Me: Extinction risk. Capabilities has of AI called the AI agents. We found out that hackers manipulated cloud to break into Mexican government systems, steal data on over a 100,000,000 people. The tool didn't just write code or perform odd tasks for the hackers. It planned and executed most of the sophisticated campaign itself. And now we're starting to see lots of control in incidents. These include agents stealing passwords, harassing developers, modifying themselves to have a shutdown in order to achieve the often mundane goals they have been given. The weekend, we found out that Chinese tech giant Alibaba produced an at in age Them: I have an extra desk over here. You guys wanna rent it? Me: Oh, yeah. Let me put that Them: Oh, I didn't send out Eamit. Oh, wow. I didn't send out the agenda. Me: I'm Brandon. Them: Why? Matiah's an adder at the same office right now. Oh my god. Me: Yeah. He's getting right across from me, actually. Them: He's he's in a s v east. I did an agenda, and I think I closed the doc for that saving it. That's silly me. Maybe not. Let's see. Oh, there it is. Hi, Earl. Mhmm. How you doing? Me: One second. Them: Who we are? Close to finishing up the updated website, getting companies in there, adding people, Although we have a few more people to add, we have to add Gary and Hopefully done next day or two. Alright. So I didn't get you. I didn't give my email address. So if you're interested in Earl at ISU, see Okay. Works too. This is not working right now. Honestly, We can we should we can use that now, Earl. Yes. Cool. This is including the read now that you're here. Is that the one you like us to use, Raul? Yes. Alright. We're gonna use for your for your documents that we're gonna use your Gmail address. Okay? The Gmail is yeah. Yeah. Good morning. Don't wanna change that. Alright. So why is this active mirrored? For some reason, I can't text into the I can't text the agenda into the chat. Something just showed up. Could you look at but I can't see it. Oh, it's like a Yeah. It's like a document. It's an image. Não sei. They can do it. Oh, paste and match schedule. There it is. That's weird. Sorry. Okay. Alright. One zero five. We should start. So FlashPass is done, signed, and monies are gone. Me: Great. Them: We got our money back from Me: Dome. Them: Dome What's that? You plan to distribute that, or or or do you just hold it down? No. Scope. This could be another text. Dome came in. FlashPass went out. Same dollars. And it wasn't return. It wasn't a return on investment. It was just a return of Me: Yeah. It was the same investment. Right? It's Them: There's no no reason to no reason to distribute that. Okay. So we talked a lot about the due diligence process last week. In terms of what we're gonna do. Has anybody been giving any more thought over the past couple of days? Me: Yeah. Me, Dax, and all we met this afternoon, kinda talking about you know, the the AI agent space within Notion. All right? So yeah, before we get into the deals, right, like, one of the the other things that I was I was working on or I hadn't been thinking about is a broader vision of that. It's more like the ICE Adventures operating system right where you have more than just to do evaluation, where you can have you know, currently, we have data in Notion, Google Drive, email threads, WhatsApp, and frankly, you know, your memory, Ed. Right? Like, the the fast deals, I think, the the consolidated sheet you sent me, they were like, 940 companies across four or five different Notion sheets. You know, we have this 37 companies across the fund one and fund two, but there is not a single path where an LP or anyone that interested in the fund can log in and see everything. Right? Like, Earl for instance, if you wanted to look up a company that we reviewed six months ago, would have to dig through the old notion or ask you to have access to those. So I really wanna build more like, the IASV operating system within IASV Co where you can log in and you you were able to see, you know, the full deal pipeline, a portfolio dash board where you can track different things. The the AI screening results from both the batch screens and the deep evaluations that that we've been doing with the Cloud Code skills. One of the things we were talking about is also, like, a wall of failure. Looking at the companies that we passed on that maybe were good investments, Them: Hey. Me: we learn from those and help that feed into how do we improve evaluation of future deals. And this light LP portal where, you know, every LP can can log in and see at a high level, you know, portfolio info, and, yeah, or had some comments around the too much transparency within the hedge fund and how maybe too much information the hands of investors is not ideal. So I think we we can we can work it out what that would look like. But the reality is just just Them: I'm sorry. I didn't understand what you meant there. Too much information in the hands of investors. Me: Yeah. Earl, you want you wanna talk about that? Them: Yeah. So, let's see. So I am I remember running a hedge from this. As as in auto, I'm giving this much transparency to sponsor So I gave I gave him a day I gave him, like, daily trades. Trades every day. I expose that to gave it to investors. And I just confused it basically is confused with him because he didn't instrument how the strategy works. It got more confusion in that case. Are you talking about, like, Me: Yeah. Like, Them: reports out to the LPs is what you're saying? Yeah. Me: like, the level of data on transparency and the information, like, what the analogy he he used is a lot of the the LPs or investors, they they're fine eating the sausage, but they don't wanna know how it's made. Right? And and they just throw in all that information might just add confusion. It's it's gonna be more, yeah, more things for you to manage. But, I mean, we can we can specific specify exactly what that LP portal can look like in the future. But Them: Oh, a portal. Mean a portal. Okay. Me: yeah, but the idea Them: Yeah. Me: the idea of of, like, you know, using AI agents within Notion itself There's a lot that it can do, but it's not really suited towards external research. Right? Like, so it's really good at at, you know, filtering, evaluating, moving things around within the notion ecosystem. But it's not really