- Duration: 39 min
- Platform: meet
- Source: Read.ai 01JCGZ67CAQATZBC6V4AFCCQF5 (https://app.read.ai/analytics/meetings/01JCGZ67CAQATZBC6V4AFCCQF5)
Transcript
[Jason Syversen]: Where are you based? [Michael Houck]: Are you in New York? Yeah, I’m in New York. [Jason Syversen]: What about you? How do you know Andrew? Just through his community meeting stuff. I actually never met him. He’s kind of a little bit of a Wizard of Oz figure. And so you show up to one of those events and you see a bunch of people and everyone’s like, who’s this Andrew guy? You never see him. And I just met a nice young woman one time and we hit it off. Found out that she was Andrew’s girlfriend. And at the meeting, she’s like, you really should meet my boyfriend, Andrew. Like, I think he’d really, you guys would really hit it off. So she made an email intro to him. We chatted. He loved my background and story. And we just kind of became friends after that. So found out we’re both fans of basketball and poker and strategy games. And so I’ve been playing polytopia on the phone with him. Organized a basketball meetup thing for founders and investors who play poker together. You’re in New York too then? No, but I’m in the city probably every month or two. I’m in the Boston area, so I try to make it down to New York. I was actually born in Hackensack, and so I still have family in Northern Jersey, and I like New York a lot. It’s a fun place to go because it’s one of the only places I can ever go where I can literally just feel heel-to-toe meeting all day long and walk to each one. It’s just such a cool place from a concentration of smart people doing cool things in the city. So I’ve really enjoyed it. [Michael Houck]: Yeah, I’ve been here for a little over four years. I was in SF for three before that, and LA for two before that, but originally from Philly. So back on the East Coast now. Very cool. [Jason Syversen]: Yeah, I’m going to be in us up for the Cerebral Valley AI Summit, I guess, next week. Yeah, so I’m going out for like Wednesday, Tuesday, Wednesday, Thursday. I’m trying to figure out when my next trip to New York is going to be. It’s not too far off. [Michael Houck]: If you’re definitely going out, shoot me like a little email that I can forward. So Ivan, who runs Cerebral Valley, is a good friend of mine. He was part of my founder So I’m sure he’d be able to show you the best spots to hit up while you’re there for it. [Jason Syversen]: That’d be awesome. Yeah, I appreciate that. [Michael Houck]: Would it help if I give some background on me? Because I don’t know how much context Andrew gave you or anything like that. [Jason Syversen]: Not much. Just good guy. [Michael Houck]: You guys should talk. Cool. Classic. So let’s see. So I’m Michael Houck. I did the big tech thing for a while after college. I was part of the early team at Uber Eats. It’s a very small part of growing that business from a couple cities. We were literally loading burritos manually into people’s cars in the early days in the parking lots of these restaurants. Then I jumped over to Airbnb. I was a product manager for Airbnb Plus, helped build that. Our whole Plus team got laid off during COVID, and I decided I’d had an agency business right after college, so let me take a venture shot this time. And so I built a company. Called Launch House. We ran co-living residencies for founders in big mansions in Beverly Hills, Chelsea here in New York, Tulum, Lisbon, Aspen, Miami, other places. Did that for about two and a half years. And we raised a venture fund to invest in the founders coming through the community, wrote about 50 checks out of that fund. And we also raised a series for the actual holding company from Andreessen Horowitz. I moved on from that company about a year and a half ago. Since then, I’ve been basically building up a personal holding company. So I have a couple media assets, including a newsletter that goes out to 60,000 founders every week. That’s kind of the creator thing is kind of how Andrew and I met each other. I also have a couple agency businesses for things like ghost writing and email growth. And then a SaaS product called Megaphone that helps you sort of get the word out about your content on social media with big creators and your niche kind of engaging with your content. So a lot of different things. More recently, I started to want to dip my toe back in the investing world. And so Ryan Hoover over at Weekend Fund is a friend. And he asked for some, hey, what are you looking at right now? And I said, I don’t know. Let me start talking to some founders again and see what’s going on out there. So that’s where this has come from I basically asked Andrew, who are some great founders I should talk do. And we got connected. So what we’re doing with Ryan, what I’m doing with Ryan is we’re looking forward to write 200 to 400 k checks, you know, sector agnostic, as long as it’s software driven, and pretty high conviction, you know, not looking to see like, Oh, which firm is leading, you know, all that kind of stuff. That’s, that’s us. That’s our background. And that’s, you know, town here. [Jason Syversen]: Very cool. Yeah, neat. Sounds like a really neat background. So is your undergrad like in technology or business or? [Michael Houck]: Yes, I did computer science with a business minor at the University of Miami, graduated at this point 10 years ago. Okay, great. [Jason Syversen]: You’ve done a lot in 10 years. 