Summary

Alpha School’s core claim is that children can reach top-1% academic performance in two focused hours a day, then spend four hours building life skills they value. Joe Liemandt says a grade-level subject typically takes 20-30 hours to master when an AI tutor keeps each learner at an 80-85% success rate. The bargain for students is simple: real engagement in exchange for “time back.”
The investable moat in Liemandt’s account is not the chatbot but a learning-science engine joined to a motivation system. Only 5-10% of children naturally engage with products such as Khan Academy, he argues, so even excellent EdTech fails without incentives. Alpha’s design starts from “don’t waste my life. Don’t waste my 12 years,” then uses personalized pacing, competition, money, or afternoon passions to sustain effort.
Elad Gil warns that unrestricted ChatGPT is actively harmful in school because most students use it to avoid learning. “It is a cheat bot. It is not a chat bot,” he says. Liemandt agrees that the current form is terrible and argues that AI does not eliminate the importance of stored facts and writing: critical thinking requires facts, while “writing is thinking.”
Removing classroom lectures changes school labor economics as much as pedagogy. Alpha pays guides at least $100,000 to provide motivational and emotional support, while software handles personalized instruction; guides are accountable for love of school, 2x learning, and life skills. Larger groups can then support higher-paid coaches because individualized academics no longer depend on class size.
The commercial wedge is an already-large private market supplemented by expanding school-choice funding. Liemandt cites a US private-school market above $50 billion, $100 billion of annual Texas education spending, roughly $1 billion of forthcoming Texas vouchers, and perhaps $200 billion of federally enabled vouchers. Alpha is expensive by design, but newer formats run from $15,000 to $25,000, with the low end below average public-school spending.
Alpha embraces extrinsic incentives where they build habits, repair self-belief, or finance creation. Middle schoolers can earn $1,000 for reaching the top 1% in a subject, invest real money through a child-oriented Robinhood account, or fund passion projects such as an all-teen Broadway musical. Liemandt’s sharpest argument is that a reward can break the belief “I’m not smart enough,” even after the money is spent.
The scaling opportunity is enormous, but so is the adoption risk. Liemandt says 99.9% of parents initially resist abandoning the teacher-at-the-front model, and it took Mackenzie Price two years to win him over. Public-school transition could take a decade. He says he supplied “the first billion” as seed funding, yet expects rebuilding education to require hundreds of billions, 10,000 buildings, rigorous trials, and a generation of builders.

Deep dive

1. GenAI turned a family experiment into a billion-child ambition

Liemandt’s path began with a high-school AI paper declaring neural networks “decades away,” followed by Stanford, dropping out, and building what he calls the first 1990s AI product to sell $1 billion. After 25 years in enterprise software, GenAI convinced him that neural networks could finally scale education rather than merely improve a local school.
He initially rejected Alpha for his own children because apps without classroom teachers looked implausible: “Everybody knows good school equals good teacher.” It took Mackenzie Price two years to win him over, illustrating the behavioral barrier now facing the business—families want familiar schooling until a sufficiently large community proves the alternative.
Becoming principal was product discovery, not a ceremonial role. Liemandt wanted to understand fifth-grade fights, parent complaints, buildings, peer groups, and adult support from the ground up: if families still drop children at a shared location, how should those 12 years be redesigned to “unleash human potential”?

2. Alpha sells children two hard hours in return for their time

Alpha’s foundational commitments came from an early co-founder: “Kids must love school,” and once they do, “your expectations of your kids are too low.” Today, Liemandt says 96% report loving school, while 40-60%—depending on the vacation and recent workshops—say they love it more than vacation.
Fifth graders supplied the product specification. Asked what would make them love school, they answered “less school,” eventually accepting two hours of genuinely engaged academics for “four hours of awesomeness.” Liemandt’s team then had to fit the academic load into that negotiated window and make the learning durable.
A day begins with “Limitless Launch,” described as Tony Robbins for children, followed by personalized math, science, language, and reading. Once the apps turn green, afternoons shift to leadership, teamwork, grit, entrepreneurship, financial literacy, storytelling, public speaking, relationships, physical challenges, and other project-based workshops.
The through-line is that school remains a bundle—academics, socialization, physical activity, life preparation, and childcare—but lecture consumes too much of it. The product promise is not merely faster coursework; it is returning most of the day for activities students and parents consider meaningful.

