Joe Liemandt on Alpha School, AI Tutoring, and Rethinking Education

Joe Liemandt

The biggest-picture answer, if you talk to parents about why there's agitation for change, is that this AI world is coming and we're all terrified of what it means. We don't know exactly what it means, but what we do know is that the education system we all went through isn't going to prepare kids for that world. That's a huge driver. I talk to lots of kindergarten parents, and they're looking a dozen years out saying, 'If I put my kid in the standard school I went through, is that going to prepare them for this AI world?' And the answer is no. So there's a drive that we need something different.

But even 10 years ago, before AI hit the scene, people felt it — something is not working with the school system. And schools are failing on pure academic results. Test scores keep going down even as we spend more and more money. There are a lot of reasons, but a school is a complicated bundle, and academic performance is only one element — one that schools have basically decided doesn't matter that much anymore. Everybody in the ecosystem has been willing to let the standards drop.

Joe Liemandt

Part of it is how the system is built. There are a couple of truisms about the standard school system. The first is that it's a time-based system: you move up every year and you have a teacher in front of a classroom. In that system, two attributes make you do well. One, it's IQ-coded — higher IQ is better. Second, it's conscientiousness-coded, that big-five 'are you a natural grinder?' trait. If you have those two attributes, you do really well. If you don't, you don't, and the system doesn't adapt for you.

The second dimension is resources. Factually, independent of any politics, one of the strongest determinants of your educational outcome is your family's income. The whole system is geared that way. So you have these two forces, and people say, 'Okay, we'll spend more money.' But if you have a time-based system coded on IQ and conscientiousness, you can pour all the money you want in and it doesn't change those things. It doesn't give kids higher IQ or more conscientiousness, so it doesn't fix the problem.

Joe Liemandt

This is where it gets complicated, and the US is different from the rest of the world. In America, everyone says our schools suck — there's an unending stream of articles about bad scores, and COVID made it worse. Everyone knows about grade inflation: 80% of students at Harvard get A's now. So we all know the standards have dropped.

But the thing that's different in the US is a big disconnect between educational outcomes and earning potential. They're not as tightly correlated here as they are in the rest of the world. So there's a whole set of parents who feel the academics don't matter that much — the life skills and other things matter more — so they don't really care.

Joe Liemandt

A good story about this: when I became principal three years ago, our first product, the way I positioned it, was '2X learning' — your kid will learn twice as much. And I got incredible pushback from parents. Surprising amounts. I was doing hundreds of parent-teacher meetings, sitting with mom, dad, and Johnny, saying, 'Look, Johnny's not learning 2X, because when he misses a question and a video comes up explaining why he missed it, he just skips it. He could learn twice as much if he watched the video.' And the parents would look at me like, 'Why are you pressuring Johnny to learn 2X?'

It took me hundreds of meetings to realise the parents don't value it. Now, 10% of the market totally cares about great academics and demands it. But we switched the message, and that was our biggest product-market-fit pivot. Same Johnny, same mom and dad, but now I'd say, 'Look, I'm trying to get Johnny out of here in 2 hours so he can do all the cool stuff the rest of the day. He skips the videos — if he just watched them, he'd be done in 2 hours.' And mom says, 'Johnny, what's wrong with you? Don't you want to get out of here?' People want to get out of the academics. When it's only 2 hours, everybody's aboard. We still teach 2X in 2 hours — we just emphasise the 2-hour part. We literally dropped it and called it '2-hour learning,' where you learn 2X in parentheses.

Joe Liemandt

Alpha School is a high-end private school expanding nationwide. We'll have 25 campuses around the country this year. The way to think about it: if you could revisit everything and say, 'We need to build an educational system that's going to be ready for when AI comes,' and do a ground-up rewrite — ignore everything you and I went through, the six hours a day in class, the homework — and just ask what skills the kids need, how they should spend the day, and how to do that, what would it be? That's what we're doing at Alpha.

We also took off budget constraints. It's a high-end private school, so we designed as if money were no object. We're going to drive the cost down over time, but the first version is the high-end design. And we've built a school totally different from every other school out there.

The number one rule, if you're going to put your kids in this for a decade, is that kids must love school. That is our core principle, and it makes everything else happen. When I talk to principals and superintendents, I tell them if you don't believe that, all the rest — the two-hour learning and so on — won't work. We survey our kids every eight weeks. It was 96% saying yes, but that got too high to be a useful metric, so we added a second question: do you love school more than vacation? We get 40 to 60% of our kids saying they'd rather go to school than take a vacation. That's the metric we wake up to now.

The highlight of my time as principal was this May, when two-thirds of our high school kids sent me a note asking, 'Mr. Liemandt, can we keep the school open this summer? We don't want a summer break.' I hated high school. I would never in a million years have asked that. There's a lot to unpack about why they wanted it, but that's the core. If we're going to put our kids in for 12 years, we as parents should say they ought to love it.

Joe Liemandt

I had that view. When I first put my kids in, I was talking to Brian, who started Alpha, and I asked what I didn't know about education that I should. He said, 'Kids must love school.' And I said, 'It's spinach.' He said, 'You don't get it. Your kids are going to come here and you'll figure it out.' Now I'm total religion on it. It makes all the magic happen.

What's interesting is when I talk to other business executives, they say no, it should be spinach. And I ask, 'Is that what you think for your employees? Do you wake up and say this job sucks and it should suck?' No — they'd never say that. But they turn around to their kids and say, 'Suck it up.' It's ridiculous. We think you shouldn't like school just because we didn't like it. If we're rebuilding it, we need to undo that thought process. Kids should love school.