10 years post college then than I was at your age. So congrats. [Michael Houck]: Hopefully the next 10 years are just as crazy and wild as the last 10. [Jason Syversen]: Yeah, I think a lot of it, you know, you sound like you’re doing a lot of the right things about kind of how to, you know, increase our optionality and exposure to smart people doing cool things. I think if you do that, you’re, we’re going to keep having that success. I totally agree. [Michael Houck]: Um, well, look, man, I’m super excited to get to know more about you and obviously how you got to, to sports physio and sort of where it goes from here. I took a look through the deck a couple of times. I think it’s super interesting. I love sports. My wife, to her detriment, I love watching sports. And I’ve been around basketball, played basketball for a long time. So I’ve actually been thinking about going to one of Andrew’s things here in New York. So I want to hear the story and then the background, and then we can dive in. [Jason Syversen]: Yeah, sounds great. Yeah, so I may be one of the only founders that I feel like my backstory is more interesting than my pitch deck. I was born in Akinsack and lived there until I was in Bergenfield, New Jersey. Some guy in a pickup truck tried to abduct a girl living on our block and circled the neighborhood three times, trying to find her when she ran away. I thought that was cool because I got to go to the police station. I’d memorized the license plate and got to meet the detectives and tell them about it, but my parents were not as excited over the situation. They were about a program to voluntarily fingerprint children at the school and my parents were like, you know, I don’t think this is where we’re going to raise our kids. So they packed up and moved to rural Maine. My dad was a mechanical engineer, started working for the paper mills. And when they all were shutting down and those jobs were going to China and India, I mean, Canada, he ended up out of work. So we lost our house to the bank, bounced around. He’s cutting Christmas trees, selling them in parking lots. He’s a machinist. He’s working at a state. We’re just on welfare. We were homeless when summer, living in a pop-up camper in someone’s backyard, food stamps, not a whole nine yards, cutting frozen blocks of wood in the middle of winter so we’d have heat. I got a free ride at University of Maine for computer engineering, stayed there for college, and then came down to New Hampshire about an hour outside of Boston, working at Lockheed Martin, which is their electronic warfare division. I built jammers for aircraft and SIGINT equipment, intercepting communications. I started hacking company network first week on the job so I can get some software I needed to do my job, which led to more than one visit from HR to my desk and also a reputation with the IT team as the hacker du jour. Sometimes I would get messages, hey, the network’s down, is that you? I’m like, I’m not doing anything. I ended up taking over the corporate network of 5,000 people at one point out on a dare from the IT teams after I had pointed out some And they were like, I’m telling them this problem. They’re like, no, you can’t do that. I’m like, yes, I could. They’re like, prove it. I was like, give me that in writing. I will do that. So ended up starting this mobile warfare group inside the company in my mid-20s. Built that up to 10 or 20 guys. Had an older gentleman who was a director, and I was like the lead whiz kid tech guy. Cool story, which I won’t go into, but I got recruited by Cell Team 6 in 2005 to join Dev Guru in Virginia Beach. Didn’t take the job, but two years later, was blessed and very fortunate to get recruited to go to DARPA, which is this building here. So ran $100 million portfolio of classified programs, all special access black programs, transitioned those to Air Force, Army, Navy, CIA, NSA, another classified DoD group that officially doesn't do anything in cyber. In 2009, and started my own company, Siege Technologies, which was this company up here In 2010, we did a million in revenue, bootstrapped the company, mostly doing bespoke R&D, commercial defense intelligence contracts, building cool tech. We retained all the IP to all the products we built, licensed some code. We spun some stuff out to a venture-backed firm that raised 18 million. We did a lot of stuff with DARPA, Air Force, intelligence agencies. I sold in 2016. To