See pricing
Reinventing School for the AI Era | Joe Liemandt, Gagan Biyani & Erik Torenberg
The a16z ShowEducation

Reinventing School for the AI Era | Joe Liemandt, Gagan Biyani & Erik Torenberg

Joe Liemandt argues that if AI can compress core academics into two hours a day, the real design problem is what to do with the rest of the time. The episode is really about a broader product lesson: when technology makes the old workflow cheaper, you should redesign the whole experience, not just speed up the old one.

October 27, 2026 · 57 min listen · 16 min read · Joe Liemandt, Gagan Biyani
0:00
–:––

Context

Joe Liemandt, founder of Alpha School, and Gagan Biyani, co-founder of Horowitz and Andreessen Academy, discuss what education should look like if AI makes traditional instruction dramatically more efficient. Their argument is not just about schools. It is a useful product lesson about what to do when a new technology collapses the cost of a core workflow. For PMs, the episode matters because it shows how to redesign an entire experience around motivation, personalization, and real-world outcomes instead of simply inserting AI into the old process.

The Big Idea

When AI makes the old core task much faster, the winning move is not to keep the old container and finish earlier. The winning move is to redesign the rest of the experience around the human capabilities that are still scarce.

Joe applies that logic to school: if core academics can happen in roughly two hours, the rest of the day should focus on agency, teamwork, communication, entrepreneurship, and doing hard things. For PMs, the same logic applies to products and workflows. Efficiency gains only matter if you rethink what users should do with the capacity you just created.

Key Insights

AI changes the time budget

Joe's core claim is that cognitive science, learning science, and LLMs make it possible to teach standard academic material far faster than the one-teacher, one-classroom model.

  • He argues the underlying learning engine can teach kids 10 times faster than the traditional model.
  • At Alpha, students spend roughly two hours a day on core academics instead of six classroom hours plus homework.
  • Joe also says students learn twice as much in those two hours as they would in a conventional setup.

Why this matters to PMs: when a technology radically lowers the time cost of a core job, the strategic question is no longer how to improve the old step by 10%. The strategic question becomes what new value should replace the time that is no longer needed.

Motivation beats curriculum design

Joe's framing is unusually blunt. The first design question is not whether AI is impressive. It is whether leaders are willing to redesign the entire day so users love the experience.

  • Alpha's first promise is emotional, not academic. Joe says 90-some percent of students report loving school.
  • Roughly 50% of students say they would rather go to school than go on vacation. One campus reached 65%.
  • Joe insists this does not come from lowering the bar. His claim is the opposite: kids like doing hard things, especially with friends.

That is a useful product lesson. Motivation is not just delight layered on top of utility. Motivation changes whether people persist long enough to get the value in the first place.

Personalized challenge drives engagement

Joe describes a simple target for keeping students engaged: the work should feel challenging but winnable.

  • Alpha aims to keep students in an 80 to 85% correct range.
  • If performance is near 100%, the work is too easy and learning stalls.
  • If performance drops below about two thirds, Joe says people start to disengage, similar to how game designers think about failure loops.

The system adapts difficulty per student. An eighth grader who missed a fifth grade concept does not stay trapped in age-based pacing. The system gives the missing fifth grade lesson, gets the student to a win, and then moves forward.

For PMs, this is a strong reminder that users do not quit only because a product is bad. They also quit when the challenge curve is miscalibrated. Too easy feels pointless. Too hard feels punishing.

Projects teach what classes miss

Joe's most concrete point is that schools often claim to teach teamwork and leadership, but their actual structure teaches the opposite.

  • In traditional group projects, one student often does the work while everyone else waits. Joe calls that anti-teamwork.
  • At Alpha, third graders do escape room exercises while AI analyzes the transcript of their collaboration.
  • Joe says less than 20% of the language in an early session is team-oriented.
  • After an eight-week teamwork workshop, students must reach 85% team behavior to pass.
  • Joe says teams also finish the escape room three times faster once they collaborate well.
  • By fourth and fifth grade, students take the same Wharton teamwork simulation that Joe says 50% of MBAs fail.