Joe Liemandt

Our kids love school, and academically we're the best-performing school in the country, bar none. We're big believers in standardised tests, because when you hear how Alpha works, nobody believes it will work. A lot of parents don't even believe in standardised tests when they arrive. They say, 'My kid's only doing two hours a day — how do I know they're learning?' I say, 'Trust me,' and they say, 'No, I don't.' So when the kids score 790 or 800 on the math SAT, they say, 'Okay, it's working.' On the NWEA MAP test, our kids are top 1% — every grade, every subject, every year. And we publish all these results.

Our kids in two hours are learning twice as fast as if they sat in school for six and did homework. The MAP test is a 300-point scale. A normal 75th-percentile fifth grader might go up five points in a year; at Alpha they go up ten. It's quantifiable — parents log in and see their kid learned twice as much in two hours a day.

A lot of private schools just select for good students. The other way is to develop them, and ours is all about growth. We take kids from bottom-half to top-half to top 1%. If you download our results and put them into ChatGPT and ask it to analyse Alpha's growth rate against any other school, it's about four standard deviations off the track, because nobody takes kids who are behind and catches them up while also moving the top 1% years ahead. I have third and fourth graders who outscore 50% of the high school graduates in this country.

And the product keeps getting better. My oldest daughter went through the school and is now at Stanford. She came back and was looking at some third and fourth graders' scores and said, 'They're getting the scores I got in seventh and eighth grade.' Human potential is unlimited, and we have that improvement cycle going.

Joe Liemandt

Selection effects are a great discussion. In most of education there's a view — maybe because I come at this as a product guy — that you must serve everybody all at once or you'll serve nobody, and that selection effects are bad. As a product guy, I think you must use selection effects to build your product: you have to select the right person you're designing for. There's no school without selection effects. Yes, at Alpha we have an easier cohort than some others we'll talk about. But compare us to every other high-end private school in the country, not to the hardest situations. Take other schools' top-1% kids and see why ours learn three times faster.

And on the 'what if you're not rich' part — we have another school that isn't Alpha, because I have to get this down to everybody. I'm going to spend 20 years on this and get it to a billion kids, so I have to solve all the demographics. We have Texas Sports Academy rolling out, a dozen campuses so far. Tuition is $15,000, and with Texas's new voucher program of a billion dollars, it's a $12,000 voucher for income-eligible families — rich people can't get it. So it's about $300 a month for parents. These kids generally come in the bottom half academically, but they all love sports. It's a middle school for D1-aspirational athletes. Back to the core — I have to build a school they love more than vacation, so I need to know what lights them up.

Joe Liemandt

Same system, same two hours. So now we have a wider funnel of kids from different academic backgrounds, and we'll publish those results too. The problem with kids learning isn't the kid, it's the system. The system is coded for high IQ and high conscientiousness. We've built a system that breaks both. If you have a mastery-based AI tutor that only lets you advance after you've mastered the material, that makes it effort-based, not IQ-based.

We run a program called '100 for 100.' A middle schooler comes in and I say we'll pay $100 if you get 100% on the test. They say, 'I can't get 100.' I say, 'Pick your grade level.' They say, 'You'll let me do a third-grade test?' Absolutely. They get 100 on third grade, then fourth grade, then on fifth grade they get a 75 because they have holes. The AI tutor offers to generate the lessons they're missing, and they say, 'I can do that in a week — for another 100 bucks?' We get more 100s on state standardised tests in a building of 200 people than a district of 100,000. If you ask alpha parents whether their kid can get 100 on the state test, fewer than 10% say yes, because in our generation only the gifted-and-talented kids did that. Ask alpha kids and they all say, 'Of course.' It's effort-based, not IQ-based. There's no concept in US Common Core through eighth grade that everybody can't understand. If you're having trouble, it's usually because you don't have the prerequisites.

Joe Liemandt

That's 100% of the problem. Knowledge is cumulative, so when you slow down it's because of your prerequisites. Teachers are awesome — the teacher in front of the classroom is terrible. You have to separate those two concepts. That's why everyone wants small class sizes: you basically want individualised lessons. Rich families who hire after-school tutors — all the tutor does is give the fourth- and fifth-grade lessons. But a sixth-grade teacher has to hand out sixth-grade material.

I went to see a sixth-grade teacher at a school with bottom-10% kids. Bottom 10% in America means you can't read and don't know 7 times 8. She's handing out sixth-grade material, and that kid needs a first-grade phonics lesson and the next needs second-grade maths. There are actually laws in various states that you must teach at grade level — it's illegal for a teacher to teach below it. It's insane.

We've known from learning science for 40 years that kids could learn two, five, ten times faster, but not in a teacher-in-front-of-a-classroom model. Bloom's two-sigma is the famous one — everybody fights about it, but it's a good paper. If you use a human tutor with mastery, you get two-sigma-better performance. The problem is you can't give everybody a human tutor, and you have to enforce mastery, which is hard.

Joe Liemandt

Do you know the material cold? Think about sports. If you're a point guard going down the court, what percentage of the time will the coach let you lose the ball? Whatever that number is, that's mastery — it's not very often. In academics, a kid gets an 80% and we say, 'He seems to know it,' sometimes even an A. If a point guard were losing the ball 20% of the time, the coach would say, 'Master the basics — we're not working on dunks.'

In a time-based system we say, 'It's time to move on,' and the holes compound. Look at the learning curves in the US: they slow in middle school and flatline in high school. On the MAP scale, if you take an eighth grader, the median kid goes up one point by senior year. One point — they learn essentially nothing in high school. Your top 1% still go up maybe six points a year, but the median kid isn't learning. UC San Diego, one of the top universities in the country, just published a paper that 50% of their freshmen can't answer fifth-grade questions.