good at say, okay. I have this company out here. Go and research the LinkedIns of the founders, create a bibliography, research, competitive analysis. All this kind of stuff lives outside of Node I think they're they're setting up something like that where you can customize your agents towards that. But you're gonna get beat up on token costs on on each of those agents is gonna be three hundred months a month. Them: Yeah. I think doing this all in Notion is gonna be really a Me: Yeah. No. That's that's kind of the the consensus that me and and Them: expensive, I imagine. Me: and Earl reached earlier. And and Notion is great for kinda, managing the pipeline part. But all of that intelligence layer, I think, you know, Claude code and and and other features are probably much better at it, and we can integrate with it in a way that makes sense. Them: Mhmm. You know, one of the things, like, for example, I went to this this group down in in the park last Thursday. Where it's it's kind of a lower quality angel group slash and people present But out of the 10 companies that presented, there was one or two that were interesting. So I would just love to, like so I just they they sent the follow-up on all the companies. Like, just take this. Throw it somewhere, have it analyze it, and then do a which I I wouldn't look at this anyhow. Right? I wouldn't look at these companies. Me: Mhmm. Them: Right? I got like, I went to that one thing. I won't go again because there was some interesting ideas, but not enough to Me: Yep. Them: drive down to Menlo Park. Right? And they do this every week. I could just cut pace, and it would screen It would pull up one or two. It you know, our prompt has gotta get better. Like, we should we should take your prompt you did, Matías, and and distribute that and refine it. Right? Me: Yeah. Refine it. Yeah. Let's Them: That that's probably that's probably an exercise we should probably Me: Mhmm. Them: do together. Me: That's that's the Them: Think. You know? Me: the idea that Earl and and and, yeah, Dax and I, we were talking about this idea of recursive Them: Yep. Me: self improvement, and and and the same thing with the wall of failing. Right? Like, the wall of failure, the wall of successes, what, like, are the companies that we invested in that did did great. Companies we passed on that also did great? What can we learn from that? And and use that to further refine the the the the prompt. Right? And and I've been doing this kind of stuff manually so far, like, Ed I go to and and even, like, Nicolas sent me the latest plug and play one. Right? It's 30 companies presenting. It's a webinar. I'm I'm I'm having lunch. I just have that webinar running. I'm recording the the whole transcript. Them: Oh, yeah. Yeah. That's good. Yeah. Me: They sent afterwards the link for the the company. So I just sent, oh, here's the link we the websites for the companies and the founder information. Here's the transcript of that meeting. You know, run the DealEvaluate batch screen. Right? So it it goes over that 30 companies, runs that, and finds, well, maybe within those 30 companies that we're for, that are worth doing a deeper dive in. And then out of those four, then it runs the full deal of evaluate where it goes further with, you know, deep bibliographical research. Competitive analysis, even the true potential score that you mentioned, which was, like, why the company can expand to different verticals. Their true potential is much broader than what they're currently doing. Or at the same time, it might be that this company is just really focused on this wedge and doesn't have anything that they can do beyond that. So that would harm their their total potential score, right, and and and other things like that. But, yeah, those Them: I would like to you know, one of the things I like to do is, like, if I get an email, I would like to forward it to a place maybe in the subject title, say what I want. What I wanted to do. Me: Thank you. Them: Know what I mean? Or something like that so I can so it's easy for me to just, like, send it to an email address or send it or communicate it somehow. Right? And I can just give, like, a sentence in terms of how it should be treated. Me: Yeah. Yeah. Them: Don't how we do that. Me: I know that I know that, you know, on on the hub, right, because you're you're still part of the v c v c hub, like, deal hub, they they have their Chrome extension that if you receive an email, you can just, you know, click the Chrome extension, and we'll extract that information from there and put it on your deal pipeline with Them: I don't I don't wanna use VCLab. I don't I wanna use anything Me: within the hub. But yeah. Okay. Them: I don't I don't wanna use anything from VCLab. Sorry. Me: No Alright. Them: You know, I I think it could be just we could create a new email address. And we send it to an email address or something like that. Me: And that's the the IESV claw. You know? The open claw, IASV that we're yeah. Them: Right. Or or it could be it could be a WhatsApp. You know? But I I think probably email is probably Me: Yep. Okay. Them: better. I don't know. That's you guys are smart and I am on this stuff. So Me: Yeah. I think the the Them: okay. So what are next steps then? Me: Yeah. I think that the I can I can set some time with you? I created the present the presentation. I can walk you through it afterwards. But the idea is, like, even with the the IACV Ventures portal, can have something that also modernize the current website because that's a bit outdated, right, with battery company. So we modernized the website and then we'll do a, you know, a log in there where the current LPs can sign in with Google Auth, and then they will have access to something similar to to Notion, but also looking at the the current portfolio and, like, the updates and things like that from the companies, and they can add that there too. Right? Like, so if, Gary is responsible for one particular company, he could do the same thing. He received an email from them, forwards it to the same IASV email address, and then those updates will show up in the the IASV portal. Right? Them: Wondering whether it it would be better if if there's a prompt that prompts back into in terms of reads it, and then decides it has questions to you in terms of where this is from, how do