our private equity-backed firm called New Maya Security, had an eight-digit exit in that transaction, took enough to live on, and donated the rest to a charitable foundation called SMA. My wife and I are both Christians, felt strongly about giving back. Our church had built an orphanage in Haiti that cost $8,000 to build and $500 a month to run and had 32 kids living in it. And that was really inspiring for the kinds of impact you can have if you're careful with where you allocate funds. I retired in 2019 after Golden Handcuffs came off and did that full time. So I was chairman of the board, chief investment officer for the charity, and then also took over Tenex Venture Partners, which is an angel group, started a small venture fund. The charity is a limited partner in the fund. And so I've been investing out of charity assets in tech companies, as well as a fund in my retirement family office. I've done 42 investments in tech startups at And then five in real estate and private equity. Ran for state senate in 2020, lost by 3% to the incumbent, which I was fine with because I'm a hacker and really had no interest in politics. [Michael Houck]: What about your own? No obligation. [Jason Syversen]: I really felt like, it's like voting. If you don't vote, you can't complain about who's in office. It's true. I, in my district, was running unopposed. And I really felt like God was like, go do this thing. And I was like, have no interest in doing that, but I was unique in the developed world that we don't pay. They pay a hundred dollars a year to be a Senator in New Hampshire, even though it's a full-time job six months out of the year. So the only people who can do it tend to be older, retired people. I worked for a special interest group that was willing to pay his full salary while he worked in the Senate. And then he voted on issues related to the special interest group. I was like, that's kind of gross. Like, who do we have here in Africa? [Michael Houck]: Like a banana Republic here. The answer is yes. Exactly, right? [Jason Syversen]: And I'm like, here I am, this like, like, you know, financially independent, middle aged tech guy, who's built companies, and we're foster parents, we've adopted our twin boys, I came from poverty, it was a cool story. I'm like, and I'm not right, no one's running against him. And people are pleading me to run. I'm like, if not me, who? Like, so I didn't want to do it. But I was like, you know, I think it's the right thing to do. And so I ran. And I got out spent $200,000, $100,000 and lost by 3%. But again, after that, I was free to do other things and was doing the investing thing, but I needed to slow down from a portfolio theory perspective. Thought of raising a fund, like a larger venture fund, like the fund I run is only like 500K or a million bucks or something. And so I was thinking of raising a $20 million fund, but I haven’t grown a pair, I haven’t bootstrapped my company. I only knew a handful of like super wealthy, it could be LPs. My angel portfolio was pretty early at the time and needed some seasoning. And I was like, you know, people are pegging me as a hacker cyber guy. I should start another SaaS business and raise venture capital, go through the whole process. It’ll make me a better investor because I’ll have actually raised money before. I’ll actually build a true product company and not this hacker R&D thing. And then God willing, if it goes well, I can either do a larger family office where I can hire a team and go back to just investing for the next couple of decades and or raise a fund if I want to. And yeah, I looked at a bunch of options. Obviously started with cyber, looked at other software companies and B2C, B2B, and settled on sports physio. I’ve been playing basketball for about 30 years. I’ve coached for almost 20, probably 17 years, from kindergarten to high school, both soccer and basketball, although mostly And yeah, I had the idea, called some customers and they were like, oh my gosh, we would love this. We’d pay good money for it. And then I called some PhD nerds from when I was at DARPA and they were like, you could totally build this. And we’ve been like 50 people for 10 years. This is a couple of people on a weekend. And they’re like, no, you know, it’s two, three high-end people for, you know, three to four high-end people for a year to get a MVP out the door. I’m like, oh, that’s exactly kind of the range I’m looking for. I can bankroll that myself. And then I looked at the market and I did a bottoms up analysis of all the customers in the state of New Hampshire, came out to like $500,000. And then I multiplied that. So there's 1.36 million people in New Hampshire at the time, there's about 366 million Americans. So you do the per rata multiplication and came out to about $500 million domestic market for basketball, which is about one third of the global sports spending market and one third of the small market, so