And there'll be no bounce-back from COVID. It used to be that a sixth-grade teacher had a range — a gifted kid needing seventh-grade material, someone behind needing fifth. Post-COVID it's a seven-year span. You have people in middle school who can't read, and if you can't read, going to school is a waste of time until somebody teaches you to read. Everybody got socially passed, too.

Joe Liemandt

Exactly. And with a mastery-based AI tutor, you go to those kids and say, 'I'll make sure you get the third-grade phonics.'

Joe Liemandt

It figures out what you're strong at and weak at, then teaches you what you don't know. It won't get you ahead in things you already do until you've caught up on what you don't. And you don't know it's teaching you at a third-grade level, so there's no social stigma.

The learning science backs all of it. There are 10,000 papers on how kids can learn better. The zone of proximal development says give them questions they're getting 80 to 85% correct. At 99% they're not learning because they already know it; at 50% they disengage because it's too hard. Every video game developer knows this — keep the kid engaged. So imagine an AI tutor giving your kid an unending stream of questions at 80 to 85% accuracy, adjusting difficulty up or down. Think of the knowledge graph of everything they need to know. AI tutors are good at assessing what you know, so the kid takes a test, the tutor finds the gaps, gives lessons until they're filled, then moves them to the next grade, at whatever rate they go.

Here's the crazy part — kids can learn ten times faster. Take seventh-grade science. In a normal school that's 180 days, an hour a day, a couple hundred hours. The average kid in our system, with Timeback — our app — learns it to mastery in 22 hours. Across K-8 you should think 20 to 30 hours per grade level per subject. That means you could basically be done with the school year in September. The material just isn't that much. There's a great maths app called Math Academy — Justin wrote a 500-page book you can download that goes through all the learning science, and they publish how many hours each grade takes. It's like 28 hours, 22 hours. Not many hours to master this material.

We know the teacher-in-front-of-the-classroom is the least effective way to teach — you can argue whether it's 5% retention, but we're wasting 90% of kids' time. For 40 years we didn't have a better way; you couldn't give everyone a tutor, so there was no excuse to tell everyone they could learn ten times faster. Now there's no excuse. We can get everybody learning that fast.

Joe Liemandt

It's really hard to change the system. First, nobody believes it. The first thing you do when you see our results is say 'selection effects, it's only rich kids,' and reject it. Even at high-end private schools charging $50,000, whose largest donors want this, the boards and heads of school struggle to believe it works, because they're not trained in learning science. They haven't read the 10,000 papers, they don't have a team like ours — seven of the ten best learning scientists in the country — and they don't have the data loop. We measure every single thing: how long a kid took, what the scores were, the whole closed loop. So there's massive scepticism. It's a classic disruption problem.

Joe Liemandt

Families like yours are our early adopters — we call them founding families. Homeschool families know two hours is all you need; that's all they've ever done. Their question is why Alpha is so expensive. The other group is athletes. Our head of high school used to run academics at IMG, where big-time athletes do academics in the morning and play sports all afternoon — that's how they catch up, and that's how our sports academy is designed.

Joe Liemandt

More important than the AI is the learning science — it's the combination that you really get online. But let me be clear about what I don't mean by AI. Our school does not teach people by giving them ChatGPT. Chatbots are cheatbots. 90% of kids using chatbots for academics are cheating, not learning. We do academics in the morning and life skills in the afternoon, and I tell them: if you're using a chatbot in the morning, you're probably cheating; if you're not using it in the afternoon for your life skills, you're probably failing. Chatbots have a use — just not on the academic side.

How we use AI to teach is the generative part, not the chat part. We generate personalised lessons for every kid. Imagine an LLM fed four things: the curriculum you're teaching, the kid's knowledge graph — what they know and don't — the kid's interest graph — what they like, say baseball — and, coming in 2026, cognitive load theory. Analogies are one of the fastest ways to learn, so if I teach you through something you love and are good at, you learn faster and stay more engaged. Cognitive load theory treats your brain like a chip with working-memory slots — everyone has a different number, and each person needs a different number of reps to move something from short-term to long-term memory, like flashcards. An LLM sitting on all that data generates the perfect next lesson for you.

Those lessons are all built on learning science — engaging, not passive; a worked example in maths, because that's the best way to teach; short videos with text and audio overlay. If you get something wrong, it goes back, and it will slip in questions to check whether you really have the prior knowledge. You might get the second-grade lesson while another kid gets the first-grade or twelfth-grade one, because the AI gives each kid whatever they need.

The second element is the most expensive part of our AI — right now we burn about $10,000 a kid, and we'll drive it to under $1,000, then $100. Think of streaming the screen to one of the best models and prompting the kid on how to learn. It's a coach. It says, 'You keep skipping those videos — you're not learning as effectively as you could.' In the bottom corner there's a little waste meter telling the kid they're wasting time.

Joe Liemandt

Our product name is Timeback, and the most important part of educating a kid is motivation. Every educator will tell you that you need a motivated student in lessons of the correct difficulty. AI does the second part well. But how do you solve motivation?

My first week as principal, I went up to the fifth graders and asked, 'Do you love school?' No. 'What would make you love it?' 'Less school.' 'How much less?' 'None.' So we negotiated two hours. I said, 'Would you engage in the apps for two hours if I gave you the other four hours back to do awesome stuff?' They said, 'That seems fair.' I went to my team and said, 'We've got two hours. Read every learning science paper and figure out exactly what we need so they can learn all the material in two hours.'