you view this, have you reviewed it, whatever. You know, like a back and forth so you can so there's more clarification in terms of what Me: Yeah. Them: this is. Me: And what are you supposed to do with it instead of just starting Them: Yeah. And and did you give me the right information, or were you lazy and just like, sent it without thinking, you know, type thing? So a prompt maybe that would Me: En el Yeah. Them: maybe detail it out a little more in terms of so that Me: Yeah. All right. Them: it would do a better job at screening it. Me: Yeah. Or or even, like because because I've what I've been doing for this deep devaluate ones, like, I get the the link from from YC, but I also get our our comments around it. Right? Like, so for instance, one of the companies that I was evaluating here today, was this this OpenSpec Right? So, you know, they have 36 like, Mark comments around it were already, oh, they have 36,000 stars on on on GitHub. That's great, but I don't know if there's a business plan around it. Right? So I I commented that I I put the whole thing into the do evaluate, so it already looks and says, okay. Yeah. Mark's on the right track. There is currently zero revenue. There's a solo founder. No real pathways towards revenue or or building a business out of it, and then it agreed and creates a score there. Right? Like but all that context all that context is also relevant, and it helps it guide where the the the deep dive, the deep research within a company is going. So we we put our thoughts on how we're thinking around that too. Them: My my only caution on a formal scoring system is that it may become misleading And and, like, if if it's a if we're scoring, Matías, like, the stuff that you're doing, which is fantastic where you're looking at, like, looking at, like, 40 companies we don't have the time to look at. Me: Mhmm. Them: Then the scorings then the AI scoring system makes infinite sense. It's perfect for that. You know, we may wanna fine tune it a little bit. Check its performance, etcetera, etcetera. If it's a company that we're all looking at because it's part of our practice, Me: Yep. Them: then I mean, we can put a score there. I guess it doesn't hurt anything to put a score there. In fact, it may be a good way to fine tune our scoring system. Me: Mhmm. Them: But, Me: Superficial. Them: just don't want to be overly Me: Yeah. Them: overly indexed on the score that it produces. In fact, maybe that's Me: Mhmm. Them: maybe that's our opportunity to improve the scoring system. If we're all actually looking at it. Me: Yeah. Earl was saying earlier something more along Them: Yeah. Me: like, the the emperor of Rome, right, where it's just like a a thumb, and then it goes like, up and down and and maybe, like, a ranges. Right? Like, this is a this is a strong strong thumbs up or it's, you know, halfway through or just like because the score we can look at, okay, anything above a four, but it's just it's the same. Right? Anything above a four will be, you know, that green space. Anything between a three and a four will be a dig deeper, and anything below a three is probably going to be a pass. Out of five. But we can do it more in ranges and more as recommendations rather than, oh, we just see a company that has this you know, a 2.8 score, and we immediately discard it without actually looking through. Right? Them: Yeah. Yeah. We should do some looking through because that's Me: Well, Them: how we're gonna improve the scoring. Right. Me: And and it's exactly what I was talking to Earl about this earlier today too. Right? Like, VC is a is a contrarian business. You know, we were even discussing the Pyrus, deal. Right? Like, where now their valuation is at a 125. Probably everyone is putting their prompt on ChatGPT, and ChatGPT is telling them that this is the most amazing company has ever come across their desk. And everyone's jumping at it. Right? And and you know, there must be a way to to also find diamond in the rough or other opportunities that maybe other funds are passing on. That are actually substantial. Right? And and and how do we further enhance our own process of of finding those those nuggets where our, you know, undervalued for what they already have or, you know, have a true potential of being much more than they're currently have, even though they are at still zero traction or pre revenue and things like that. That will will substantially you know, underscore the risk and and give us a nice return profile for the particular investment. Them: Another thing I'm thinking about is, like, our database is getting pretty valuable in terms of what we've we've put in there. I mean, it's and I think as we do this, it's gonna get more and more valuable. We're gonna have a great history of of companies. Which we can use, a, to analyze current companies that we're looking at Also going back in history, I'd love to start analyzing looking at companies from the past to see if there's things there. But also in specific areas that we like that we're kind of trying to research, like, material science is something this is in, you know, robotics. Material science is interesting to me. So I'm doing a lot of work there. Me: Ajá. Them: And, you know, looking at existing companies. But so two things there is, like, being able to go back and take some of the information I'm looking at the company now going back and looking at other information we've acquired. From other companies and see how they relate to this company. Also, as we as we wanna learn about, you know, maybe specific LLMs that people are building for health care or for financial, whatever. Me: Yep. Them: In terms of how, you know, how people are doing it, stuff like that. Me: Mhmm. Them: That's kind of the next the next level of analysis to kind of being an agent. I can see an agent just going through and kind of helping us. Like, I'm looking right now at a Me: And Them: a start up at out of Berkeley, which is doing material science. In both pharmaceuticals as well as for, you know, actual materials, like, Me: Yeah. Them: you know, new new elements and, you know, who the team is and and learning more about that and analyzing that against Me: Yep. Them: of the other companies that doing material science. Me: Yeah. And and then, like, one of the limitations we were talking about of of just, like, web