The product lesson is that soft skills rarely emerge from slogans. If a capability matters, you need a designed environment, clear feedback, and a measurable standard.

Believability unlocks adoption

Gagan praises Joe not just for the idea, but for making the idea believable. Joe says his background in product and entrepreneurship helped him package the model in a way parents could understand and trust.

A few specific signals matter here:

  • Alpha grew from 12 campuses to 50 in a year.
  • Joe says some early info sessions had 1,000-plus interested people but still struggled to enroll the first 25 families.
  • One New York City campus opened with roughly 10 to 12 students, then later filled to about 150 once results were visible.

This is a useful PM lesson for any category-creating product. The hard part is often not the invention. It is making the new behavior legible enough that a cautious buyer can believe it will work.

College should center real building

Both guests argue that higher education has the same redesign problem, but with a different user context.

  • Joe says elite interns from places like MIT, Stanford, and UT told him they were scared university was not preparing them for an AI-shaped job market.
  • He says the students liked seminars, clubs, people, and side building, but saw big lecture classes as a charade.
  • Joe's line is that a university is a bundle of products. The network and environment still matter, but the lecture-heavy academic bundle looks increasingly weak.
  • Gagan argues that college-age students need to build, interact with frontier operators, make mistakes, and learn in service of real work.

For PMs, this is a familiar unbundling story. A product can retain value overall while one major component becomes obsolete and needs to be replaced.

Optimism beats defensive planning

One of the strongest shared ideas is that students should not pick a path by asking which field is safest from AI.

Gagan argues that is a defensive frame. Joe adds that the goal is to raise a generation that sees AI as a tool for building the future, not just surviving it.

The logic is:

  • AI makes domain learning faster. Joe calls this cognitive onload.
  • AI also gives students builder leverage much earlier.
  • That combination means a motivated student can reach expertise and ship something meaningful faster than past generations could.

For PMs, this is a reminder that strategy built around fear usually underperforms strategy built around agency. The question is not only what AI threatens. It is what AI now makes newly buildable.

Mental Models & Frameworks

The 80 to 85% zone

Joe's engagement model is simple and reusable.

  • Too easy: near-perfect performance means the user is not really learning or growing.
  • Too hard: frequent failure causes disengagement.
  • Right zone: around 80 to 85% success keeps people challenged and progressing.

Use this model when designing onboarding, tutorials, learning loops, team stretch goals, or workflow automation. It is a good check against both over-handholding and frustration.

Education as a product bundle

Joe's framing of university as a bundle of products is broadly useful.

The bundle includes things like:

  • academic instruction
  • network and relationships
  • identity and prestige
  • access to ambitious peers
  • real-world building opportunities
  • exposure to experienced operators

When AI weakens one element, such as lecture-based knowledge transfer, the right response is not necessarily to kill the whole bundle. It is to ask which parts still create value and which parts must be redesigned.

Adoption starts with believers

Joe repeatedly comes back to a classic adoption-curve point: early cohorts matter more than broad appeal.

  • Alpha did not need everyone to agree at the start.
  • It needed a small set of families who already believed the old system was broken.
  • Those families became a community asset and proof source for later adopters.

Joe also makes the marketing implication explicit. A strong product definition should attract the right users and repel the wrong ones. That is often healthier than softening the product until nobody objects.

Raise ambition, not just skills

Joe references the idea that one of education's most important jobs is to raise a student's ambition for themselves.

That is more than inspiration. It is a design principle:

  • bring students close to people doing meaningful work
  • make hard goals feel reachable
  • show that impressive operators are still ordinary humans
  • increase the user's sense of what they can attempt

This model applies well beyond schools. Good products do not just improve execution. They expand what users believe is possible for them.

Decision Principles

Principle: Redesign the whole experience

When: a new technology makes a core step dramatically faster or cheaper.

Why: Joe's argument is that speed alone does not create the full value. The real leverage comes from deciding what the user should now do with the time, energy, or budget that technology freed up.