So when a new student is screwing around and not engaged, their two-hour learning becomes three-hour learning, and they say, 'Whoa, I thought this was two hours.' And the waste meter says they're wasting 50% of their time. Change these behaviours and you're out in two hours. At the sports academy, the meter has to go green before they get to play basketball or soccer. Their coach says, 'Guys, stop screwing around, get it done so we can all go play,' and their friends are already out playing. Now you have motivation. It's not hard, because we keep them at 80 to 85% — they just can't screw around.

Every kid tries to game it — they scroll to the bottom looking for the questions, trying to skip reading the article. You tell them, 'You actually have to read the article if you want to learn to read.' But we can also generate the article so they're interested. Take a second- or third-grade boy who says he hates books. Ask his favourite movie — the Avengers — and who his buddies are, and now it's a choose-your-own-adventure where he and his buddies save the world, and he'll click page by page for an hour. You keep the Lexile level within 10% of his difficulty so it's not too hard, and you funnel new vocabulary in.

One of our design principles is that parents give us 12 years to fill their kids' heads with awesome knowledge, because we should make a kid's head an interesting place to hang out — they have to hang out there the rest of their life. And there's a learning science concept that the more you know, the faster you learn. The average kid's curve slows down; ours accelerate. At high school we'd be at 5 to 10x, partly because the other kids aren't learning so the denominator drops, but also because ours keep learning.

If there were three things I could put into a parent's brain that they don't believe today: one, your kid can — and must — love school, more than vacation, and that should be your standard. Two, they can crush their academics in just two hours a day; the six hours plus homework isn't necessary — we have the highest test scores in the country on two hours. If you go more, it's only to catch up. Three, the key to your child's happiness is high standards.

Joe Liemandt

If you're building a school kids love, there's a view of 'just let them play all day,' but that won't make them want to come during the summer. They have to work on things that are ambitious and significant. If they're just scrolling TikTok or playing Fortnite, they won't come — they can do that at home. Kids are awesome and they want to do awesome stuff; we need to give them the environment.

We start in kindergarten. The thing that freaks everyone out — we do it on a shadow day so parents can test it — is we have a 40-foot rock wall for kindergartners. It's all safe, harnessed in. We teach growth mindset, we mic the kids up, and halfway up they start to struggle. Our guides and coaches — think of a mentor whose whole job is to hold high standards but give high support — coach them through it. When the kid gets to the top and comes down, they're beaming, because they accomplished something hard.

Every child-development expert will say the cycle of struggle and failure — I add, as a principal, sometimes crying — on the road to success, supported by a caring adult, is what child development is. It builds self-confidence and resilience. It's why 50% of America puts kids in after-school sports; everyone expects that cycle in sports but not in school. Every kid needs it.

In middle school we have an epidemic of scrollers — 'consumers,' we call them — caught in the dopamine loops of TikTok and Fortnite. So we run a values chart, a Japanese Ikigai, a vision board where a fifth or sixth grader asks who they want to be. They all want to do awesome stuff and they diagram it. Then we make them do a 168-hour project — track every hour of your week. And they realise, 'I guess I'm going to be the world's best TikTok scroller,' and they're disappointed, because you are what you do. That's our job as a school. Because they're not in academics all day, we have afternoon workshops where they find their passion project.

One kid told the next cohort that if you'd told him at the start, 'I'll pay you $100 an hour for the rest of your life to play video games until you die — just plug into the matrix,' he'd have sold his soul for it. Then he went through the program and said, 'Now I don't even have time to play video games,' because he's engaged in who he wants to be. And back to high standards — that kid has never worked harder than on the project that pulled him out of the loop. Kids want to do it because it's hard.

Our afternoons are project-based workshops teaching life skills in five categories: leadership and teamwork, storytelling and public speaking, grit and hard work, entrepreneurship and financial literacy, and socialisation and relationship building. Our second-grade teacher, Faith, said, 'All my second graders can run a 5K.' I was sceptical. She said kids are limitless, and she'd teach them Atomic Habits — most of us read it at 25, but sixth graders understand 1% better every day. She signed them up for the Jingle 5K at Christmas back in August, walked the track with them the first day, and every one finished. Then a quarter-lap running, then half a lap, and by December they're all running 35-minute 5Ks. When they cross the finish line, they tell their parents, 'Dad, I can do anything, because I can be 1% better every day.' They learn life skills we don't learn until late, through hard things they love.

Joe Liemandt

There's a huge teacher problem — burnout, can't hire enough. Coming from the outside, HR 101 says either the spec's too hard or you're not paying enough. To be a teacher in a standard classroom you have to be good at five hard things: be a domain expert in seventh-grade science; be good at teaching, which is separate — plenty of researchers who know a lot are bad teachers; connect with and motivate kids; deal with parents; and handle administration. Five hard things, and then we underpay them. No surprise no one applies.

So how do we re-envision the role? Teachers didn't become teachers to grade seventh-grade science quizzes. They became teachers to connect with and transform kids' lives. We all had one or two who did that for us — not because they marked your paper well, but because they connected with you, got you to see a bigger picture, and pushed you to greatness. So we narrowed the scope. You don't need to know seventh-grade science, because AI does that. You don't need the 10,000 learning-science papers, because the AI tutor does that. Are you the world's best at connecting with and motivating kids? Can you hold high standards and high support? If so, we want to hire you.

On the parent side, we have a dean of parents, so guides don't have to do that. When I first arrived, someone said, 'This teacher's great with kids but not with parents, so don't promote her; this one's not as good with kids but great with parents, so promote him.' I said no — we'll build a school where they're great with kids and we'll find another way to handle parents. Those are different skills, and the more important skill is the kids. We call them guides and coaches because they're not doing academic teaching.