agents, right, is that a lot of information sometimes is confidential that we might get access through. For instance, like, with FlashBass, you received a lot of more insight from the the lead investor. Right? Like, if you're doing that with Claude code or in a folder, you can put all that information in, and it's gonna use that as as context for that evaluation rather than just doing a web scraping and finding competitor there. And and we were also talking about the the deals that we passed on and revisiting. Right? Like, Notion could do something where it just it just gives you an alarm and see months from now to say, hey. Look at this company again. But we were talking about looking at that from a broader perspective too. If there's a market event, right, like if they just released an update or or they released that, oh, they just closed the deal for $5,000,000 with client x or y and z. Like, we should be running those periodically to resurface this kind of deals of companies that we might pass because they're too early. And then, okay, they closed the deal. They are interesting to reach now. But if we waited six months to reach to them, maybe they lost the opportunity. Right? If that opportunity is resurfaced right then, you know, on Friday, we have that insight. On Monday morning, we send an invite to the founder. You know, whenever they open the round, we're the first ones that they're gonna talk to. Right? These kind of things. So so there's there's a lot of process around that too, but no, we we just, frankly, running running kinda like this web scraping on every single update that came out out 940 companies, that's gonna eat our our token usage, and it's gonna be very expensive if we're running those daily. Right? So you know, how can we batch it, maybe, like, run it periodically look at companies maybe in an interval of at least a three months or or things like that, right, not to not to get our our expenditure too crazy with AI tokens as well. Them: No. Thank you. It's it's a bit set up, like, these Google name of the company or whatever. It says Me: Okay. Yep. Yeah. Google, Alex. Them: Google or gives you some data updates, and then it gets used in caller pages. Great context, and then we run run the same periodically. Me: Yep. Them: Okay. And one of the things, I would like everybody to if you're if you're comfortable with it, every meeting you have recorded because you can always turn it off But I'm just finding like, for example, on the the flash pass, I did have a recorder when I talked to the person. Actually, I think I was on the phone. Well, I was on or I was on the phone using Me: Think he recorded it. You because you sent the Fathom the Fathom with the founder, you sent it to Them: no. The fountain. No. No. With the investor. Me: me. Ah, with you did, but with the founder you did. Them: With the investor. I I I didn't. And so it would I think I did it on my phone, and I didn't I didn't get I mean, actually, I don't even know. Adam doesn't come up automatically if I if I fire up a a a, you know, a a meeting. A a web or a video meeting. So I need to figure that out. But, you know, you can always turn them off. But I think it's probably a good thing. Because I had I had I had a pick my brain in terms of, okay. What do we talk about? So it wasn't as accurate as it could be. So I think that'd be a good thing. Okay. Let's move on. So next steps it sounds like you're kind of the point person on this, assuming that you grab whoever you want. To to and who's willing to help. So keep moving that forward. I think there's a lot of potential here. Really make it'll make our analysis so much Me: Yep. Them: better and better investments better investments for sure. Okay. So I'd like to turn let me see. I'd like to turn the stage quickly to Mark. It was interesting he he's working on interesting project. He's he's he's a cloud call Clyde whatever whatever it was called. Your so, Mark, why don't you just give you a quick overview of what you're doing? I think it's it's pretty interesting. Yeah. I'm trying to my screen. I gotta be off camera, folks. Because, my computer's really eating crap right now. But, anyway, can you see you see the screen that I I put up there? Yep. Awesome. Okay. So this is, this is just sort of like a little thing I was noodling around with. Like, all of you are probably aware that Texas has no in state income tax, which is oftentimes outed as an extraordinary benefit. Of living in Texas. But the part they don't really discuss that much is that Texas actually has a very substantial property tax in virtually every county. I think typically, it's at least one and a half times or more, sometimes as much as double as what the property tax can be in California. And, furthermore, there's no 13 protection. So they will reassess you every single year. And so this creates a funny dynamic in Texas. So taking Austin, which is Travis County, taking Travis County, Austin as an example, There are 400,000 parcels in Travis County, and every single year, they reassess them And every single year, 30 to 40% of the people who own property in Travis County file a formal tax protest where they demand that the county lower their tax assessment. So it is this peculiar sport which has developed in Texas, where, over the course of ten years, you're going to be protesting to the tax authorities, you know, 30, 40% of the time, and asking them to lower your taxes. And this has created a little miniature industry in Texas where, there are professionals and services that help you do that. And my assessment of them was that they are all antiquated, and none of them really, are using the level of data that's available. So I just I said, well, what if I wanted to use available data could I produce? And this is my alpha that you're that you're looking at here. It's sort of a work in progress. I'm gonna give you an example that, shows what this can do, and it's a cooked example. Because it's something I noticed, like, ten minutes before we started while I was testing it. So let's type in an address. It's 1400 Fultz. And, what it does is it shows this zone here, which is TCAD. If I Refer To TCAD, that's a Travis County Appraisal District. This is TCAD's view of an a comparable neighborhood. So things in this zone are comparable. And by the way, to get your orientation, here's downtown