The Intelligence Layer for Learning
BlueberryML builds a framework for adaptive learning — the mathematics, the implementation, and the technology that surfaces it. Deployed through partners as ClassGrade.
300,000+
students
150+
schools
50,000+
learning outcomes mapped
4
curricula
The Framework
A geometric model of knowledge
BlueberryML's framework maps every learning outcome in a curriculum, tracks each student's position, and determines the optimal next step. Built on Riemannian geometry — where the metric tensor encodes the pedagogical cost of moving between concepts.
How the framework works →The Pattern
Deployed through partners
The framework reaches schools through ClassGrade — the partnership pattern under which partners deploy it on their own infrastructure, under their own brand, with full data control. Our flagship deployment is Chrysalis, operated by EZ Vidya across 300,000+ students and 150+ schools in India.
See the partnership pattern →Technology
What powers the intelligence layer
Knowledge Graph
50,000+ learning outcomes connected by prerequisite edges. The framework navigates this graph in real-time.
Learn more →Adaptive Engine
Field-theoretic intelligence: divergence, entropy, and Jacobian estimation determine what to teach and when.
Learn more →Curriculum Mapping
Cambridge Primary, Cambridge Early Years, UK National Curriculum, and Common Core mapped to a single coordinate system.
Learn more →Vision
Our thesis: learning is navigation through a knowledge space. The geometry of that space — how concepts connect, how complexity varies, how understanding propagates — determines everything about how a student should learn.
The framework is an active intellectual project, not a finished asset. Products are the commercial expressions of where the framework currently is.