Joe Liemandt

We get the top 1% of teachers who want to move over. We tell them, 'You're not going to be a teacher,' and that's always the biggest surprise — they say, 'This is what I want to do all day.' During the two hours, since they're not in front of class, they pull kids out one-on-one.

Joe Liemandt

Coaching. 'How was your weekend? How was the softball tournament? We've been working on this goal about your vision, and you don't seem to be engaging — do you really not like it, or is it too hard?' They connect, then help the kid understand what they're trying to do and how the guide can help them hit their goals.

Joe Liemandt

Our guides have great relationships with the kids. In middle and high school we ask parents: do you trust your Alpha guide to hold high standards so you, as the parent, only have to provide unconditional love and support? As kids hit adolescence it gets hard for a parent to be both. Take my example — my oldest daughter was putting in her Stanford application, and I asked to read it. She said no, because kids don't want to be judged by their parents. So I just gave her a hug and said good luck, because I knew Chloe, her guide, a Harvard grad, would hold the standard. My daughter trusts Chloe, and even now that she's in college, when she comes back to Austin she goes to lunch with Chloe.

Everyone thinks 'AI tutor' means robot terminators. But predict 20 years out: parents will still drop their kids off at a building with other kids and adults. School isn't going away — kids need to be around other kids, and they need caring adults to mentor and coach them. What changes is what happens in those six or seven hours. Right now we waste 90% of that time; we're just going to stop that. So our design has three parts: they must love it; they need to learn ten times faster, which is the Timeback engine; and there's a set of life skills that matter beyond pure academics.

Joe Liemandt

The biggest difference from others is we try to quantify everything, because a lot of life skills are soft and I can't run a scalable system on 'maybe it works.' Everyone wants their kid to have grit, but reading Angela Duckworth's book doesn't mean anything. So every third grader in our school can do a Rubik's Cube, every kindergartner can do a 100-piece puzzle, and every eighth grader has passed a teamwork-grit test — they all run a Tough Mudder and cross the finish line at the same time. Some of these high-end-private-school moms say, 'My daughter's never rolling in the mud.' Then at the end they're all covered in mud, bonded as a team.

All our fourth and fifth graders can pass Wharton's leadership and teamwork simulation that Wharton MBAs take. For entrepreneurship, can you make $1,000? Can you raise $10,000? Character building is the hardest to measure but the only thing parents really care about — is my kid a good kid? We have middle and high schoolers building an app to convince a hundred million teenagers that the key to their happiness is contributing to their community. They're learning to vibe-code with Cursor in the afternoons, working with some of the biggest influencers on the planet, and they don't want to go on summer break because they're working on something hard and significant.

Joe Liemandt

I love that story. If every parent thought like you, we'd have a different school system. One thing I've learned as a principal is that your mindset is very rare. Parents in general want low standards. It's why even private schools react — why does Harvard have grade inflation? There's pressure from parents and students to lower standards, and it's hard to keep them high. The 'participation award' problem.

Where it manifests is middle and high school, where kids are disengaged, addicted to TikTok and Fortnite, with mental health problems. Those are the outcome of low standards, not high ones. People think it's too hard. But if humans don't engage in challenging things they love and can accomplish, they lose meaning. There's no meaning in a TikTok loop; you can't eat sugar all day. We need schools that show kids it can be fun. If you took out 'kids must love school more than vacation,' parents wouldn't send their kids because they'd be terrified of the high standards. It only works because their kids come home saying it was the greatest thing ever. Parents of the sports kids say, 'They used to skip school constantly; now they wake me up saying we'd better not be late.' Alpha kids ask, 'Why don't we go to school on Saturdays?'

Joe Liemandt

The monitoring is really interesting. Kids like the monitoring and coaching, because they want to get better. Parents don't like AI monitoring — they're terrified of it, there's a boogeyman around it. But the kids say, 'I don't want a human to look at me, because I'm judged by adults. Let me talk to the AI.' We have an app called Orai that grades your public speaking, and if you ask whether they want a human giving feedback or Orai, they choose Orai until they hit 90% — it flags filler words and pacing. It's not judgmental to them, so they don't feel the pressure. Parents are the opposite. I tell them, your kid wants this, so get over it.

But the monitoring alone isn't enough — you need the motivation, the coaching on why they should do it. If you put Timeback into a standard six-hour school day, it won't work. EdTech has generally failed because kids aren't motivated to engage. If you don't give them their time back, they won't engage. That's why it's hard for a normal school to adopt — it's a full rebuild, and there's not enough proof yet, and you have parents who don't want screen time. Alpha is the most talked-about school in the world; give us one more year and everybody will realise they need to do this.

Joe Liemandt

A few practical reasons. First, they've been lying to their parents. High-end private schools give kids A's they don't deserve. We had hundreds of kids join recently from $40,000, $50,000, $75,000 private schools, and we gave them our assessment. If your transcript was an A, some students were one grade level ahead — awesome — but many more were three years behind. Seventh graders missing a lot of problems on the fourth-grade test. If you got B's, you were three to seven years behind — high schoolers who need third-grade material. Parents don't give out C's, D's, and F's anymore, so A's and B's have to cover the whole range, and parents are being lied to using their old point of reference.

So a school can't put in the AI tutor because it'll show the kid needs fourth-grade material, and the parent doesn't know that. As soon as they put it in, the assessment makes it obvious. That's actually what causes a lot of people to move to Alpha. We opened in San Francisco, and one father said, 'I've had my kid in a $50,000 school for eight years,' and the kid couldn't answer fourth- and fifth-grade questions. Not the kid's fault — the school gave A's, good brand, everything looked fine.