here. Here's the river or lake, and then this is this is Zilker Park if you've ever been to a concert and gotten at Zilker Park. And, this is the neighborhood right here So it's kind of a cool neighborhood. So I'm looking up at this 1400 Folts Avenue. And it tells you a bunch of I this is all pulled from public data. And I'm going to do, like, a little comparison to show you an example of how this person may be overpaying on their taxes. So I'm gonna click on where is it? It's the one I'm looking for is 12 o two faults. So this is a property comparison tool that I I built with QuadCode. And this shows you that if our property is 1,400, there is this other property on the Block 1202 which has $80,000 lower valuation, which translates to about $1,200 a year. But it's very, very similar in many basically, every other respect. And in fact, the comparison property down the Block has additional living quarters built onto it and, should should be more valuable. So if I'm going to do my tax protest, I am most certainly going include this as one of my three or four comps that I'm saying you are overcharging me for taxes. And so this is a it's early on. I'm gonna build it so that you can kind of collect your comps. And you can poke at the data automatically in a number of different ways because it's all built on a SQLite database. And, so this this is what I'm screwing around with right now. I'm gonna try to figure out if I can charge people for this. Right now in the state of Texas, it's about people spend about 200 to $250,000,000 a year. So it's a micro mark market in terms of the things that we look at. It's not something that we would look at. But it's or that any VC would look at really, which is part of what makes it attractive as a single person business. So for for 200 so within Austin, I think it's oh, I figured out the number. I I don't remember what it was off the top of my head, but it's it's, you know, tens of millions of dollars in Austin that are spent every year on these tax protests. And so, I'm gonna see if I can produce something that I charge 50 or $100 undercut these sort of hacky tax prep offices that basically do something like this anyway except not automated. And, and see if anybody wants to spend money on this. And if they like it, Me: And Them: then I can expand out to the other counties in Texas. My Me: are there other states where where, like, tax brokers like this are similar Them: project people Me: similarly in place as well, Mark? Or is this very specific to us to to Texas? Them: Yeah. My gut says you can you can do this any pretty much anywhere in the country or I'm guessing. But I imagine California is Yes. Like, if I if I wanted to There are many this stop sharing the screen. Me: Yeah. You're cutting. Them: Your your your audio goes in and out tomorrow. Me: Cutting a bit. Them: I I gotta figure how to stop sharing the screen. I see this thing. Here we go. Okay. Got it. Okay. Yeah. There's there's I think, you know, there's probably several states because they're nationwide firms that do this for people. And so there are I can't hear you. You're you're all over the plate. Me: Yeah. He fell. Them: Oh, I think he left. Me: He fell off. Them: Yeah. Now we fell off. Okay. Okay. Give me a second. That did solve the steering problem, but not in the way he was hoping. Yeah. Wait. So you can use this outside of Austin, Texas? Me: También Them: Is that where Mark I think he I think he's I think he's you can bill use it most most Me: account, but he Them: places. Okay. You should you should build use in most most places. Think you'll find I think you'll find that all of the states that have very high property taxes and low income tax or no income tax have similar processes because if property tax is their primary resource, for for the government, then there will obviously be you know, a a way to contest whatever value they're assigning. I know California has a contest. You know, they send you a note every year in, like, March or April and say, your valuation is this. When your property tax comes due in November, if you haven't protested, this is what your bill's gonna be. So California definitely has a similar process. You're muted, Mark. This cap, though, he kinda raises it at 2% every year. 3%. Yeah. So that essentially, like, moving to a state like Florida or Texas or I forget what some of the other ones are. I get evens out. So you're not really saving that much money because they just charge you a higher property tax. They even out the low to no income tax. Is that Yep. Yep. I mean, Texas it's a question of, you know, all of these tax policy things. This the government always has a, you know, as a need for money. The question is who do they want to take that money from? Property tax versus sales tax versus income tax hits different people in different ways. Yeah. Okay. How can I get you in all three simultaneously? Yeah. Excessive income tax, plus the excessive property taxes. What is the property So, Mark, what what is the average property tax in in Texas? In terms of percentage of of value? I think it's about can you hear me, guys? Yeah. Hear me, guys? Yeah. Okay. Me: I hear you better now. Them: Oh, yeah. It's I think in Austin, it's, like, 1.8 or 1.9% And in California, it's, like, 1.1 or two or something like that. Yeah. Okay. 1.2. It's one and, again, California is 1.2 And because of prop 13 capped, if you've been there very long, your valuation has nothing to do with the value of your property. So California may have a, quote, unquote, high property tax, but for the people that have owned houses more than a decade, it's a high value against a low valuation. Right. So I think there's 20 or oh, there's a couple dozen states that do this annual thing, where they they they reassess you. So, theoretically, this could be bigger and go to more states. But, I mean, hey. I just wanna see if I can get people in Austin to use it, and then I'll think about scale. I mean, just think about it. If if I could protest, if you could get the right information, if you could cherry pick the information, okay, a person pays $50. Right? The program does everything for you. Right? You just give your address. It goes around. It cherry picks the ones that are lower in terms of value. It puts it it's plucks pictures of what things are all about. And you know, you can