Optimizing College Matches: Troove CEO David Hurwitt on Boosting Graduation Rates

0:28Welcome to the AI Chat Podcast. And about three years ago, started this software-based company to help people get into college. That's super fascinating. Making that kind of career, I feel like was that a bit of a career pivot? Had you done anything in software before? Was this a completely new move for you? Well, if you think about it in the context of what I do versus where I do it, it's the same thing I've been doing all along, which is to see a customer need and to try to fill it with a better product or service. I have applied that literally in everything from, from turbines to toothpicks.

1:01I say sometimes windmills to washing machines have some of the things I've worked on, but you've worked a little bit on for general electric. Is it right? I work it on washing machines. Actually, well, I worked at general electric on some other stuff, but I worked at Whirlpool on the washing machine business. Okay. Okay. That's what it was. I was seeing that, you know, we, we looked at, I was the head of the Whirlpool washing machine, the brand manager on the Whirlpool washing machine business in 2000. And Whirlpool was the largest producer of front loaders in front loading washing machines in the rest of the world, but didn't sell any in the US where they were the number one company in top loading washing machines.

1:37So I asked the sort of simple question of why is that? And is there a reason why we should try to change that? So we did some fascinating research about why Americans might want to use a front loading machine, why they weren't already using them. And then we actually just developed a brand new product, which is that large format front-loading washer that's in probably 30-plus percent of American homes today and has done over a billion dollars in sales by simply changing the format and bringing this technical advantage to the American market. That's super interesting. And so what I'm feeling is I know you're working on a new startup now, a new company, which is Truve.

2:14And I'm assuming you took a little bit of that thinking over to your new company. But I'm curious, what inspired you to start Truve, inspired you to get into that market specifically. It has nothing to do with, you know, washing machines or some of the other, yeah, some of the other areas. I'm curious. Yeah. You know, I originally worked in admissions when I was in college. And then when I was in business school, I was on the admissions committee as well. So I was reading applications. And I'm old enough for that to have actually been right at the beginning of the internet in the world. So I went away from it for about 20 years, had a child, built my career doing some of the kind of work I just described, and then came back into this as a parent and said, I'm expecting the process of finding and enrolling in the perfect school for my daughter to be decidedly different because we live in a different world today than the one in which I applied to colleges.

3:09Right. Typewriter. Right. Right. And what I found was grossly disappointing. It was really the same old thing where she was entering it into a keyboard, but it was the same limited information about a limited number of schools. The information flow, the process, the technology was all terrible. And just, you know, quick anecdote that kind of really got me thinking, taking her to visit the first school on her list. And we put a lot of thought into exactly what should be on that list. She had a lot of advantages that most kids don't have in terms of two parents who went to college. a good guidance counselor, a private guidance counselor as well.

3:44And despite that, we got to the first school where we thought we'd hit the bullseye with the kinds of things she was looking for. We drove onto campus after a six-hour drive. And she said, no, no, no, Dad, don't even get out of school. Or excuse me, don't even get out of the car. This is not school for me. I said, well, look, let's go ahead and get out of the car, take a tour, and then we'll go to dinner and talk about it, right? And so as we're driving to dinner, I said, all right, tell you what, you jump on Spotify and find the perfect song to lighten the mood. And I had jumped on Yelp to find the perfect restaurant in this town we didn't know.

4:18And we got to the perfect dinner with fantastic music. It was just right for the occasion. And it occurred to me that the technology used to match us with those two relatively consequential things was so much more sophisticated than the technology we were using to match her with what in the US is potentially a quarter million dollar investment. And so I really started to think about how we might bring other technologies perfected in other industries and in common usage today into the decision about what am I passionate about? Where do my people go to school and where should I apply and where do I have the best chance of not just getting in, but getting out with a degree?

4:58I love it. So tell me, well, I guess give everyone a little bit of context and background on what Truve is doing and how you guys are currently using AI. and then I'll have a couple other questions for you. Sure. Yeah. You know, we are a two-sided platform for students and for schools. And we operate under the simple premise, like I said, that A, the goal is not so much just to get in, it's to get out. And if that's the case, then it's pretty self-evident that the largest part of every school community are the graduates. They are the people that have found the success and happiness in that school's unique culture that might be of interest to prospective students.