Second — and I fought this. My first year I went to the seventh graders and said, 'You have holes, I need you to do earlier work, some fourth- and fifth-grade material.' Try motivating a kid and making him love school when you tell a seventh grader he has fourth-grade material. Non-starter. The only people who hate it more are the parents. 'My kid must be at grade level! Why are you putting him in earlier stuff?' Every parent's been trained that their kid must work at grade level, but that's not what learning science says — you have to go back and fill, master the basics. I still don't know how to fix the parents, so I went around them, and that's how 100 for 100 came out. Third grade, they did it; fourth grade, they moved up. And it builds self-confidence — they realise they can get 100 on any standardised test.

We advance kids at 90% to the next grade level, because a 90% shows mastery, but you get the 100-for-100 bonus if you get 100. So kids think, 'That's just a couple extra questions in an hour or two.' It sets the mindset that of course they can get 100. The reason we get so many 790s and 800s on the SAT is that our kids think, 'Of course I can get a perfect score — it's just how many hours.' It's an hours question, not a capability question.

Joe Liemandt

Totally. The SAT has a little IQ coding — any timed test does — but 1350, which is top 10%, no problem. 1550, top 1%, gets a little dicey. For 90% of kids, off the charts. In our high school program, non-honours is a 1350 minimum, honours is 1550. The difference is three hours a day for honours — 1550 plus fives on your APs — versus two hours for 1350 and a four or five. To kids it's just an effort issue: do you want two hours or three?

The after-school tutoring market is about $50 billion, which is insane. The percentage of kids at high-end private schools who also have a tutor — aren't you sending your kid to school to learn? We commit that your kid will learn during the day, so they don't need the tutor.

Joe Liemandt

Math Academy is the best maths app for fourth through twelfth. There's a new one called Physics Graph — the physics version of Math Academy. AP physics is ABC or 1-2-3; the first is algebra-based, the third calculus-based. I'm not sure they've released the calculus one, but for the algebra-based, use Physics Graph.

Joe Liemandt

Even though we have high support, our kids are self-driven learners — in the two hours they have to learn how to learn on their own. Every eighth grader knows how to get a five on an AP by themselves. They know everything they want to learn is on YouTube, so we say, 'Pick some random AP you're interested in and go learn it.' That's how we show you're a self-driven learner who can learn anything for the rest of your life without being spoon-fed.

Our first class had 12 graduates. Eleven went to college; the twelfth is one of the world's best water skiers, building his media empire. The others all got into their first-choice colleges — Stanford, Vanderbilt, NYU Shanghai — because their academic scores are through the roof. They also have these Alpha X passion projects they work on for four years, so they look like a completely different character from the average college grad. We may have a high school student who's the lead researcher on something submitted to Nature — no high schooler has ever been. It hasn't been accepted yet, so I shouldn't have mentioned it, but if she were a PhD it would be one of the most impressive things of her career, and she's 17.

I met with them at Christmas. Everyone had a 4.0 and loved college — sororities, the social side. When I asked the worst part, they said, 'Those lectures are a total waste of time.' Four of them had written editorials to their school papers asking to stop the charade. Two hundred people in a class, none paying attention, all scrolling. My daughter, a maths-English major, says she just uses Math Academy. She used to have disdain for me for skipping class, and now she says, 'I get why you skipped — it's a waste of time.'

Harvard did an A/B test of one of their best physics-101 teachers versus AI tutors and published that the AI tutor is better — kids learned twice as much. So the 100-series classes at universities are going to go away once everyone figures this out. Small-group seminars they love, so universities won't disappear — some kids still need time to grow up and to socialise — but the big lecture classes are gone.

Joe Liemandt

The world's worst. Anyone who knew me at Stanford can't believe I'm doing education. I dropped out. Every kid has a passion for learning — it's just not always academics, and that was me. I didn't care about the classes. In high school our grading gave four points for an A, no A-minus or A-plus, and an 89.5 rounds to a 90. I'd literally do the minimum to hit 89.5, not 89.6. One time I didn't finish all the questions on a test because I knew I didn't need to for the 89.5. I was a nightmare — whatever the minimum was.

I got into Stanford because I wanted to start a company. In some Spanish class I thought, 'What am I doing?' and dropped out. Then I worked 100-plus hours a week, sleeping under the desk, super passionate. I found my spark, and I'd never worked harder in my life. My dad wasn't happy I dropped out, but 18 months later he said, 'I'm so glad you did. I was worried you were a lazy shirker, and you found what you love and you're working your ass off on it.'

My view is that what I found after dropping out — that spark — we want to get into kids in middle and high school. As they hit adolescence, who do they want to be, what do they love? Let's get them started on it. There's some minimum academics — you want to start a business, you need maths; you want to write good proposals, you need to write — but you do it motivated by what you love. We're competing with TikTok and Fortnite, but kids' vision of who they want to be is awesome, so associate the learning with that.

We have a high schooler doing an all-teen-produced Broadway musical. She sourced the singers off TikTok, got a million streams the first week, and filled the venue four times over. She's learning about music rights and legal issues — harder than any high school class — and she'll spend endless hours because it's hers.

Joe Liemandt

Exactly. Take our sports academy. You tell a 10-year-old boy, 'We're going to teach you public speaking,' and he's turned off. But say, 'At our sports academy, post-game press conference — are you ready?' and he's in. Or, 'We've got to work on your Nike $5 million contract,' and suddenly he wants financial literacy. Kids will spend endless time on what's in their interest graph. We have a kid who knows everything about World War II battles — show him any picture and he'll say, 'That was Poland, 1942.' Kids learn endless amounts if they're interested; our job is to attach what we want them to learn to projects that are hard and meaningful to them.