you can include pictures of your house if you want, if your house isn't that good inside or whatever, then it automatically submits it Right? You track that, and it it comes back you know, through your through your portal to the customer. They don't do anything. Yeah. I mean, to the honest thing is, you know, your your big debate, you said, you know, I wanna charge $50 and see if I can win. Quite frankly, if it really was a fully automated, click here, and we'll submit their thing for you. Should be a percentage of whatever the reduction is. Because people just pay one to 2% of whatever they save. How do you collect that, though? How do you collect it? Know how you collect it. There must be some kinda contract. Because I looked at the competitors in this, and there are people that do it. That's structured. But this is this is AI could automate this whole thing. So easily. Yep. And here's here's another thing, Mark. Assuming this starts going crazy, right, and a lot of people start signing up for it, Guess who's gonna wanna have a program which is going to take these requests and automate those requests to then re then send back a response to these homeowners that are sending it. In other words, you could you could also have a program for the other side, for the municipalities, right, to automate the whole thing. So you're automating You're creating tanks on both on both sides. Arms dealer. Yes. Exactly. Right. But danger of that which they'll never, you know, they'll never do it. But the argument is if the government had this, then they'll figure out the arbitrage value of like, how how much can I raise it before someone will use this product to complain? You know? So They do that. Do that. What's the beauty of AI coming forward? Right? People are gonna start using this for assess assessing. I think the ideal would be is if I could give it away for free. To the TCAD, to the government because and let and if I can tell people, you know, this is the Me: That's Them: this is a tool that Keypad is using to, look at your shit. And so perhaps you should be looking at it too. There's there's no but there'd be no better sales argument than, you know, why not take a look at exactly what TCAT is looking at? Yeah. Because people want pee people wanna do that. People just wanna on a button and get it done. You'll have to you have to so what happens is the pro the way the process works is that you file your, you know, your best comps and you say, need to lower my taxes by a $150,000 valuation or whatever it is. And then TCAD will come back you, and they'll say, these are the comps that we're using. To come up with evaluation. So you need to look at their comps in order to diffuse their claims. So you could say, well, that comp has a swimming pool. That comp is, considered a higher class of building. Well You know, so on Or that's part of your program that you're gonna do. The person's on gonna do that. You're gonna do that. I'm gonna do that except they have to the their I don't have API access to TCADs. Reply to them. So there'll have to be some some human intervention to look at that. They'll have to look at Specifically for that homeowner? Yeah. So TCAD's gonna send them a letter an email or whatever saying this is the this is these are our valuations, and so I need to ingest that. Well, that can come through your that can come through your portal. Right? You can you can be communicating to them via your portal. Right? Right? Possibly. So I think it's an interesting idea. I think it's this is gonna be an you're gonna have to advertise but I don't I've never seen an ad for something like this. Never. So and then you could do an analysis of property values that are higher on the country or have increased a lot over the last year. You focus your advertising there. This could be a big a quick big moneymaker, I think. I don't know. That's my thoughts. What what do you guys think? Your other interesting market is insurance because they're the other people who care about the value of the houses. In terms of selling it, Oh, in other words Home insurance. So in other words, you've charged me too much. This is what the value of my house really is. Or, my cost to rebuild my house is higher than you're insuring me for, and I'd really like more. Right. Yeah. This is this is, yeah, replacement value. Oh, insurance. But you're you buy insurance based on the the the value that you determine on your policy. Or they determine Right? My boss Every year a policyholder to start contesting with the insurance company. Sees as a value. It's it's just another potential scenario. Right. Okay. Well, thanks guys for the future. Me: Yep. Them: I really appreciate it. It's really an alpha form. And and to in in full disclosure, there are several different kinds of workflows that people go through to protest their their tax. And this covers the main one. But there are others as well, and I'm gonna have to figure out how much I can suck into this system you know, to automate. So so it's it remains to be seen still, but Me: Es automático. Them: I'm I'm really excited by it. Pretty cool. Cool. Okay. I think I mean, I if this thing works, you can make millions overnight. I really do. I think you could. Complaint. Alright. Let's go to anything else before we go to our So one of the things I think we we're talk about Rosa, to the demo day and the last YC batch. Is that, you know, the validation seemed to crept way up there. Yes. Three three minute three minute base, forty minutes, cap that type of thing. So part of the question there is was the was the who's driving that? Is it because it's gonna be easier for the project the issue for the companies to hit, like, the they know our market cap. The reason why the reason my opinion is that YC is doing screening work and they're they're going through thousands of companies. So investors, they're doing a lot of the screening YC's doing a lot of the screening for him and YC has a brand. So as a result, they're they're paying much higher premium But if you notice very few of these companies are getting large investors involved. It's mostly smaller ranges. That are investing in this stuff. And smaller ranges don't have the deal flow. So I think for us, it's there is opportunity there. But the what I see in my mind is a learning experience us occasional company. And, you know, the