Principle: Spend saved time deliberately

When: an efficiency gain creates obvious spare capacity.

Why: Joe says sending students home after two hours would miss the point. The scarce value is in what fills the rest of the day. PMs should treat freed capacity as a strategic asset, not as empty space that will somehow fill itself.

Principle: Sell users and buyers

When: the people who use a product are different from the people who approve, fund, or de-risk it.

Why: Joe notes that students may want a new school model quickly, while parents need a separate belief-building process. Many PMs face the same split between end users and economic buyers, and both audiences need different proof.

Principle: Start with conviction segments

When: a new product category feels risky or strange to mainstream customers.

Why: Joe advises targeting people who already feel the current system is not working. Those users tolerate inevitable early hiccups, give better feedback, and create the first credible success stories.

Trade-offs & Nuance

Fun needs real rigor

Joe pushes back hard on the idea that a lovable school must be soft or unserious.

The trade-off is not fun versus rigor. The trade-off is shallow entertainment versus meaningful challenge. Joe's claim is that students love environments where they do hard things with friends and can feel themselves improving. If the work stops being demanding, the model loses credibility.

Screen time needs outcomes

Joe knows that some families hear AI, screens, and software and immediately worry about another failed edtech experiment. His answer is that the tools only work if the school day is redesigned around them.

That creates a real tension:

  • some families will reject the model on principle
  • some families will accept the tools if the outcomes are visible
  • the wrong families can become a drag on an early-stage product

The product lesson is that controversial enabling technology needs a strong outcome story, not just a technical story.

K12 and college differ

Gagan stresses that each age band needs a different design. Joe agrees and says Alpha's K-8 model is different from its high school model, and college should be different again.

The useful nuance is that principles may transfer while implementation does not. Motivation, agency, projects, and real-world learning travel well. Exact structures do not.

Safety can become risk

Joe says many parents still see the traditional path as safer, especially once a student is already near college admission. But he argues that in a changing labor market the old path may actually be the higher-risk choice.

The nuance is timing. Parents of younger students may be open to alternatives. Parents of students already at the five-yard line of the old system may optimize for avoiding disruption.

Common Mistakes

Mistake: Layering tech onto lectures

Joe says many past edtech efforts failed because they were inserted into the old school day instead of changing the structure around them.

The failure mode is easy to recognize: a new tool improves one task, but the surrounding process, incentives, and expectations stay the same. In that setup, the tool looks disappointing even if the underlying capability is strong.

Mistake: Confusing ease with motivation

Joe's model is not about making school easier. It is about making progress visible and attainable.

When teams confuse motivation with comfort, they often remove the very challenge that makes an experience feel meaningful. Users do not stay because everything is effortless. They stay because the effort pays off.

Mistake: Faking teamwork instruction

Joe's critique of normal group projects is that they claim to teach collaboration while actually rewarding freeloading and uneven contribution.

The better approach is to observe collaboration directly, coach the behaviors, and set a real bar for passing. If a team capability matters, do not assume it will emerge from group seating charts and shared slides.

Practical Application

Audit your time budget

Pick one workflow where AI has already made your team materially faster, such as writing first drafts, customer synthesis, or prototyping.

Then force three explicit decisions:

  • what time was actually freed
  • what low-value work should disappear entirely
  • what higher-value work should now expand because capacity exists

This follows Joe's core logic. The value is not the speedup alone. The value is the redesign that follows the speedup.

Instrument the challenge curve

Review one part of your user journey or team workflow where drop-off is high.

  • Identify whether people are failing because the task is too hard, too easy, or poorly scaffolded.
  • Add support for users who miss prerequisite knowledge, just as Joe describes giving a student the lesson they actually need.
  • Track whether completion improves when the task feels winnable but not trivial.

A target like Joe's 80 to 85% success band is a useful starting benchmark.

Turn skills into projects

If you want a team to improve at communication, ownership, systems thinking, or collaboration, do not rely only on workshops or abstract coaching.