it’s about $1.5 billion market for basketball. Volleyball adds another $400 million. You add in soccer, baseball, football, lacrosse, it’s an $18-ish billion market. Call it $10 to $20 billion market. That's a venture scale business, right? It's not cyber, which is $100 billion plus, but there’s not thousands and thousands of companies competing for the same market share. I’m competing with a couple of volunteers low-end hype manually pay people or a couple of private equity companies that pay thousands of people in Nepal, India, Pakistan to type in a computer and send it back. That’s what Huddle and Synergy do, or StataLinux, which are kind of the incumbent competitors in the space. I’m like, I really like targeting an industry that’s a few billion dollars, you know, billions, low billions of dollars, and I’m competing with legacy players that do things manually, and I can leverage my tech background to build some advanced you know, technology to build something disruptive. So, yeah, I started a company three and a half years ago, put a half a million of my own money in, built an MVP, raised a pre-seed round with an AI fund out of Boston for two and a half million, got the product built and launched last summer. We raised a seed round with Sapphire Ventures out of San Francisco. We have a $180 million sports tech fund. Which led the round, both Hyperplane and Sapphire kind of on the board or joining the board. And then we were only doing about 10K in revenue by the end of last year. So we let our head of growth go. It was like an SMB guy that a friend knew that we hired. And he's like, I'm a scrappy startup guy. I'm like, well, we'll see. But he's really more like a marketing guy for your lumber store, not really a venture backed startup. I'm like, yeah, you're not the guy. He's going to help us scale. So let him go. And I did a search and ended up hiring Sean O'Connor, who was global marketing director for Microsoft. Basically, he ran Western Europe for them. And he's got a sales background, sports guy, super sharp. And he's helped us scale from 10K last year. We're at 375 right now. And we're tracking to break 700 by the end of the year. So pretty good growth from last year. Volleyball this month. We're launching a coach's product later in this quarter, and then we're launching a premium product first quarter next year. We have a baseball API thing we've built. So yeah, that's kind of the story. And then bringing the full circle, my goal is to take whatever I make from the company and pay myself back, obviously, and then drive the rest into nonprofit causes. Our CTO and COO have both had seven digit exits in previous startups, and have pledged to also give whatever they make above what they need to retire on. One of them lives in SF, so his retirement budget's like, not too good, because everything is expensive, and our COO is not as far away. He lives in New Hampshire. But they both pledged to give above what they need to charity as well. So ambition-minded folks that love sports, but must have a deep tech background, and trying to apply that in a specific vertical? [Michael Houck]: I think it's a great story, Ben. And I think the market makes a lot of sense to me. I've seen folks who've gone after this, but mostly with, I think, what is a more manual approach, sort of similar to what you're saying. My curiosity is, why has someone not taken an AI-first approach to this market? You see images of the little boxes around players and sports. You see that, but it sounds like what you're saying is, hey, maybe that's not actually as AI powered or technology powered as it could be. [Jason Syversen]: No, it's a great question. And you're absolutely right. People have tried to do this with AI. I don't know if you know Basu from Courtside Ventures. He's a big hoops guy in Manhattan. Oh, yeah. [Michael Houck]: Yeah, I know of him. We don't know. [Jason Syversen]: Yeah, he started a company called Crossover that does what we do manually. They tried to do this with AI. Seven or eight, probably eight or nine years ago at this point. And he said, it just didn't work. Like there was no chance it was going to work. So they gave up. He said, they'll lead our A round. You know, if we get the tech working, he's very bullish on the opportunities. Like, Oh yeah, you guys will crush it. If you're actually successful, you go cross over to like 50 million revenue and sold the huddle. And they were doing mostly manual, but he's kind of pissed at huddle because he said huddle basically destroyed the business and drove them in the ground. There have been people who have tried to do this. And the reality is, and I've learned this too, right? I'm not going to say that we have really smart people and we've worked really hard and I think done a good job. But it's hard. It is not as easy as just plugging something in an LLM. We built our own algorithms and wrote our own code. But it's not 100% accurate still. I'm not