valuations, they're I don't think we don't find we didn't find anything, I think, within that made any sense. So Yeah. Is it worth is it worth our effort? I think we need to look at it as you know, keeping keeping our pulse on things. I think it's it's worth the exercise. To keep us sharp in terms of all the new technologies that are coming in. I guess the primary thing. And making contacts, you know, Me: Yeah. Them: I think that's important as well. Okay. And I I I have a Me: From from the from the article that I shared with you, I shared with Earl and Dax as well. Them: sorry. Me: Where, like, even though YC has a brand and a name and they have a significantly better, you know, baseline startup mortality rate than the industry. From the viewpoint of of any investor that's writing checks into that batch at this high valuations, the distribution is that at least 80% of those deals are still gonna be below one one time return. Right? Like, you're gonna lose money on those. So the fact that these valuations are much higher, that means that you have to have if, like, for for Lobster Capital particularly, like, they only invest in YC companies. Like, you you are relying on more than one success per fund because in the past, like, were Them: Right. Me: getting enough out of the company that, you know, if one of the companies it out of the park, you would be enough to make up for the rest of the fund. Now because you're getting less ownership of it, that's just not enough. Them: Or they have to have a SpaceX level. Me: Yeah. Them: Of return. You know? So, I mean, their their model is flawed. Me: Yeah. Them: And but that's just what he's that's what he's doing. Their model makes perfect sense from their perspective, though. Me: Oh, yeah. From their perspective, is. Them: They have infinite money that they can raise. They just open up the open up the account and money flows in. And so what they do is they think about long term relationships Me: I see. Them: They're Who's Who's that? Y Combinator. So Yeah. They're thinking they're thinking that we're gonna take this smart kid from MIT and give him some implausible gargantuan task. Maybe he'll hit it out or she will hit it out of the park. But more likely, they'll come back to us for the second time that they that they go through this. They're building long term relationships and building a talent pool. Well, it's even more than that. They're making a return as well. Because, note, they're getting a huge chunk of the of the company for a very low amount of money. They're getting 7% for $250. Me: It's 500 now. Them: So Me: They put 500, a 100,000. Them: it's what? 250,000. Yeah. So it's 500, but they're get is it no. Don't think it's that high. Is it that high? Today we get that $500,000 guaranteed or or live I don't think it is. I think it's I think it's, like, in pieces. I think it's in two pieces. It's seventy five? It's $3.75. I was So we $3.75 for 7% is a good evaluation. There. Me: Return. Them: On the companies based on the the percentages. For the investors who are investing at those high valuations, they're not. Right? So for the people that are investing in the fund, in the YC fund, I mean, there's investors in the YC fund, I believe, but there have been in the Maybe maybe they use your own money from now on. Probably not. I imagine they still have outside investors that they're using. They have a pretty good return. You know, what is you know, what's that valuation? Pretty it's pretty high. Up. So I think that's still for them, they're making money as well. Okay. Alright. Let's go to yeah. You have a comment. I had a question. Have you all in your experience in venture capital, have you seen high evaluations like this before? Like, is this a trend that happens every five to ten years, or is it just because YC can do it? And on top of that, there's this AI craze right now. I mean, it happened in 02/2001 with web you know, with webweb1.o. So that's at least one point where it definitely happened. Actually, it happened also here recently in 2122 when the prices are just skyrocketing. And Is it yeah. It's it's it's abnormal. It's a bubble. It's and the thing is it's gonna and and the question is, will it ever come down? Because it's smaller investors for the most part. Mhmm. You know? So I think that they could probably keep doing this. If I so, add to that. If it's a bubble, what's going to be, in your opinion, the straw that breaks the camel's back? The problem with this is it pushes up the other values too outside of YC. Okay. And that and that will be a bubble where people wanna invest anymore. Right? I I think That's why valuation is so important. Yeah. I mean, the answer to what breaks the the goals back, you never know. But, I mean, there are the the things like, you know, like, Oracle right now is taking on 30,000,000,000 in debt every year to just do their build out. And eventually, someone will stop giving them that money if there's no revenue behind it. So and when when the first one of those guys says we're gonna have to cut back on CapEx because we we can't get, you know, we can't financing because there's no revenue. Somebody like that will make an announcement, and everyone will rethink their universe. Because right now, the mag seven and the AI giants are all just passing money back and forth between themselves. Yeah. That's circular funding. Yeah. I'm not I'm not I'm not sure if that really is as big a bubble as people are making out to be because I think that people are underestimating the need for AI. Compute. I think that there's there's the players like Google and everybody who right now is infinitely well funded and can afford it. And then there are the people like Oracle who can't. So some of those players should be in that game and some of them shouldn't. You'll have explosions in certain areas. Yeah. But for the most part, the big guys are gonna win big time because I I think a, they'll make money on their investment. B, things are just gonna keep growing. I mean, old chips now are going above their initial value. It should that shouldn't be. Just because of production. Right? People are gonna get their asses handed to them who are gonna now invest in ships because they see the value of the residuals on these chips as being very high, and they think that's gonna continue. Me: No. It's not. Them: