Math that sounds like your life.
Make Math Mine writes practice problems around the things you already care about: cooking, money, the garden, your team. Not as decoration. Because a problem you can picture is a problem you can reason about.
Most apps personalize the order. Not the math.
Adaptive math apps are good at sequence: which skill comes next, when to review, how hard to push. The problems themselves stay generic: trains leaving stations, widgets coming off assembly lines. Make Math Mine starts where they stop. When you practice, the problem itself is written around one of your interests, on the spot.
Why that’s hard, and measured to be hard
In 2026, researchers studied what happens when an AI rewrites math problems around a student’s interests. The math mostly held up: errors showed up at about the rate of human teachers. What broke was the world of the problem: quantities that made no physical sense, and stories that name-dropped an interest without sounding anything like it.
Their hardest finding: when they asked AI to score whether a problem felt authentic, the AI barely agreed with human judges, and the human judges barely agreed with each other. Whether a problem feels like your world isn’t something a model can score, or another person can score for you. Only you know.
The study: arXiv:2604.05160, a multi-agent pipeline for validating and refining LLM-personalized math problems.
So we built the whole thing around that
The math can’t drift.
Generated problems are built from each unit’s authored templates and problems: set structures, set parameter ranges. The story changes; the equation underneath doesn’t.
Every fresh problem gets a second read.
Before a newly generated problem reaches you, it’s checked for realistic quantities and plain phrasing, and the check runs harder exactly where the research says stories slip most, like sports scenarios and measurement units.
You’re the judge of “feels like mine.”
A thumbs-down pulls that problem out of rotation for review, and offers you a different one on the spot. Three thumbs-up from learners promote a problem for everyone. The more it’s used, the more the library sounds like the people using it.
What we don’t claim
That study measures problem quality, not learning speed, so we won’t tell you interest-based problems make you learn faster. We claim something smaller, and checkable: the problems will sound like your life, the math underneath is solid, and when one misses, telling us takes one tap and gets you a better one immediately.
See for yourself
Free Practice is free. Pick a unit, pick an interest, and look at what comes back.