going to pretend that we We've solved all the problems. The gap between a prototype and a full 98% production grade, consumer grade, any gym, any venue, numerous courts, bottles on the sideline, players walking on and off the court, crowds of people, layers of basketball hoops in a venue with a bad camera angle. It's a hard frigging problem. [Michael Houck]: Yeah, it reminds me of the self-driving car problem, which I know a little bit because we were working on it at Uber when I was there. I was friends with people on we're building that stuff. That was 2016, 2017. And finally, now it looks like Tesla and Waymo. [Jason Syversen]: Uber still hasn't gotten there, right? [Michael Houck]: Dara, as short-sighted as he was, sold it off to Toyota, whatever. But yeah, I think it's a lot of complicating factors to accurate analysis. So I guess, what are you guys doing to get past that? [Jason Syversen]: Just keep working at it. And basically it's like more training data, you know, improvement, you know, you move, uh, it's like a decaying exponential in terms of, um, return on investment. So, or, you know, so you, you start off and you're all excited because the first six months we're like, oh my gosh, we went from nothing to full amazing prototype. The demo is really well in six months. And we're like, we're cooking with God. We'll be ready in a year and a year. And we're like, yeah, we've got all the we could do rebounds, blocks, like, or not blocks. We could do rebounds, steals, assists, turnovers, two pointers, three pointers, free throws. But it was like 60, 65%, you know, jersey recognition, player recognition, 2D homography, but like 60, 65% accuracy. Like, okay, well, it's a prototype demo as well. Well, let's keep going. And then like another year goes by, we're like, oh, now we're like 70 something percent accurate. I'm like, okay. This is getting hard, we can't keep screwing around the lab guys, we're gonna launch a freaking product. [Michael Houck]: Right, like where's the asymptote there? [Jason Syversen]: Exactly, so we launched the product last summer and we basically said, look, we're gonna have humans clean up this data because they need, the tipping point is you gotta be 90 something percent accurate to really be competitive with humans. And we're in strike, we can see that, but we're not gonna be there for another year or two. And the investors I was talking to were like, yeah, this is cool, What about market fragmentation? How do you scale? What's your go-to-market strategy? Can you sell? They want to believe. How much are people going to pay? Sports ball? Is that a market? How big is that market? You get these nerd investor types you just don't know. Even our CTO, I didn't even get him on board until a year in because he's like, sports ball? People are going to pay for that? He's like, I wear a propeller on my head. He's a self-proclaimed propeller head. He's like, I don't know if there's a market. After he saw the traction and learned more, he's like, oh, yeah. Oh, I think it is a market. Like I'm seeing you get traction and seeing what you're building. So, so yeah, we launched, we're still like 80 something percent accurate in basketball. Um, and so we're still paying humans and we can deliver our, our gross margins are still profitable even with humans cleaning up the data. Uh, and all of this stuff we're doing every game, every customer we're training, right. We're getting smarter and the algorithms are getting better, but you know, you spend three months worth of work. To gain 4%, you know, another three weeks or more to get a half a percent. So you're just knocking down, Oh, it has a failure mode in like these types of Jersey colors or this lighting environment, or this kind of camera angle, or this kind of basketball condition, you know, don't, it doesn't recognize as well as this play. So like you build a training, you bang it out, you spend three weeks. Oh, okay. Now it improves. And we went from, you know, 89% to 94% in handling dumps. Like, okay, cool. Now move on to the next. Thing and just knocking it down. So for me, like some people get nervous, like we've had a couple investors pass, but they don't like capital raised to revenue. And I was like, dude, like, to me, that's, that's a benefit, right? Because the cost that we've put in, we are building a competitive moat, the IP, when I hear people say they're going to compete with us, and they want to build this with a like ball or TV, we try to use them as a partner, they were trying to lowball us on price. And finally, they said, yeah, we're going to build this ourselves. Like we have a PhD, and he's just gonna go build us. I'm like, you do that, buddy. Like you get back to me in two years. And you tell me how you're one dude is going to redo and do all the stuff that we've done with three years worth of work. [Michael Houck]: Like, good luck. Yeah, no, I, I buy that argument, the sort of building the competitive