And it will continue for a while, but it won't continue forever. It's almost it's a weird shortage dynamic. Like, if you went back to when during the pandemic, for example, you know, two year old used cars were going for more than their initial value because no one could get the chips that went in cars, so there were no cars. So you you know, there was a weird used car inversion that's right now every device that has memory in it like I said, people are buying Mac minis because they have RAM in them. And they sell for less than the value of the RAM that's in them. Me: Mhmm. Yeah. That was a great time. I I leased my car for three years, and I sewed it back for the... The the dealership wanted it back, and they paid me more. Them: Yeah. Me: Than what I paid for it. And I got it. Yeah. Them: Yeah. Yeah. During during COVID during COVID? Me: During COVID. Them: Yeah. Yeah. Yeah. There's all kinds of those kind of weird supply chain abnormalities, and then no one can predict how they shake out. Okay. So for the remaining meeting, we only have ten minutes. We're just gonna go through and allocate the companies if we could do them all. And then next meeting, we're gonna go right into the companies right off the bat. We're not gonna talk about any of the other stuff. Okay? So let's just go to access try to get to the new ones. We're not gonna even go we're that's all we're gonna do today if we can even get through that. Think we have, like, nine you said. Yeah. There's nine. That's gonna be hard in ten minutes, but we'll get we'll get to as many as we can. Okay. We're we're a ways away still from the next demo day of the lights are still YC. We also have the a 16 coming up. And then we also have Skydeck. I can the Skydeck, we're gonna kill all the Skydeck because I don't like any of them. Okay. But let's do let's do the new ones on the the y the YC. So we're killing all this side of the sky. Don't don't do that right now. Let's get through these things. Yeah. I got you. Okay. So the first company is Moka Tray. US stock perpetuals. That's global traders get synthetic twenty four seven exposure to The US equities. So trying to do this outpass. Yep. Next. Chime up if you don't agree. Okay. Next. Okay. So the next company is called our personal AI cloud computing for everyone. That's a good five or six of these already. These are, you know Yep. Cloud cloud instances of OpenCall. There's nothing special here. Jill, next. So we have Complair cause a self defense for physical products. Helps teams manage compliance testing and documentation for CPG and hardware. CPG. Se pedir, você I can see your products. See your package is good. What can you package? You can see your package kids. It's on my package case. Does somebody wanna look at it? Me: About the space. Them: Who do you think? No. Pass. Pass. Next. Ace power. So ace power. Smart energy monitoring for buildings. We've sent a few of these. We've invested in one in the previous clip fund. I wouldn't I wouldn't get into that. Or passing. Someone's interested in looking at it. I just saw a lot of these companies. Going once going twice. So it's a long sales cycle on this. Okay. I just dropped the link for the next one. It's called Lumias. A accessible three d ultrasound platform, and I'd expanding everyday diagnostic imaging access. Now we usually, like, shy away from medical related devices, but if this is of interest to anybody. Okay. The next company is Me: I just dropped the link in the chat. Them: conversion tool that increases website conversion rates behavior, and then they build new page variants. And try and do Me: That's it. Now you generated a UX Them: testing. Me: I had a friend who had a start up was doing exactly that, but they pivoted away from it. So I I'm not sure what was the reason behind it. But now they just do ChatGPT native apps. And that's kind of the direction that they went with. Them: Yeah. Think it's a pass now. I never heard the expression notion, but I like that. Me: Yeah. What does that mean? Them: Thank you. Red Ocean. It's, like, opposite of, like, a blowout ocean. You know? Blue bad. It's bad. Like, the opposite. Okay. Crowded unhealthy. Gotcha. Yeah. Something with algae. Okay. We have user land. It's AI agent that predicts churn months before it happens. I'm not sure what the unique data moat here is, and I feel like there's a lot of companies that do churn prediction already. Sounds like a sounds like a feature. Me: Okay. Yeah. Yeah. Yeah. But when you look at the the the founders profile on LinkedIn, it just says internal open claw for customer success teams. Right? So it's yeah. Maybe they they it's not personal assistant for internal teams, but Them: Alright. This task. Okay. The next one is standard signals, the AI run hedge fund. Where models research and execute trades end to end. It was kind of similar to a company we had done before called Kempton AI. That's Kempton was interesting when we went into it. You're right. It was, like, there's no model there. Yeah. I said I signed up for it. I'll play with it, but it's Me: Yeah. And, also, the Them: it's cool, but I wouldn't invest in it. No. No. No. Me: what was the one from Florida that I brought the the founder? Mean, he was very early on Chris, but it was similar. His his was also AI AI Them: Right? Yeah. Me: you know, Them: Exactly. Okay. Me: Sentiment. Yeah. Yeah. He his was focused on sentiment trading. Them: Recruiting agency. Yeah. I mean, Hire Glide, this the next one, this is Jack and Jill, so I don't understand, like, what they're doing that's anything special. There's a zillion companies doing this. Yeah. Kill. But, I mean, there's one with really solid brand recognition in it ain't them. Yeah. I just, I just had coffee with somebody who I I and I connected with him. He fought Me: Which had a big exit on the Them: on the web one o area. Marketing. So it was kinda interesting. I guess it's not important to go through it now, but yeah. Well, he he dead. Dead dead in the water. He thinks Me: Good to know. Oh, yeah. I'll I'll Them: have a good weekend. Me: I'll send you some information around potential leads for for your other idea. Alright. Not clear. Them: Okay. Sounds good. Thanks everybody for for looking at my alpha version and giving feedback. I really appreciate it. Sounds good. Well done, Mark. Me: Sí, vale. Them: See you guys.