's a bummer.

Tyler (30:24.311)
Yeah. I kind of love it though. I mean like just the worst people on earth are stand to benefit tremendously from this AI stuff, and I don't want the AI to go away, but I want the people to go away, and that might be what happens. I'm a hater, Rick.

Rick (30:36.182)
my goodness. it's we haven't had a Tyler Hate in a while. I feel like this is c it's coming back.

Tyler (30:42.116)
internally I've been I've been doing a lot of hatin' lately. Can we talk about SpaceX? No, no, we don't need to.

Rick (30:46.506)
man, we gotta I feel like we need to rethink the format of the podcast where we have like a hate a hating segment. you know, at the last fifteen minutes of every episode, it's like things I hate.

Tyler (30:55.639)
I would love that. I I don't subscribe to very many newsletters, but one I subscribe to is like a St. Louis food weekly newsletter. And I don't read any of it except there's a micro rant each time that's like, don't you hate it at a restaurant when blah blah blah? I love that.

Rick (31:09.518)
I like the concept of micro rants.

Tyler (31:14.187)
Yeah, I need to I need to build that into something.

Rick (31:16.812)
Yeah. This is the perfect forum for it. The micr micro rants. I r micro rant any other micro rants you got today, Tyler?

Tyler (31:19.265)
Mm-hmm. Mm-hmm.

Tyler (31:23.895)
Well, unfortunately I've got macro rants, which is that SpaceX is not worth two point five trillion dollars, but it's I hate it. I hate it, Rick. I hate it. It's so annoying. Should I talk about other stuff that's not gonna rile me up? Yeah, so let's go back to the health insurance stuff. So first of all, I gotta do a correction from last episode. I said the the cheapest CRM I could find that's HIPAA compliant is Zoho at forty dollars per month. Insightly seems to be HIPAA compliant at at

Rick (31:37.71)
Yeah, that's what else you got?

Tyler (31:52.534)
on all their tiers and their cheapest is twenty nine a month. So I don't think that changes that much, but that's now the the bottom tier kind of price to to compete with. other than that I'm just still pushing forward on HIPAA. That's that's my priority right now.

I don't expect major conversation from that, but that's just an update. What I did want to talk about.

Rick (32:15.0)
So you're just gonna keep is the takeaway here like you're gonna keep HIPAA kind of baked into the pr is it just a feature of

Tyler (32:20.481)
Yeah, I think so. You you the conversation with you was actually helpful. Even though you were saying the opposite, you were saying raise prices. A thing you mentioned in the last episode, by the way, I know you don't listen to our episodes again. I recommend it, at least on anywhere your topic is a major one. Cause it w it wasn't until I re-listened to it that I had the moment of clarity. but you just mentioned a couple times, like if it's about values or just like core to your company, and I I think the more I think about it, the more I'm like,

Having any kind of add on, having any kind of second pricing tier, however you want to package it, it's a bell that you can't unring. Or I mean you can, you can go back, but there it's a zero to one thing. It's like a step change to go from we have one price, it includes everything. That's what it means to be less annoying. So anyway, yes, we decided we're gonna stick with fifteen.

Rick (33:11.766)
If you're gonna do that, it's probably not the HIPAA flag that does it.

Tyler (33:15.167)
Right, yeah. Especially 'cause the vast majority of our customers don't care about HIPAA. I and I'll just repeat, if we had multiple tiers already, there's no question HIPAA would go on a higher tier. It's just that we don't, yeah. I wanna talk FMOs, field marketing organizations.

Rick (33:18.55)
Yep, exactly.

Rick (33:25.826)
Yep. No, I I agree. Yep.

Rick (33:34.946)
Let's do it. I can't believe I can't believe you're bringing more health insurance topics to the podcast today than than I did.

Tyler (33:40.183)
I know, right?

And I've been talking to we we've had I've had a couple calls with some of our health insurance customers. Just ask them like, what do these terms mean? Like, for example, if you sell Medicare and like supplemental insurance, do you think of yourself as a health insurance agent? Questions like that, because I'm trying to understand the language here. FMO is a field marketing organization. I as a summary for the listener, I believe basically if you're an insurance agent or even an agency, you don't want to get appointed with every single carrier, like United Healthcare and Aetna and all this. So

The FMO, they're sort of like a merchant of record for billing, right? Paddle figured out how to pay taxes to all these different governments. And so we do all our billing through Paddle, they handle it all and then they give us the money. And FMO is basically that, but for selling insurance.

Rick (34:25.546)
It's the the the the the agent still is contracted with the with the insurance company generally and but it's through the FMO so that the FMO can sort of handle all the downstream sort of admin. but I believe like insurance regulation requires there to be a con like a contract between the person who's receiving doing the the placement of business and

Tyler (34:34.031)
okay.

Tyler (34:44.269)
Gotcha.

Rick (34:52.674)
the carrier, the but the FMO sits in between to handle the admin. everything from the paperwork to the training to the certif like ongoing compliance, verification to commission payment, that kind of thing. And then so so all all the insurance company has to do is send like all the commission to the FMO one time and the FMO figures out who it goes to.

Tyler (34:56.483)
That makes sense.