Joe Liemandt

Totally true. My dad was worried about me being a spoiled kid — he grew up with nothing and got success later, so he was terrified of that. He said, 'If you have a good idea, other people will invest. If I invest, it just means it was such a crappy idea that no one else would give you money. So go convince people it's a good idea.'

Joe Liemandt

Yes, he was Jack's strategic planner. His son Mark and I played hockey and skied together, so I had a business upbringing, which is maybe why I knew I wanted to be an entrepreneur.

Joe Liemandt

My dad passed when I was 25, and Jack called and said, 'I'll be your mentor going forward.' Every year I'd fly up and spend a couple of days at GE and at Jack's house. There are a ton of stories. Mark and I were small for our age and wanted to switch from hockey to skiing — don't tell a hockey dad that. Jack said, 'There's a race at the end of the year, Chimney Peak. You two need to come in first and second, and then I'll let you switch to skiing.' We trained all summer, running up the mountain after school. Mark came in first, I came in second, and Jack said, 'Wasn't it worth it?' Crazy high standards.

On gross margin — GE had swapped their TV business with Thomson for CAT scanners. I said, 'TVs are the only cool thing you make; CAT scanners are boring.' Jack said, 'Let's talk about gross margin,' because TVs have none and CAT scanners make a lot of money. And I'll connect that to now: our fifth graders launched a food truck, and I asked why they picked breakfast foods. They said, 'Have you seen the gross margins on breakfast food?' Most awesome thing ever — they're building a real business, so they have to know gross margins.

Jack was both harsh and supportive — high standards, high support. I was the youngest self-made member of the Forbes 400 and got the cover, and I flew up full of myself, ready to sign him a copy. At the end he said, 'If I had to classify you, you're like a level-eight product manager.' At the time, realistically, that was it. Then he said, 'You have to decide whether you're going to sit here full of yourself and decide you're awesome, or decide you can spend the rest of your life and crush it. Is this your first step or your last step?' The whole plane ride back I wrote down a hundred things I had to do to make sure I was on the first step.

Another one: in '97 we won e-commerce product of the year, and GE Medical had used it. Jack said, 'I hear Med doesn't like your product.' I said, 'It won e-commerce product of the year.' He said, 'Joe, I don't care if it comes from heaven above. If GE doesn't get an ROI, your product, you, and your company suck. Where's our ROI?' It reset my whole view. I was a product guy who thought I'd delivered a good product, and he said if your customers don't get value, nothing matters.

I apply that to education now. Education thinks its job is to put kids through school. Ours is to actually produce learning. I think we're the only school in the world where, if a kid isn't learning, it's our fault — the system's fault, not the kid's. If our kids aren't learning 2X in two hours, I ask what I did wrong. Was my motivational model off? Do I need a different guide? Was the lesson the wrong level? Most school systems say, 'Kid's fault.'

Joe Liemandt

The pcOrder story helped me realise I didn't want to be public. One of the Trilogy co-founders, Christy, built pcOrder — selling computers online in the dot-com days — with all the biggest resellers aboard, and took it public. Then Dell put her customer base out of business, and the stock went from 50 to about 5. It ran up in the hype and imploded, and we bought it back. I've kept Trilogy private to this day, mostly because I prefer to be 100% in control. Like Marcus Lemonis says on The Profit, 'I'm 100.0% in charge.' That's my attitude.

Joe Liemandt

Correct. I'm in charge, or I'm your biggest supporter — good luck. I don't do well on 50/50 consensus. I push the edge on radical stuff, and I don't want to vote — it's not a democracy. If you want to start a school on Timeback, you're going to do 2X in two hours. People say, 'How about 1X in one hour, since 1X is as good as every other school?' I say go do that with someone else's offering. It's who I am. Reminds me of Alan Mulally turning around Ford: here's how we're going to do it, and if you don't want to be part of that, that's on you.

Joe Liemandt

We make three commitments to the kids: your kid will love school, they'll learn 2X in two hours, and life skills. Every kid. If you're in my organisation, you wake up every day to deliver that, you never blame the student, and you never make an excuse. We lose about 50% of our educators when we tell them they're accountable for the kid learning, because they're not used to it — they'll say, 'What if he doesn't do this?' Our guides are measured on whether they deliver the three commitments to every kid.

When I first came in there was skepticism, because I'd say crazy things like, 'There will be no adults doing academic teaching in this building.' With hindsight it works, but before that it's crazy talk to a parent or teacher. One teacher from public school — I'm the evil billionaire doing education, all the memes — I asked what won her over to stay. She said it was my constant focus that it had to be every kid. We'd get 90% of the kids there and I'd say, 'Let's focus on the ones we're missing — what do we have to fix to get it for every single one?' That's the model if we're going to get this to a billion kids.

Joe Liemandt

Scaling is hard. Building your first prototype is easy; building the factory that turns it out is the hard part. Timeback is a key element on the academic side, and that part scales relatively easily once we bring the cost of compute down — you could add a billion people if you had the compute. The hard parts are the physical and virtual.

On physical schools: how do you measure guides so a new school you've never visited is as good as the Austin flagship? We call them painfully insightful metrics. Every summer we ask, what's the metric or survey question that tells us the new school is as good — how do I know I don't have a bullying culture somewhere? On academics we have the best closed-loop system, so I know daily whether a kid is learning. But we hold the same level of metrics around culture and guides. Within eight weeks I know whether a guide is good. We ask kids, 'Every adult had one or two teachers who transformed their life — is your guide that for you?' We ask parents whether they trust their guide to hold high standards. And we quantify all the life skills, so it's not 'our kids have grit' but 'everyone scored 95% and every fifth grader beat a Wharton MBA in teamwork.'