3. Personalization starts with knowledge grade, not age grade

Liemandt’s diagnosis is blunt: “Age grade and knowledge grade are two totally different things.” He says transfers with straight-A private-school transcripts test anywhere from one year ahead to three behind; B students range from three to seven years behind, and some entering freshmen cannot write a grammatically correct third-grade sentence.
His reassuring counterclaim is that a grade-subject combination—fourth-grade math, for example—takes only 20-30 hours to master. A child who appears hundreds of classroom hours behind may need roughly 60 focused hours; adding a third daily hour can therefore repair the gap in about 60 days.
The engine targets an 80-85% success rate. At 99%, students already know the content; at roughly 66% or below, they disengage. An AI tutor can generate limitless scaffolding inside this “fun struggle zone,” something a teacher delivering one lesson to 30 differently prepared children cannot do.
Liemandt rejects throwing students far over their heads as grit training: giving him the MCAT would not teach medicine, while rolling him back to freshman biology might. He claims the same progression can move weak high-school writers through earlier material and, within weeks, toward work that could earn a five on AP Lang’s free-response section.

4. High standards replace lecturing as the adult’s primary job

“The key to your child’s happiness is high standards,” Liemandt argues. Parents accept this athletic ethos from championship coaches yet resist it academically; Alpha’s workshops recreate the cycle of supported struggle, failure, tears, and success—including kindergarteners climbing 40-foot rock walls.
With apps delivering content, guides provide the work memorable teachers already did best: convincing children they can accomplish difficult things. During 25-minute academic Pomodoros, a guide can pull one student aside, ask about the weekend or softball, and diagnose disengagement rather than grade another seventh-grade science quiz.
Alpha pays guides at least $100,000 and expects each to deliver love of school, 2x learning, and life skills to every child. Students are asked whether their guide has transformed their life; middle and high schoolers even interview candidates, favoring adults who will make them better rather than simply make school easier.

5. Motivation is infrastructure, and money is one available tool

Time is the dominant reward. Liemandt contrasts Alpha with an AP-and-Ivy-track high school requiring six classroom hours plus four hours of homework: offer the same student two to three academic hours, a path to 1550-plus SAT scores and AP fives, and four years of afternoons for meaningful work, and “you can’t pay” enough to match the relief.
Guo raises the concern that stickers, leaderboards, cash, and time back might crowd out intrinsic motivation. Liemandt’s categorical answer is “that’s just not true”: different students require different triggers, and creating daily habits or breaking a limiting self-conception matters more than preserving an idealized motivation purity.
Alpha Bucks begin in kindergarten as an economy for earning, saving, spending, and donating. Older students can receive $1,000 upon reaching the top 1% in a subject, then invest it through a child-oriented Robinhood account; real losses are permitted because simulated portfolios encourage consequence-free YOLO behavior.
The best specimen is Liemandt’s younger daughter, who believed top 10% was sufficient and that her older sister was “the smart one.” A $1,000 shopping bet got her to top 1%; afterward, she told her mother the lasting reward was realizing she could match her sister “if I just put in the work.”

6. The model treats disengagement and diagnosis as system effects

Liemandt says America’s time-based progression produces “Swiss cheese holes”: without fractions, algebra becomes difficult; without algebra, chemistry follows. He claims the median high-school student gains only one point on a 300-point scale over four years, while students around the 99th percentile gain roughly six points annually.
His proposed repair is “100 for 100.” Seventh graders unwilling to revisit fourth-grade material eagerly take a third-grade Texas STAAR test for $100, advance until they score 75 or 85, then let AI generate the missing lessons. For $400-$500 per child, he argues, the country could restore foundations and show that mastery reflects effort, not a fixed “smart kid” identity.
On ADHD, dyslexia, IEPs, medication, and adolescent mental health, Liemandt hedges that diagnoses may not be excessive within the existing system. His stronger claim is that many difficulties look different when children need not sit bored for six hours and instead receive individualized material plus an afternoon built around action.
One workshop combines a values chart, Japanese ikigai, and an accounting of all 168 weekly hours. A girl who wanted an ambitious life discovered her schedule pointed toward becoming “the best TikTok scroller”; Alpha’s middle-school objective is to turn consumers into creators by giving them time and structured opportunities to build.