Create a real project with visible stakes, shared outputs, and review afterward:

  • what behaviors showed up
  • what slowed the group down
  • what would better teamwork have changed

Joe's teamwork example works because the skill is practiced in a live environment, not discussed in theory.

Recruit believers, not everyone

For a new product or internal process, write down which users already feel the status quo is broken. Start there.

Joe's point is that early adopters with conviction are not just easier to close. They also create the initial proof that later buyers need. Trying to convince skeptical late adopters first often slows learning and muddies the product.

Separate users from buyers

If your product has one group that uses it and another group that approves it, build separate evidence for each group.

A student may care that school feels meaningful. A parent may care that the path still leads somewhere credible. In B2B software, an end user may care about speed while a budget owner cares about risk, control, or ROI. Treating those as the same sale usually produces weak messaging for both.

Questions to Consider

  • Where has AI already made one core task in your product or team dramatically faster, and what higher-value work should replace the time that speedup created?
  • Which part of your onboarding, education flow, or internal process keeps people outside an 80 to 85% success band, either by boring them with tasks that are too easy or losing them with tasks that are too hard?
  • Are you trying to win skeptical mainstream customers before delighting the smaller group that already believes the current system is broken?
  • If one audience uses your product and a different audience approves or pays for it, what proof does each audience need before they will adopt the new model?
  • What important capability in your team, such as collaboration, ownership, or communication, are you still trying to teach through discussion instead of through a real project with observable feedback?

Bottom Line

This episode is really about redesign after compression. If AI makes a core activity far faster, the opportunity is not to preserve the old experience and end earlier. It is to rebuild the experience around motivation, calibrated challenge, and the human capabilities that matter more once routine work gets cheaper.

Resources Mentioned

ResourceTypeWhy it was mentioned
Horowitz and Andreessen AcademySchool websiteThe college-age program Gagan Biyani is building, centered on real-world projects, relationships, and frontier operators.
Alpha SchoolSchool websiteJoe Liemandt's school model, which compresses core academics and uses the rest of the day for life skills and projects.

Case Studies Mentioned

Alpha School model

Alpha's K-12 model is the clearest case study in the episode. Joe says students spend about two hours a day on academics, learn more than they would in a conventional full-day setup, and use the rest of the day on projects covering leadership, teamwork, entrepreneurship, communication, socialization, and grit. The business lesson is as important as the education lesson: Joe says the model only worked once Alpha productized it, proved it, and made it believable to parents.

HAA one-year structure

Horowitz and Andreessen Academy lowers adoption risk by offering a one-year experience rather than asking students to make an irreversible bet. Gagan notes that students can return to college afterward, which makes the decision less like dropping out and more like taking a structured detour. The product lesson is that shortening commitment can make a new category much easier to try.

Founder School guarantee

Joe describes Alpha's entrepreneurial high school, Founder School, as a specialized program for students who want to build businesses during high school. The striking claim is a tuition-back guarantee if a student does not start a business that earns $1 million of profit while in school. The path is scaffolded through revenue milestones of $100, $1,000, $10,000, $100,000, and then $1 million. The lesson is that Joe sees entrepreneurship as a teachable stack of skills, not a mysterious talent.

People to Follow

Joe Liemandt

Founder of Alpha School and the main advocate in this episode for redesigning education around AI-enabled efficiency, motivation, and real-world skill building. He brings a product and entrepreneurship lens to schooling, which is why many of his arguments transfer well to PM work.

Gagan Biyani

Co-founder of Horowitz and Andreessen Academy and a longtime education entrepreneur. In this conversation he pushes the idea that learning works best when it is tied to building, relationships, agency, and real-world feedback rather than passive classroom formats.

Notable Quotes

Are you willing to redesign your school day so kids love it more than vacation? (Joe Liemandt)

Kids want to go do hard things, right? They love to do hard things and they love to do hard things with their friends. (Joe Liemandt)

A university is a bundle of products, right? (Joe Liemandt)

We need to create a generation of builders who look at this optimistically, right? (Joe Liemandt)

AI PM course

Everyone hears the same episodes.
Few can do what they describe.

Start for free