The method system
One spine, four layers, two engines
The whole cultivation system on one page: the path every child walks, the four layers the content unfolds through, and the two engines underneath. Why it fits how children learn lives on the approach page; the five-step flow on how-it-works; this page is the map.
The spine
From diagnostic to Demo Day
Every path, whichever program, walks the same seven stations.
- 1/7Diagnose
A five-dimension math baseline plus interests and weekly time — before any path is proposed.
- 2/7Personalized path
The diagnostic decides the starting point and which modules activate; everything else stays off by default.
- 3/7The Daily Four
Foundation → Project → Evidence → Reflection, closing the loop every single day.
- 4/7Skill gates F1–F7 / M1–M7
Fourteen capability gates, lit only by real evidence — never pre-granted.
- 5/7Evidence archive
An append-only process record: every version, every decision, kept.
- 6/7Portfolio
A publicly viewable artifact the child owns and takes with them.
- 7/7Public showing
Demo Day: the child presents the real work to family and real users.
Four layers × the books
Theory → verifiable math → practice → real project
Our programs unfold every topic through four layers. The layers are fed by a real book series — and one thing we say plainly: this four-layer frame is our product-side unfolding of the books, not terminology quoted from inside them.
- 01
Theory
Why we teach this way: education redefined for the AI era — from passive tool use to active human–AI symbiosis, from knowledge transfer to wisdom cultivation.
What the books contribute
From the series: the educational-philosophy volume, plus the learning-methodology volume — spaced repetition, adaptive testing, knowledge graphs, metacognition. - 02
Verifiable math
The claims about learning must survive formal scrutiny — stability, viable sets, what happens at the boundary when a learner is pushed.
What the books contribute
From the series: the Roll Principle volume — a mathematical framework for the stability of learning systems — plus the quantitative assessment stack (IRT, knowledge tracing). - 03
Practice
Method becomes operable: concrete routines, tools, and habits a child and a mentor can actually run every week.
What the books contribute
From the series: the technology-practice volume (concept to product), the teacher’s handbook, and the learner’s guide — prompt craft, AI-assisted work, critical filtering. - 04
Real project
Learning ends in creation: a real project with real users is the endpoint, not an optional extra.
What the books contribute
From the series: the leap from learning to creating — which in our programs becomes the child’s publicly verifiable project artifact.
Two engines
What runs underneath
Personalization is not a slogan here — it is two concrete, inspectable engines.
The math diagnostic engine
- Reads a five-dimension baseline: algebra, functions, geometry, data analysis, reading speed.
- Every error is diagnosed to a breakpoint — concept, broken step chain, calculation, or misread — and the explanation is matched to that cause, then resurfaced on a spaced schedule. Mastery can go down as well as up.
- Coach, not solver: hint ladders are adversarially audited so the first hint is designed never to reveal the answer.
- This engine drives the RUDO math companion and the starting diagnostic of every path. We never promise a school place or a test score.
The ten-volume AI-education series
- The content source of the AI-literacy line: a 10-volume main series within a larger 21-volume, 170-chapter body of work by Xiaowen Kang, written in Chinese with a full English translation per its own records.
- Machine-queryable: a built-in search index over 843 chapters and 33,575 chunks, so lessons cite the source instead of paraphrasing from memory.
- Its value layer — the Five Bottom Lines — is ours too: slow over fast, low tech debt, human dignity, transparency, reversibility.
- Honestly: this is a real, finished internal curriculum source, not a bestseller — it has no public readership yet, and we don’t market it as one.
Two reference personas
What the system looks like on a real week
Two de-identified reference personas show how the same spine serves two very different children.
“W.”
In one sentence
A K12 child (US grade 4–12 curriculum band, typically upper-elementary to middle school) who gets stuck on math homework every day — and is too shy to raise a hand. What they need is a quiet, honest companion that is always there.
A day (and a week)
Stuck on a problem → snap one photo → RUDO reads it and coaches through a graded hint ladder, never just handing over the answer → the error is auto-archived into the error book, tagged across the five dimensions → it resurfaces for practice at the right time.
What the parent sees
One short bedtime email a day — deliberately one sentence, not a report: how many problems went wrong today, which knowledge point the child is stuck on, and what to practice tomorrow. Every problem is visible to the parent, with one-click export and one-click delete.
Which line they walk
RUDO Math Companion (currently an invite-only beta), with Pathfinder as the natural next step once gates start lighting up.
“L.”
In one sentence
A rising 8th-grader, about 13, of the frontier type — not remedial, not a prodigy: equations already solid, while geometry and probability sit right on the capability frontier (easy items stable, hard items drop).
A day (and a week)
One mainline: a 12-week AI project drawn from her own daily life, ending in a bilingual public project site, a complete evidence chain, a 3-minute English presentation, and a v1.0 portfolio publish. Each day runs the four cards — Foundation → Project → Evidence → Reflection — and a day only counts after the three reflection questions are answered. Math is bilingual bridging, not tutoring: her domestic Grade-8 curriculum mapped against Algebra 1; TOEFL-Junior practice lives inside daily Foundation, never as a separate course; AI literacy arrives as four intro lessons built on the AI4K12 big ideas.
What the parent sees
A read-only parent panel: a weekly slice, the evidence stream, a diagnostic card in honest plain language, and a monthly mentor review. Every step toward public exposure is parent-consented and revocable — and parents cannot fill in reflections or upload evidence for the child.
Which line they walk
The Global Bridge Track for the math/English bridging, with the project mainline running on the AI-literacy line’s mechanisms.
Go deeper