mode doing the hard work. [Jason Syversen]: It's like the self driving thing. Like Tesla has this massive competitive edge that only Mo and like Google and a couple of companies are going to to be able to run with, but GM is way behind. These guys are not even in the ballpark. Subaru, I mean, they got nothing. [Michael Houck]: Yeah, 100%. I would even argue that Tesla has a competitive advantage over Waymo just because of the data collection capabilities, because they have all these cars on the road. They're using LIDAR. So I guess my question to you guys is, when I think about sports, the data set that you guys are training this stuff on, I've been thinking a lot about, how the rumor is that GPT five or Orion or whatever they're going to call it is actually not going to be that big of an improvement over GPT for relative to four versus three. Yeah, because it's scaling law, right? There's only so much publicly available, scrapable, purchasable data out there at some level of, you know, size, you run into that limit. So I don't know what that size is for sports video. And like, how much of that you guys could have already consumed. [Jason Syversen]: Yeah, I mean, it's a great question. For us, it's less about the quantity of training data and just about the hours to solve all the corner cases. For us, it's the handling of jerseys in this situation. I'll give you a great example, a pet peeve of mine that I've been beating up our CTO about. We still don't have physics-aware 2D homography. So I don't know how much you know about computer vision. So occlusion is a big challenge. Computer vision, that's when something gets blocked by another object. 2D homography is when you take a 3D image from a visual receptor, and you transpose what you perceive onto a 2D plane, and then you can plot objects on that plane or lines. You use that to define the boundary conditions of a court. So you want to know what the three-point line is. You want to know what the out-of-bounds and in-bounds are, so you know which players are on the playing surface, which are off, because you've got a bunch of players sitting right on the sideline. So we have been doing our action detection, and our player tracking, and our player recognition using pure optical trackers, which pisses me off. Because all that does is it looks at pixels, and it does a running counter with a Kalman filter and player models. And it tracks like a human using those bounding boxes like you see. And it'll do OCR to try to detect the jersey number on the pixels, which works great when there's no occlusion. And we are like 99% If it's like one, two, three people on a court shooting hoops, oh yeah, we can do that all day long. As soon as you get 10 people plus two, three refs and maybe a bunch of people on the side and they go around each other in a scrum and then he pops back out. Well, the model doesn't read, identify that player until they wave their Jersey and you get to get it on the camera because we're not physics aware. So what I was pushing the team on and what we haven't done yet, and this is going to be an opportunity, which I think is going to improve us. Like another 5% or 10% is our model needs to take all these objects, transpose them into a 2D plane, plot their position on the court, build physics-aware models. So when a player, number 22, goes behind number 17, we are tracking him in a 2D surface state space model. So that way, when they reacquire a player over here, we know it's 1.7 seconds afterwards. And he's moving from right to left at approximately like eight miles per hour. And then on this other side, guess what? 97% probability of confidence that that guy is number 22. And you go behind him and come back out. Our pixel detector doesn't do that because it doesn't have a concept of depth. It doesn't have a concept of physics. It doesn't have a concept of like surface and behind what's occluded. It just counts pixels and the pixel counter is great. What it sees, but it doesn't model things it can't see. Right. So our, team needs to build this physics aware tracker, which can track objects that get occluded. And I think when that happens, our detector and our reacquiring model for players after occlusion is going to go up significantly. That'll improve things like minutes. That'll improve things like layups and steals that happen behind when occlusion occurs and drive our accuracy up. But that's an example of a model we haven't built yet that the team needs to, we haven't homography model now, we haven't implemented the physics and we're in the final training of making sure that the homography model is accurate. And then we need to add this physics-aware model. And I think that's going to be an example of a great pickup. So it's that kind of stuff that just needs to get built. But we've been operating with, you know, we started with a team, like I like to joke that we've laid off more MIT PhDs than most companies will ever