Inside the engine
How DynaMath works.
A transparent look at the proprietary IP that powers months-ahead struggle prediction and personalised intervention.
1. Curriculum intelligence
Every UK KS3 → A-Level maths concept is mapped with explicit prerequisites. This dependency graph is what most platforms lack and what makes prediction possible.
2. Predictive risk model
Mastery scores propagate through the graph. Weak prerequisites compound, so we can forecast which advanced concepts will fail months ahead of any classroom signal.
3. Adaptive nano-lessons
Short 3–5 minute lessons fix the highest-leverage gaps. A spaced-repetition rhythm keeps mastery from decaying.
4. Hybrid AI–human tutor
When risk crosses threshold and the AI tutor isn't lifting mastery, DynaMath routes the learner to a vetted human tutor. Outcomes flow back to retrain the model.
5. Mastery progression analytics
Real-time dashboards for students, parents and schools — all powered by the same predictive engine. Grade forecasts tighten as evidence accumulates.
The concept dependency graph
Below is the live graph for the Number & Algebra strand. Each node is a concept; each arrow is a prerequisite. Weakness propagates downstream — which is why a 35% mastery on Equivalent Fractions reliably predicts later struggles with Solving Linear Equations.
Mastered ≥75%Developing 50–74%At risk 30–49%Gap <30%Predicted future struggle
Demo predictive engine
This prototype uses a transparent rule-based version of the model so you can see exactly how risk is computed. The production version uses gradient-boosted trees and sequence models trained on tens of thousands of UK student trajectories.