One geometry, many subjects.
The manifold is a learned geometry of how a subject is understood — the shape of the knowledge, where each learner sits in it, and where they go next. The method is domain-general. English is the first domain live and calibrated on real learners; Maths and Science have their dimensions defined and mapping underway. Each subject has its own coordinate system, built on the same geometry.
Accuracy is the universal axis. Conceptual Understanding and Representation recur across Maths and Science. Each subject keeps an irreducible axis of its own — Pronunciation, Procedural Fluency, Inquiry. The shape changes; the geometry does not.
English is live today. Here’s the manifold at work.
The closed core, the open tools.
The manifold is closed — the engine these tools read from, not for sale, not exposed. Below, it’s at work in English, the first domain live. ClassGrade is the set of tools that put it to work without opening it: author against it, grade against it, report against it. You get the capability; the engine stays sealed. These tools are in production today as Chrysalis AI, a custom-branded deployment in India — and you can run them the same way: directly as ClassGrade, or as a branded build of your own.
Replace manual exam-setting and marking with one loop.
Author an exam, print it, and grade the scanned scripts — one loop, against a structured map.
The grading map is the manifold reading the work — the teacher only ever sets an exam and gets it back marked.
Replace generic report-card comments with outcome-level evidence.
Fold every source of evidence into one per-student report — written against what a child demonstrated, not boilerplate.
Every sentence traces to evidence the manifold holds on that child — the sealed engine, as a document a school can stand behind.
Replace the spreadsheet a head of school keeps by hand.
Roll every child’s evidence up to the view leadership actually acts on — by grade and subject.
The same manifold read behind every child report, rolled up to the view leadership acts on.
LSRW is the framework's read on the four language skills — Listening, Speaking, and Reading measured live on the voice surface, and Writing read from written work.
Ten activity templates cover discriminative listening, imitative and responsive speaking, reading fluency, reading for expression, and reading comprehension. Each emits a position on the five-dimension manifold the framework uses for English. Writing is read the same way — graded from scanned work in ExamBuilder, against the same space.
CEFR-aligned at the surface, so partner curricula and international assessments map to the same coordinate system underneath.
Five dimensions
Accuracy, Pronunciation, Fluency, Vocabulary, Content. Every activity returns a coordinate in this space; reports and dashboards aggregate over it.
Ten activity types
Across Nursery, PP1, PP2, and Class 1, extending upward. Reading fluency includes WCPM, prosody, and phoneme-level scoring against CMU Dict.