Joe Liemandt

Exactly — and asking the hard question. It's why we moved from 'Do you love school?' to 'Do you love school more than vacation?' Comparing to your old school is too easy; better than vacation is a hard standard. You're always trying to find that extra question. We opened many new locations, and to know if they're good we have six and a half hours of testimonial videos from families who joined in the last 10 weeks saying their kid's life transformed. We show all our grades publicly. And the other part is core culture: do you believe in delivering the three commitments to every kid? If a kid can't get through a lesson, the person who made the lesson gets on a call with them. Some of the learning-science researchers quit when they realise they have to deal with kids.

Our growth team works in three concentric circles. First, is there anything needed to deliver the three commitments to every existing child? If so, do it before worrying about a new campus. Second, have we filled every school we already have, because the number of classmates matters? Only then do we ask whether to open a new location. We're the most talked-about school in the world, demand through the roof — the Secretary of Education has been down, ten governors are coming — but I insist on high quality at all times. Never relent on quality.

Joe Liemandt

I can't reach a billion kids next year by building schools, so there's the virtual path. First, we'll let people build on the Timeback offer. Anyone who wants to build a school — now's the time. We're hiring a thousand coaches in Texas for the sports academies. We have 500 gifted-and-talented schools, a Montessori school, a wilderness school. Think Shopify for microschools — that's what Timeback will be. Then there's pure virtual: homeschoolers and everyone else. That's what my kids want — the two hours on weekends.

The problem is that if 10% is the EdTech and I let it out of my environment, I lose the motivators. If I own all six hours and give you four back, I have the motivation. At home I don't. So we've spent all our time on motivation, and when we release Timeback next year the set of motivations will be awesome.

Joe Liemandt

Take middle school boys — their number one motivation is video games. Imagine 150 of the world's best video-game designers building the world's best game on top of this, where kids play two hours a day and become top 1%. Or after-school tutoring — everyone hates Kumon in a dingy building. What if you went to Dave & Buster's, did your hour with the tutor in the birthday room, earned your card, and went out to play? Or online, you get into a 10,000-person Twitch stream with your favourite streamers, earning social cred.

As a parent, what if you could put in the money for a 100-for-100, and the money went into a Robinhood-for-kids account to teach investing? Or it funded your trip to New York to meet the producer for your musical, or your robot war? Or more mundanely — screen time. Every parent struggles with how much. What if Timeback managed it: get to top 1% and earn more gaming or TikTok time? There are a hundred ways to motivate kids — social, peer, gaming. We have a multi-user version where kids compete on academics, and the 10% who love academics run out of curriculum and blow through the top of the charts.

Joe Liemandt

There's a huge debate, and I believe we'll have the biggest real dataset in the world. There are papers on both sides. There's one you can look up that says there's no such thing as intrinsic motivation, because even if you're intrinsically motivated, it's because of something — even a good student wants the accolades of being a good student. Is it a true love of learning or a love of accomplishment, and what's the accomplishment getting you?

It's like sports, where parents hold a kid back a year so they're bigger and win a little, and once they win they get jump-started and start to like it because of the accolades. Academics is the same — you just jump-start them. In our middle school we'll pay any kid $1,000 to become top 1%. When they transfer in, top 1% seems too hard, but you pay them and they get there, and it changes their internal view of themselves: 'I am a good student.' Then you don't have to pay them anymore — they do it themselves because their self-image changed.

That's the point of extrinsic versus intrinsic: if you can change a kid's internal view of their capabilities, it's worth using whatever motivation you can. Guides pull whatever lever breaks through. Kids have blocks — 'I'm not a maths-science girl.' That brain chemistry doesn't exist. Whatever the currency — $1,000, an off-campus lunch, social peer pressure — the guide uses it to get them to top 1%, and then it sticks. Our kids at college aren't getting paid, and they're all 4.0 students.

There is a bad way to do it. Starting in kindergarten, our kids earn 'alphas,' because if you want to teach financial literacy — earn, save, spend, invest, and donate — they start that cycle young, with bigger dollars as they get older. A phantom-stock scheme in seventh grade is terrible, because it's not real money they earned, so they'll YOLO it — that's not investing, it's gambling. But real money they earned, that's hard, and they'll invest it. We do let them YOLO and lose it, and it's the best lesson — better to learn that at seventeen than at twenty-seven. Maria Montessori believed kids should be involved in real-world things and that adolescents should earn money — that work has value is good for development. Ask alpha parents: we're not creating coin-addicted kids who won't learn without pay. That's just not how it works.

Take Maddie, who's doing the musical — sorry, Maddie, I just outed you. She got a 790 on her math SAT. She was below it, but she needed money to fly to New York to meet the producer. Timeback told her how many hours to get to a 790 — we were actually going for 800 — and I said, go do it, earn the money, and we'll buy your plane ticket. She got there. She was happy at 750 or 760, and then you just move it up.

Joe Liemandt

Both things you said are correct. Without an AI tutor, you can study a long time and your SAT scores don't move — it's ambiguous. We're not ambiguous. Our dashboard says you're 17 hours away, or six weeks from finishing; do an hour of homework and you're a week away. It's definitive and small. Go look at Math Academy — it'll say '28 hours away.' They just released their SAT program, and I think it's about 180 hours. That's it. We're 10 times faster — 20 hours, not 200.

Joe Liemandt

That we transform education for a billion kids, so it's nothing like what we had. My tagline is that our job is to make this the best time in history to be a 5-year-old.

Joe Liemandt

I loved it. Thanks for the time.