5:36And so we can use data science and use AI to go quantify the culture academically and socially of a school and then make that available to a student to take a quick, you know, in our current format, they take a short quiz about their learning and social culture preferences. And then we're able to say to them, the people that have the same kind of answer patterns that you have, have found happiness and success in these schools. And this is where you should start your college search, because you'll find that as you get engaged with that school, it's going to be check, check, check, check, check.

6:12These are all the things that just really light you up and say, wow, this is incredible. And more likely than not, you didn't know about this school in advance because it might be 50 miles away from home, might be 500 miles away from home. But most of us have such a limited perspective on what schools are available to us that we miss out on a bunch of amazing opportunities. So that's our goal is to use AI to better match you with the outcome that you're looking for, which is graduation. And we believe that the right way to do that is to match you with the people that have already graduated because they are the model for success in the same way that AI has shown us that the right place for me to go to dinner is where people that have my same tastes and preferences have already gone to dinner.

6:53Very cool. Very cool. I love that. I think that's such an amazing application. So talk to me a little bit about the technology that you guys are using in that. I understand you've been working on this for a few years now. When you kind of started Truve, was the idea just being like, hey, how do we match people with a quiz and a survey kind of thing? I know there's a lot of very simple technologies. They got a database with a couple of check marks and you take a little quiz and it'll just kind of match you with something. Or right off the get-go, were you looking to do something a little bit more algorithmic, a little bit more in-depth with an AI play on that?

7:27Yeah. So we, you're absolutely right that there is a very simple, think about the Venn diagram of this, right? Do I want to go to a big school or a small school, city school, country school? And that creates a sub list of the schools that meet the criteria that I, I requested on a very sort of limited demographic kind of basis or geographic basis. And that's, that's where we started by necessity. That's sort of where you have to start. And that's where the industry is today. Right. And, you know, we are a bootstrapped startup. And so we don't have massive amounts of technology and budget to be able to invent a lot of things.

8:02This is about plugging in into this market technologies which have already been proven and pioneered in other markets. So we are in the midst of that transition now. If you think about the acceleration of AI technologies in the last six, nine, 12 months. And as we look at our product plan over the next year, there is obviously phenomenal application here to take friction out of this system, where right now, as you suggest, we started with this pattern matching between what happened at the school for its alumni, recent alumni and current students, and what a prospective student may want. The AI side of this gives us the ability to eliminate that friction or greatly reduce that friction.

8:46So we can go to a school and say, here is your pattern. Here is what we have found to be the pattern of success academically and socially from your graduates without you having to lift a finger. And then here for the student, instead of taking a 10 or 12 minute quiz, they could take a one or two minute quiz that is much more kind of a generative AI kind of experience that would really be exciting for them. Yeah, that's very cool. And then take over the control of that. And part of this, Jaden, is we want to get rid of this junk mail model that the U.S. higher education system is built on. And the junk mail model, what I'm referencing there is if you I'm not sure where you went to university, but most American students are absolutely deluged with spam and physical mail from colleges and universities that have historically just sent them massive amounts of junk.

9:39Right. understanding that some of that's going to get through and that will create their class. For a lot of reasons, and that's maybe another show, that model stinks. And they all know it. The students know it, the schools know it, but they just don't know what to do. True is what you do to fix this, right? And as we create effectively a online dating app that allows a student to very quickly take a quiz and to match with schools and then approve the schools that they like, you know, swipe right on the ones you love. And then they can, that opens the portal for the schools to talk to the students directly online, as opposed to us saying, we're going to sell that students matching information to a school that puts it in their CRM system, and then blast spam at them until they finally respond.

10:21Yeah, where the industry is now. And that's part of what has to get fixed with technology. Yeah, that makes perfect sense. I think that's a pretty awesome application. And I think also your strategy of, you know, you obviously have to with these startups, you got to start somewhere and then you kind of build into the AI aspect. I had a similar thing on my last startup where it was a, it's kind of like a mindfulness app. And then with kind of chat GPT launching and a lot of other things, we started kind of building some of our own technology. But then also harnessing, you know, OpenAI's API into chat GPT.