7. Private schools and vouchers provide the first scalable market

Alpha itself was designed by pretending “price is no object,” but lower-cost brands vary the afternoon bundle. A sports academy needs fields and coaches; a gifted-and-talented school uses robotics, Math Olympiad, books, and academic enrichment. Current variants cost roughly $15,000-$25,000, with Montessori another planned format.
Personalization is not one cultural template. Liemandt’s older daughter would happily study all day and created equation-based writing for STEM students; his younger daughter wants minimal academics, millions of TikTok followers, relationships, and entrepreneurial projects. Morning mastery earns each type more of the afternoon that motivates her.
The economics change because software personalizes academics. Liemandt says the high-school guides he asked would prefer twice the group size if it meant double pay and a more awesome coach; the $15,000 sports model approaches a $10,000-$11,000 Texas voucher with only $400-$500 monthly parent contribution.
Liemandt explicitly wants for-profit companies with “mission and money as a purpose.” TimeBack will package the academic system so builders can build school models around it, while a triple-A game team is building a free-to-learn layer it believes could become the most profitable game ever—premised on 10 million US mothers paying $100 monthly for top-1%-to-5% outcomes.

8. Unconstrained ChatGPT undermines the knowledge AI needs

Elad Gil’s warning is unambiguous: give students ChatGPT for normal academics and “90% will use it for cheating.” “It is a cheat bot. It is not a chat bot,” he says. Liemandt agrees that the current form is terrible and says the system should be rebuilt around learning science rather than simply deploying ChatGPT.
“Writing is prompting” is the dangerous shortcut; “writing is thinking” is Liemandt’s rebuttal. Students restructure knowledge by composing it, while critical thinking cannot operate without stored facts. Parents who want children to know nothing but reason well are asking for the human equivalent of an LLM hallucinating without a factual base.
Multiplication tables are his clearest learning-science example. If 7 × 8 still consumes working memory during an advanced problem, the student is “doomed” to overload and careless mistakes. AI can install facts faster—“Neo uploading the Matrix”—but it does not eliminate the need for long-term schemas.
Most EdTech, he argues, ignored this science and then sold into bureaucracies with low ARPU and endless cycles. His positive example is Math Academy: 28 hours to master fourth-grade math, 26 for fifth, and 22 for sixth and seventh, backed by what he describes as a 500-page account of its learning science.

9. Public adoption requires evidence, incentives, and a long runway

Liemandt expects public-school transition to take a decade because the bundle embodies conflicting community priorities. Parents may defend a failing school because a teacher transformed an older child’s life; a superintendent cannot optimize a single outcome when every family values different combinations of academics, relationships, childcare, and neighborhood identity.
His policy request is “pharmaceutical-grade randomized controlled trials for education” at million-student scale. Early public pilots focus on MTSS Tier 3 students—the bottom 10%, where schools have flexibility—but even successful software needs gift cards, games, or another reason to engage. “If you don’t solve motivation, all this EdTech doesn’t matter.”
Liemandt says he put in “the first billion” as seed capital, while rebuilding education could require hundreds of billions and 10,000 buildings; countries may also want sovereign LLMs, with education among their central uses. His recruiting close is personal: after three years as principal, transforming the next generation feels “10x” more rewarding, and he expects this to be “the best 20 years of my life.”

Stripe's Payments Foundation Model: How Data & Infra Create Compounding Advantage, w/ Emily Sands 20VC: Nvidia Invests $100BN in OpenAI, Navan IPO & Notion's $500M ARR