Tegore Is Building Duolingo for Math, and the Classrooms Are Already Using It

The Macro: Math Education Has a Tutoring Problem

Here’s the thing about math tutoring: it costs between $40 and $100 an hour for anything decent, and most families can’t sustain that. The kids who need help the most are the ones least likely to get it. Private tutoring is an industry worth north of $100 billion globally, and the distribution of that spending is about as equitable as you’d expect.

The obvious response is “just use Khan Academy.” And look, Khan Academy is genuinely great. Sal Khan built something important. But Khan Academy is a library. It’s organized by topic, not by your kid’s specific class, not by the pacing guide their teacher is using, not by the chapter they bombed on last Tuesday. The gap between “here’s a video on quadratic equations” and “here’s help with the exact thing your Algebra I teacher assigned tonight” is wider than it looks.

Photomath and Mathway went after the homework-help angle, but they mostly became answer-lookup tools. Parents figured that out. Teachers figured it out faster. There’s a reason schools started banning them alongside ChatGPT. The product worked too well at the wrong thing.

What nobody has really nailed is a product that sits between “full human tutor” and “search engine for math answers.” Something that actually teaches, follows the curriculum, and adapts to where a student is struggling. That’s the gap. It’s been obvious for years. The question has always been whether AI was good enough to fill it without producing garbage explanations or, worse, confidently wrong math.

The Micro: One Founder, Bay Area Classrooms, Real Students

Tegore is a one-person company right now. Ende Shen is building an AI math tutor that aligns directly with K-12 school curricula, covering grades 6 through 12. The product uses voice and interactive visuals to walk students through lessons, and the teaching approach is dialectic. Instead of just showing you how to solve a problem, it asks you questions to figure out where your understanding breaks down, then rebuilds from there.

That last part matters more than it sounds. Most AI tutoring products dump an explanation on you and call it done. The dialectic approach means Tegore is trying to do what a good human tutor does: probe, identify the actual gap, back up, and teach from the right starting point. A student who can’t factor polynomials might actually be confused about the distributive property. A product that just re-explains factoring is wasting everyone’s time.

Shen is part of YC’s Spring 2025 batch and has already gotten the product into Bay Area classrooms. Ryan Stagg at Tide Academy in Menlo Park has endorsed it. A ninth-grader named Sofia said it “can explain a topic without a real human needing to teach,” which is both the dream testimonial and the core product promise in one sentence.

The product covers the full 6-12 math curriculum, which is ambitious for a solo founder. The mobile app is still coming, so right now this is browser-based. No pricing is visible on the site, which probably means the school partnerships are the current go-to-market and the consumer product is still being shaped.

The “Duolingo for Math” comparison is one Tegore is making explicitly, and it’s a smart frame. Duolingo proved that structured, curriculum-aligned, bite-sized lessons can drive retention at scale. Nobody has done that for math in a way that stuck. IXL is the closest, but IXL feels like a worksheet engine, not a tutor. The experience gap is real.

What I like about the approach is the specificity. This isn’t “AI for education” in the vague, pitch-deck sense. It’s “AI that knows your kid is in Algebra I, Chapter 7, and is confused about slope-intercept form.” That level of curriculum alignment is hard to build and harder to maintain, because curricula change by state and district, but it’s what makes the product useful instead of just impressive.

The Verdict

I think Tegore is pointed at a real gap. The math tutoring market is massive, the existing products either cost too much or teach the wrong thing, and AI is finally good enough to have a real conversation about a quadratic equation without hallucinating the answer.

The risk is obvious: this is a one-person team trying to cover an entire K-12 math curriculum while simultaneously building a consumer product and servicing school partnerships. That’s a lot of surface area. The classroom traction is encouraging because it means the product is being tested by actual students and teachers, not just demo’d to investors. But scaling curriculum coverage and maintaining quality with a team of one is going to hit a wall at some point.

Thirty days from now, I want to know how many schools are actively using it and what the student completion rates look like. Are kids finishing lessons or dropping off after two minutes? Sixty days, I want to see the mobile app. Browser-based math tutoring for teenagers is a friction problem. Ninety days, the question is whether Shen has hired anyone. The product ambition here requires more than one person, and the timeline between “promising classroom pilot” and “Duolingo-scale consumer product” is measured in years, not months. But the starting point is strong, and the problem is worth solving.