10:52And we built out essentially an AI life coach, a pretty sophisticated tool. But it was, it's so powerful where you can plug into some of these existing products. And I think that the software companies that will succeed over the next five to 10 years are, well, we've already seen it with the major corporations, but even when we come down to the smaller and indie-sized companies, it's going to be the ones that, whether they were doing it before or not, they're starting to leverage all these new AI tools and integrating them to scale and improve their product. When you, and actually have a good idea for you, but I'll give you in a second of someone I've seen in an interesting way they started integrating.

11:27But I'm just curious what your, where your mindset is and what your vision is on this. Like what AI technologies are you planning on integrating or using? Or have you thought about how that implementation kind of looks beyond just the concept of AI matching people? Yeah, I'll give you one kind of peek behind the curtain. One of the areas, one of the reasons why this industry, this college search process is so difficult is that there just are not enough great guidance counselors out there. The average ratio of guidance counselors or rather students to guidance counselors in the United States is 475 to 1.

12:04So the average guidance counselor in American high school is overwhelmed by the number of students they're trying to help. And they're giving academic counseling and behavioral counseling and college counseling. So it's a really challenging role. So one of the things that we want to do is to be able to start with this really simple, engaging kind of dating app looking quiz for a student that allows them to express themselves quickly and easily in an entertaining way. And then to be able to actually speak to a school and speak to a guidance counselor about those results. And I would say the relatively new ability to have that actually be AI, as opposed to saying we have to find and train physical guidance counselors to start to fill that role online, a la telehealth kind of model.

12:53So if we could do that generatively, and if we could have, for example, a new school, a prospective school, you know, if a student matches with the University of Vermont, and then a kind of a bot from UVM says, you know, hey, Jaden, it's so great to meet you. I appreciate that you swiped on us. We love to see the match here. You know, we see you want to do this, this, and this. Would you like to speak to somebody directly? Would you like to speak to a student who majors in what you want from your hometown? You know, we have a few of those. There's so much you can do. But our ability to then sell that to the school is significantly enhanced by our ability to automate some of those processes.

13:31So the school can literally just sort of step into that as opposed to us saying, buy this speculatively. You can say, here it is. It's already working. And now you can tap into it. And our ability to extract value from that goes way up. Yeah, I love that. And I love the, I also love the fact that with, with a model like that and a play like that, um, you know, let's say a student matches with a school, they can of course go talk to a guidance counselor over there, but you could also get just literally an API to chat GPT and give it like a page long of instructions or a couple of pages that are like, Hey, you are a guidance counselor at, you know, the university of Texas.

14:08Here's all the statistics about the university of Texas that people usually know, right? Whatever's already in your vector database, you plug it in. And it's like, what questions do you have? Are you interested in this? You know, we saw that this was on your quiz. Have you thought about this? And it's just, it's essentially just pulling everything off of there, but having a really fluid conversation with them, which I feel like is the way a lot of people will be gathering information in the future is beyond just like, you know, instead of getting a two pager from a university of all this information about them, where not all of it's relevant, being able to just sit there and kind of query it and ask it a couple of questions specifically and get responses and natural language would be really beneficial.

14:44So that's very cool. And then integrate that with my calendar. Like, tell me, what are all the filing dates? Like, so I love this school. Do I want to apply early? What does that mean? And if so, what are the dates that matter? And there's a lot of complexity in the process of applying to college in the U.S. You know, about the financial side of it and filling out these forms and getting your parents involved. So there's a lot of stuff that tends to bog down most students. And the interesting thing is I got into this a few years ago. I thought the problem was getting into college in America. But that's the storyline that the press wants to tell you, right?

15:19You hear about Harvard and 3 % acceptance rates and Stanford and MIT and so hard. Those schools are less than 5 % of the total market. The average acceptance rate in the U.S. college or university is about 70%. The problem is that the average graduation rate, the on-time graduation rate at four years is 45%. And so the problem isn't getting in, it's getting out. And when students run into, and a lot of that I feel is driven by misfit with where they go to school. And so when they run into the inevitable problems and the challenges of being 18, 19 years old, being out on your own, trying to fend for yourself, having money problems, having relationship problems, all kinds of things come at you.