hire. We've already hired two MIT PhDs with 20 years experience, both of whom I let go because they were kind of expensive and they're slow and they get caught in rapid trails and they want to think deep thoughts on things. And I'm like, guys, we're shipping product here. I already have two PhDs. I don't need four. You know, our AI lead was chief scientist for AI for BA systems, North America. Our CTO was a colleague of mine at DARPA. He ran a 200, $250 million portfolio. He briefed a SecDef. One of the smartest guys I’ve ever met. I’m like, we don’t need more PhDs. Like let’s get rid of these guys, bring in more like master’s degree undergrad, like Python on coder types who can just write code and implement algorithms and get more training data. It’s the other thing. Early first year or two, we just didn’t have a lot of data. So we’re like, we have algorithms, we’re going to have training data. Now we finally have some training data. But then they were like, just trying stuff. And I’m like, OK, we need ground truth. So every time you build a new algorithm, I want to benchmark that against known ground truth training so you can compare improvements. So now we finally have that. So we’ve been building processes and improvements and refining the algorithm. And so it’s a lot of the engineering work, but the good thing is we’re there. We have training data sets now. We have a strong team. We have three full-time AI engineers, plus our CTO, plus the AI lead who’s part-time working on patents. We have a five-person AI team. And we’re at mid-80s. We’re at like 99% for court detection, basket detection, ball tracking. And then the jersey detector is like 87% and our action detector is like 95%. So you multiply those through, it ends up being a cumulative probability in that like 85, 86, 87%. Our customers really want to be 95. So that’s why I’m like, you know, the bridge round that helps us scale a revenue to two, two and a half million next year and gives us a little more time to improve that. But by the time we do that, then I think when we raise the series A, like I’m pretty confident that we can get the cashflow positive. Because once we go full AI on these models, our margins go from like 10, 15% to like 50, 60, 70%. And, you know, scale that revenue from two, two and a half to five, six, seven at 50, 60% margin. You know, we’ll have a cloud storage thing for premium. That’s 90% margin. We have a coach’s portal. That’s 80% margin. An API product for baseball, that’s high margin. So like, I think by the time we get there, middle of the A, mid to late in the A, we’ll be, you know, netting three, four or five million bucks, you know, at that point, and we’re like cash for positive, and we’ll probably still keep raising money. But at least we have the option, right? [Michael Houck]: We kind of control our own destiny. Yeah, it makes total sense. And always good to have that optionality, especially these days, more founders are preferring that, which I think is smart. Just real quick, I do need to hop, but real quick. Give me some context on the bridge rounds. I don’t think it was actually, I guess, fully laid out in the deck. So just curious, like, how much are you guys raising? How much is left in the round? What’s the cap on the note? [Jason Syversen]: All that kind of stuff. Yeah, I do have to jump to actually. So we’re doing a bridge round. We did a 19 million posts last summer. When we were launching the product with Sapphire, we’re at 23 million posts now. We 1.1 was taken by insiders. They took like two or three X their pro rata. Our Hyperplane, Harvard Alumni Group, Sovereign’s Capital have all taken at least two X their pro rata portion, which got us the 1.1. So we have, and we have new money that’s come in. Andrew came in, Alumni Venture Group, a few others took about 100K. So we’ve got about 900K available. And we have some folks we’re talking to right now. That are random smaller checks. We also have a couple of bigger checks and a customer play on sports that has said they might want to take the whole round. They’re talking to us about kind of a distribution partnership. So yeah, we’re kind of trying to see what our options are and get some good folks on board and wrap this up this year so I can focus on the product for six months before I go back to racing series A next summer and close that out next fall. [Michael Houck]: Cool. Okay. That makes sense. As I mentioned, I think we’re 200 to 400 We’re pretty high conviction. So we wouldn’t need to wait on the Yeah, it sounds like they might be the lead for the round the pro rata. [Jason Syversen]: Obviously, it’s exciting to hear two to three X pro rata. [Michael Houck]: That’s right I’ll chat with Ryan. I’ll send a note tonight. We’ll chat tomorrow get back to you ASAP You definitely hear from us before the end of the week and we can go from there. [Jason Syversen]: All right, sounds good my crew Look forward to talking again, right?