Tyler (35:07.959)
Gotcha. Okay, let me

Tyler (35:16.727)
That actually helps close a loop for me when we're writing up all this, like the hi the BAA, the the contract that you have to sign for HIPAA. By the way, I when I re-listened to our episode last week or last episode, we did not define any of like what what is HIPAA. Anyway, sorry, listener, you better know what HIPAA is. the BAA, the contract that you sign to like tur become make this CRM HIPAA compliant, there's the term business associate and the term covered entity.

Covered entity is the, as I'm probably gonna butcher this, but like the primary entity interacting with the patient. And then the business associate is like downstream of the covered entity. So for example, if you go to a hospital, you're the patient, the hospital's the covered entity. But then if the hospital use Epic as their electronic medical record system, Epic is a business associate of the hospital. The insurance carrier is the covered entity.

The insurance agent is a business associate of the carrier, so it makes sense that they have to be appointed with Sorry, I just said something that no one is going to follow or care about, but that helped me.

Rick (36:24.862)
Okay, so it's it's contract like all appointments are just for translation are contracts to to like reseller agreements basically.

Tyler (36:31.477)
Mm-hmm. Mm-hmm. And the the key thing is the the in the customer of insurance, the patient, even though it's a weird term, but the patient, their relationship is with the carrier. And then the carrier has a relationship with the insurance agent. The patient does not have a direct relationship with the insurance agent. Policy holder. Okay. I mean that's fine, but in the in HIPAA world they call it patient. I think that's what every everything I've read. Anyway. Okay, cool. Cool. I I hate this project.

Rick (36:48.802)
Policy holder. Policy holder. Yeah.

Tyler (37:00.749)
Cool, let's go. FMOs, as a result, they're basically an like in more general business terminology, they're an association with lots of insurance agents underneath them. So Less Annoying CRM right now has probably something like 2,000 insurance agents using it. it'd be nice it well, that's going in one by one, getting a bunch of independent insurance agents, or maybe getting like small agencies, like five or ten people at once.

The real dream is to partner with an FMO that has thousands of agents under them. Well, the the real dream is that they pay for the CRM for all of their downline. the middle dream is that they recommend us to all of their downline and then the downline buy it themselves. So I have more to say about that, but yeah.

Rick (37:46.904)
We have a Legup Health has a general agent if you would like to talk to them.

Tyler (37:51.33)
I would love to in a bit. I I wanna n I I s I'm still like learning how to say the words so I don't sound like an idiot, but yes, that actually sounds great. Do you know how many agents or agencies or whatever would be

Rick (38:03.502)
I haven't talked to them in a long time. JD probably does. but but they're Utah primarily Utah focused.

Tyler (38:11.681)
Yeah, I mean, great. so the conference we're going to is organized by an FMO. And I think this is actually pretty common that part of the value add that FMOs offer is they're like, We will run conferences and do continuing education so that like agents have to do these classes every year to maintain their license, right? the FMOs provide a lot of that stuff. So anyway, we I got on the phone with this FMO group.

That runs this conference, thinking like it's just about to sp we're just sponsoring the conference. That's the whole thing here. And they were like, you know, our our network definitely needs a CRM like this. Can we put together a CE, like a continuing education class that you guys run and talk about less annoying CRM? Can we get you on webinars? I thought we were gonna have to spend years like building trust in the industry before it was worth talking to FMOs. And this just happened on the first call. So that's good. That's a good sign.

Rick (39:10.488)
Heck yeah.

Tyler (39:11.547)
and then coincidentally, our affiliate program, which we've had for years, we had someone reach out to our affiliate program like, hey, I want to be an affiliate. I'm an FMO with 3,000 agents and I want to recommend it to all of them. And we're just coincidentally, like literally the same week that I think it is. We haven't none of this is public yet. Yeah.

Rick (39:25.228)
Is it coincidence?

I mean, we talked about on the podcast. We talked about on the podcast. So so I'm just there there I think like when you focus on something like this, like with like very much intention, it good things happen and you can't tie it back to anything. but I doubt it was accidental and coincidental.

Tyler (39:44.491)
Yeah, it it I agree in the sense that it would be a crazy coincidence. After sixteen years in business, we've never heard the term FMO before. And then as soon as we start talking about it, we get I don't I don't know if 3,000 agents is like a big or small. I don't know what like where that fits in terms of the s the size of these groups, but that feels pretty good to me. so yeah, we've got two FMO leads now all of a sudden. this has been a thing that we

Rick (40:07.926)
Yay.

Tyler (40:12.833)
Have tried in the past, it has failed. Trying again. But I think the big realization here is like we we need to focus on our ICP, which is one to five user accounts. But associations of one to five user accounts are a great distribution channel. They're not our customer. That's what we messed up in the past is we were like, we're gonna try to do enterprise sales, we're gonna try to, you know, if we can get a 500 user account, which we have before.

Let's go do that. And and it's it's always not been worth it. It's failed. It's been a low hit rate. We can't build a business on that. But thinking of it as a distribution channel rather than this as our customer, I'm hoping will be a new approach that doesn't bomb the way the old one did.

Tyler (40:58.142)
all right. I've got more, but I feel like this is a good stopping point unless you got anything else. All right. Yeah, good talking to you, man.

Rick (41:02.146)
I got nothing else. This was a good episode for me. I I might I might chop out the bit to go listen to it again and pull the transcript so I can start thinking start building out a plan.

Tyler (41:08.492)
All right. Yeah, go do some AI stuff and keep us posted. All right, man. See ya.

Rick (41:14.051)
Will do. Have a good week. Bye.