16:01if you are not in your right place, surrounded by your kind of people, you're much more likely to bail. Do you have any, I know like, you know, new startup and you have obviously kind of this niche of being like a match between students and college and stuff. Do you see your roadmap expanding into helping students like once they get into college or like going any further down the pipeline? Or do you think you'll stay pretty siloed on just the kind of admission process? No, I would love to see this expand. You know, in its essence, what we're trying to do is we're trying to use trying to digitally quantify the culture of an organization in order to put the people that have the greatest chance for success in that culture into it.

16:45Right. And so that could be if you follow the education training, you could go up to into grad schools and other things like that. You could help with student persistence and show schools. These are the things that your graduates found to be important. So do more of it or find early intervention for, you know, for students who are struggling with that. But it goes beyond that. Right. You can say, well, these are the companies where your particular passions and interests and values and priorities would have the greatest impact. And so these are the companies where you should consider working. And these are the career types.

17:18And this extends to major selection as well. You know, I don't know of a single teenager that's ever used the verb to major until they got into the college admissions process and tried to figure out what that means. There's a lot of pressure around that. So helping students to find their passion and then showing them what a career might look like and then showing them the kinds of schools where people with those same passions have found success. And then to follow that into these are the kinds of careers, these are the jobs, these are the specific companies where your particular passions, interests, talents would be a greatest impact, I think would be a huge game changing value for people and for companies.

17:58So being in the kind of education and AI space, I'm curious because I've heard a lot of kind of controversy around this. And I'm just curious what your thoughts are. So like, let's say a student, they use this, they find an awesome university, they think it's great. They know that, you know, they chatted with an AI bot. They kind of used AI to get them to this point where they're now like applying for school. What do you think the impact of, like, let's just say chat GPT, but AI is on like, actually the school admissions process. Like, would you ever have any sort of tool to help people with admissions?

18:26Or I guess what's your thought on like students, let's say just using ChatGPT to write their admissions essays or, or, you know, that whole, that whole side of it. Yeah. You know, I, I tend to be a sort of glass half full person as, and so I, my, my instinct is not to go to the negative and think all the different ways that you could abuse the system with technology. It would, an example would be, I had a bot write my essay for me. Now today, wealthy students oftentimes get someone else to either write them or appreciably write their essay for them. So that problem exists today, but it's limited to people with extreme means.

19:04And so to say, well, that problem is going to proliferate because it's now cheaper, it's essentially free to do that. Yeah, that's an issue. I think what the real opportunity though, for schools in particular is to understand who they are and what types of students do best in their culture and to be able to see that in those students. So I think that I had to actually had a the college president asked me the other day, he said, you know, would you envision this technology, if not today, but ultimately replacing my enrollment team? And I said, look, I'm not a futurist, but I don't think so. I think what we're talking about is endowing your enrollment team with much, much better information and better access to the kinds of students that you're looking for and allowing them to make better decisions because they have better inputs.

19:54And so it's not about the machine making the decision. It's about the machine making available better decision making data and matches for the human team to be able to make their selections. So, yes, clearly there are negative potential implications and teenagers will find them. Um, but ultimately it's sort of, you're only kind of harming yourself. If you lie to get into a school, however you want to define lie, but if you, if you present something other than your honest self and you get in on false pretense, you're unlikely to stick it out and you're just creating problems for yourself. So I, um, I'm maybe a bit naive on that.

20:39Um, but I, I really think that the opportunity is far greater than the risk of infusing this particular old system with better technology. Do you view, yeah, that's a great, that's a great answer. Do you, do you think that schools, like, you know, you mentioned you talked to that professor and he said, you know, this is going to replace my enrollment team. Do you think schools will ultimately move away from, I mean, and I know a lot of them perhaps already are, so maybe that's, maybe that's the answer, but do you think schools will move away from reviewing individual applications and go towards just AI essentially, reading the application and deciding if it's a good fit based off of your database?

