Adaptive Learning Platform

Part of the worlds-biggest-software-project initiative.

An open, AI-native adaptive learning engine that personalises curriculum, schedules spaced review, and tracks mastery without locking content to a single publisher catalogue.

The Adaptive Learning Platform is a standards-based engine for K-12, higher education, and corporate training audiences who need real personalisation rather than linear courseware. It pairs a configurable knowledge graph with adaptive item selection, mastery-based progression, and AI tutoring to close the gap between cognitive-tutor research and what most institutions can actually deploy. The project targets the long-standing problem that effective adaptive learning today is either expensive, narrowly scoped, or trapped inside proprietary publisher ecosystems.

Why Adaptive Learning Platform?

Content lock-in dominates the market. Knewton, Smart Sparrow, and Cogbooks were each acquired by a major publisher (Wiley, Pearson, Cengage) and now only work with that publisher's catalogue, leaving independent educators and content creators without a serious adaptive option.
Best-in-class engines are subject-siloed. DreamBox covers K-8 math and early literacy; MATHia covers grades 6–12 math; neither addresses cross-subject transfer or secondary science, humanities, or workforce learning.
Pricing excludes budget-constrained buyers. K-12 SaaS pricing of $15–$50/student/yr and higher-ed pricing of $50–$200/seat/yr put rigorous adaptive learning out of reach for many districts and smaller institutions.
Spaced repetition tools stop at recall. Cerego and similar platforms apply memory-science scheduling but offer no branching paths, adaptive assessment, or knowledge-graph reasoning for procedural and conceptual learning.
Standards adoption is uneven. Although LTI 1.3, OneRoster, xAPI, cmi5, and the emerging IEEE P2247 reference architecture exist, no open implementation ties them together into a coherent adaptive engine.

Key Features

Mastery-Based Learning Core

Mastery-based progression with configurable proficiency thresholds per skill node
Standards-aligned knowledge graph (CCSS or custom learning objectives) with prerequisite mapping
Adaptive item selection engine that chooses the next question from a learner performance model
Adaptive re-assessment so learners can demonstrate improved mastery after remediation
Spaced-repetition review scheduling to counter knowledge decay on previously mastered content

Instructor and Institutional Tooling

Instructor dashboard with per-student mastery status, time-on-task, and at-risk flags
Predictive at-risk identification with automated teacher alerts before end-of-term assessments
Configurable knowledge graph editor for instructors to define content sequencing
Cohort-level analytics exportable to institutional data warehouses
White-label deployment option for institutional branding

AI-Augmented Learning Experience

AI-powered Socratic tutoring chat as a learning companion within adaptive sessions
Affective state signal detection (response latency, rapid guessing) with configurable intervention triggers
AI-generated micro-content (worked examples, practice items) for identified knowledge gaps where no authored content exists
Cross-subject transfer modelling to surface readiness signals across curriculum areas
Real-time knowledge graph refinement informed by aggregate learner response patterns

Compliance, Accessibility, and Integration

LTI 1.3 integration with LMS grade passback (Canvas, Moodle, Brightspace, Blackboard)
IMS OneRoster roster import from district SIS
xAPI and SCORM packaging for portable activity tracking and legacy LMS compatibility
FERPA- and COPPA-compliant data model with verifiable parental consent for under-13 users
WCAG 2.2 AA accessibility conformance target
Multi-language learner interface with community translation framework

AI-Native Advantage

Incumbent platforms rely on hand-curated knowledge graphs, one-dimensional spaced-repetition algorithms, and fixed content libraries. An AI-native engine can dynamically infer prerequisite relationships from response patterns across many learners, calibrate difficulty using multi-modal cognitive load signals (latency, error patterns, session timing), and generate targeted micro-lessons on demand when no remediation content exists. It can also detect frustration or disengagement from interaction signals and trigger interventions before a learner drops out, and model longitudinal mastery transfer between domains to enable genuinely cross-curricular adaptive paths.

Tech Stack & Deployment

The platform is intended to be deployable as self-hosted open source for institutions with data-sovereignty requirements and as managed cloud for districts and businesses that prefer a SaaS model. Integration is built around open education standards: LTI 1.3 for LMS launch and grade passback, IMS OneRoster for rostering, IMS QTI 3.0 for portable assessment items, xAPI and cmi5 for fine-grained activity tracking, and SCORM 1.2 / 2004 for legacy compatibility. The architecture aligns with the emerging IEEE P2247 reference for adaptive instructional systems. Adaptive content can be exported in IMS Common Cartridge format for portability.

Market Context

The global adaptive learning market was valued at approximately $5.13 billion in 2025 and is projected to grow at a CAGR of about 19.8% to roughly $12.66 billion by 2030 (MarketsandMarkets), with a competing SNS Insider estimate of $22.33 billion by 2032. Incumbent pricing ranges from free (Khan Academy) through $15–$50/student/yr for K-12 SaaS (DreamBox, MATHia) to $50–$200/seat/yr for higher-ed and corporate platforms (Area9, Realizeit, D2L). Primary buyers are K-12 curriculum directors, higher-education provosts and instructional designers, corporate L&D leaders, publishers embedding adaptive engines in digital courseware, and government education departments funding personalised-learning initiatives.

Project Status

This project is in the research and specification phase.
Contributions, feedback, and domain expertise are welcome.

Contributing

We welcome contributions from developers, domain experts, and potential users. See CONTRIBUTING.md for guidelines.

Important: All contributions must be your own original work or clearly attributed open-source material with a compatible licence. Copyright infringement and licence violations will not be tolerated and will result in immediate removal of the offending contribution. If you are unsure whether a piece of code, text, or other material is safe to contribute, open an issue and ask before submitting.

Licence

Licence to be determined. See discussion for context.