I listened to Shane Parrish’s interview with Joe Liemandt on The Knowledge Project, Alpha School and the Future of Education. Liemandt founded Trilogy Software, and he’s now betting heavily that AI can help kids learn much faster. At Alpha School, students spend about two hours a day on AI-driven, mastery-based academics, and the rest of the day on leadership, entrepreneurship and projects they care about.
What struck me wasn’t the technology. It was how cleverly the school is put together around it.
Hire people for what they’re best at
Alpha’s staff are deliberately picked to excel at one job. The adults who work with students are guides, much like Acton Academy’s approach (I wrote about my own run at a Socratic school in Only Ask Questions). They motivate, coach and encourage. They don’t lecture.
And here’s the part I liked most: the guides don’t have to deal with parents. A separate team, hired for exactly those skills, works with families. Anyone who has worked in a school knows how much of a teacher’s energy goes into parent communication. Splitting it out lets each person do the work they’re best at.
Kids will admit to a tutor what they won’t admit to a teacher
Today’s kids are comfortable with tech and social media. One result: they are more comfortable sharing what they don’t understand with an AI tutor than with a teacher in front of their classmates.
That matters, because it attacks one of the biggest barriers in schooling. A student falls behind on a concept, but moves up a grade anyway, and nobody goes back to fill the gap. The gaps stack up. A patient tutor that a student is willing to be honest with can find those gaps and close them, one at a time, before moving on. No lectures, and no moving forward until the material is mastered.
Confidence, tracked and acted on
As gaps close, kids’ confidence builds, and Alpha tracks all of it. Liemandt describes students moving up the scores across the board.
That’s the lesson I’d underline for any district: measuring is only half of it; acting on the measures is what moves students. It’s the same principle behind Actful-KPI, where each KPI is built around the question “what would change what we do tomorrow?”
Not without a struggle
Of course, getting this into a regular school is harder. I asked a teacher in our extended family whether her school might consider bringing in an AI tutor. Her answer was a useful reality check. Her province is moving to restrict social media and AI chatbots for youth, and her school has gone the other way entirely: a nature-based approach, with Chromebooks taken out of classrooms and a shared tech cart for research.
I understand the instinct. Kids and screens have a rocky history. But I worry that the students who never learn to work with AI will graduate into a job market that’s already being reshaped by it. This is the fastest-moving tech wave I’ve seen, and large tech firms are already cutting staff as AI takes on more work.
The beneficial side of tech won’t surface on its own. Some schools will lead, some will wait, and some will push back. In my experience the smaller, more adaptable schools and districts are the ones most ready to try.
What we’re trying
At Celerity we’ve been experimenting with Claude, an AI assistant you can teach specific skills and then let loose for users to ask questions. We’re testing whether it can be shaped into the kind of patient, Socratic tutor described in the podcast. I’ll share what we learn.
If your school or district is looking at AI tutors, or deciding not to, I’d like to hear why: info@celeritysw.com.
References
- The Knowledge Project, Joe Liemandt: Alpha School and the Future of Education
- MobileSyrup, Manitoba to ban social media and AI chatbots for kids (April 2026)
- Related posts: From Obedience to Curiosity and Only Ask Questions