21:21Like, do you think there's a place where maybe your technology or a similar one could be used essentially, you got this database on the user, and this is why you guys match, you have the database on there, and then it just runs an AI through their application to determine if they are eligible to enrollment in the university? Yeah, I think some version of that is certainly possible. I mean, if you think about today, some of the larger public universities, for example, it's hard to get great data on this, but the average estimation of the amount of time a reader spends on each application is just a few minutes.

21:53And so what are they doing, right? They're scanning through to say GPA, does that match? Test score, does that match? Let me take a quick look at the transcript, does that match? They're just buzzing through something in the way that essentially a computer would buzz through it. I'm sure they automated that system too, to a large degree, you know, as well. Right. And so that already somewhat exists. And the question is, not so much does the student have the academic aptitude? It's are they the right student for our particular learning and social culture? And that to me, we have a long way to go before we're able to really quantify this in a way that allows for pure automation.

22:32It's not our goal at TRU, but I think we are trying to help in that sense. If you look at the data now, the percentage of students who stay and graduate from their first enrolled school is 51%. So it's essentially a flip of a coin with the current system, right? So it's wrong as often as it's right. So we have to do better. We can do better. And so that's where I think the technology is going to help is not so much, do they qualify for admission, but are they likely to stay and graduate? So asking a different question is better. Yeah. And when you think about using data to improve those odds of that flip of the coin, what data are you inputting into the AI?

23:18Like what is it asking about them? What is it learning about them to try to help make a better decision? Because I know you said, you mentioned, is it a cultural fit? Is it, there's a bunch of different things that go beyond perhaps just like the GPT score and whatever? What kind of data do you envision? Yeah. So just think about the, well, we talk about the learning culture of a school. So, and some of this is impacted by major, but oftentimes it is a school specific kind of answer. And you see a pattern of data to the school that says, do you prefer a big school or a small school? And that's not a great question.

23:54When you ask instead, said, how important is a relation or sort of a learning relationship with your professor to you? When you have a problem in a class, do you go to the professor? Do you go to the TA? Do you go to Google? Do you ask a classmate? How do you solve your problems? And when you're studying, do you prefer to study alone? Do you prefer to study in groups? How do you form those groups? Do you just group with your friends? Do you find the smartest kids in the class? Do you look for diversity? when you start to ask questions around the idea of how you learn, you see those patterns start to sort of tease apart from school to school, as opposed to saying, well, the students with a 3-7 go here, the students with a 3-2 go there.

24:42That is the system that is a flip of the coin. And so when you start to ask deeper questions, and importantly, those questions cannot be sort of direct, if you will. You can't say to a student, choose from the following four types of students. Which one are you? Because the level of self-awareness of most teenagers, or even honestly, most adult learners is not that high. So when you ask them, what would you do in this scenario? You'll get a more clear read. I love that idea of asking questions around the topic, not just big school, small school, but everything around it, because I think that's maybe they have a preconceived opinion on the big or small school, whether that's a good or bad thing, but it might not actually match up with what they prefer for learning style.

25:28So I think that's good. And I think that's actually a really good piece of advice for anyone listening, building applications anywhere, querying your customers is not necessarily asking them the question directly, but questions around it with similar implications. So I think that's really interesting. My question for you is, you know, how do you see, like, do you have any evidence essentially, or is this something that will come of the fact that your questions, you come up, your quiz that you've built actually improves their odds of staying in the school, right? Like, I mean, theoretically, it's like, we asked them a bunch of questions that match them up better with the university.

26:08but of course you invented those questions. So like you have your own biases and thoughts in there. Like, do you have anything that actually proves that like your quiz keeps people in there, in their school longer? Or is that something that you will develop over time as you probably do like surveys after people begin and end university? Yeah. Um, we, we do not have the data yet, obviously that says the student that we matched with a school who enrolled there, um, persisted, graduated. And ultimately, you know, if you, well, let me just finish that thought. That request to be able to follow that student and monitor, not at the individual level, but at the aggregate level, so we're not dealing with data privacy issues here, but to say the students that matched with Truve, were they admitted at a different rate?