The Adaptive Learning System

What the dev team actually builds: the runtime (Maestro Journey) that takes a finished course and runs it for students — rendering, adaptive delivery, the Evidence Check flow, milestone submission, AI-assisted grading, and the SME and learner surfaces around them.

The 30-second version

The runtime consumes a Published Course Package and runs it against each learner. It renders the nodes and every format in them, uses per-node DNA to decide what comes next, holds a learner at a topic until mastery is shown, lets them set their own pace, and runs and grades assessments with AI. The litmus test: anything that exists because a specific learner is interacting is Journey's job.

Rendering the authored course

The runtime shows the nodes and every format inside them — video, text, interactive — exported from Studio, consuming the produced learning objects in their common envelope and respecting the learner-visibility flags. It never reaches back into Studio's internals; it works entirely from the package.

Adaptive, mastery-based delivery

The runtime uses the per-node DNA (prerequisites, mastery statements, linked assessments, likely trip-ups) to decide what comes next. It holds a learner at a topic until they have shown mastery and routes them back to learn more if they have not; it lets the learner set their own pace, moving on as they demonstrate progress. The full routing logic is in Adaptivity & Routing.

The Evidence Check runtime

At the heart of delivery is the Evidence Check runtime. For each node it:

Renders the formative Evidence Check object (text, structured-response interactive, or artifact fragment).
Captures the four signals — response, reasoning, confidence, process.
Invokes the AI Diagnostic Matcher through the-code's trust-layer component.
Records the evidence record, updates the learner model, and drives routing.

For engineers — three assessment systems, not one

Build the Evidence Check flow, the milestone artifact submission flow, and the SME grading interface as separate systems. They serve different purposes and produce different outputs: Evidence Checks produce mastery-relevant diagnoses (no scores); milestones produce SME-graded artifacts. Do not merge them into one "assessment engine."

Milestone submission flow

Each summative milestone renders a Milestone Guide (assessment brief, rubric, checklist, examples, AI-use rules, and live readiness status). The runtime evaluates the readiness gate against the learner model, handles submission unlock, receives and stores the summative artifact, and captures AI-use disclosure and process evidence throughout production — decision log, revision history, acknowledgements. The trail is the evidence, not just the endpoint.

AI-assisted SME grading

Summative artifacts are SME-graded against the rubric with AI assistance. The grading interface runs an AI evaluation against the rubric, presents per-criterion recommendations with reasoning, accepts SME adjustments and overrides, and captures the final grade with SME sign-off. It also surfaces the formative behavioural summary alongside the artifact so the SME has the integrity context. The final grade reflects artifact quality — it does not re-derive mastery (mastery → grade).

SME dashboards & review queues

The SME (in their delivery role) needs surfaces to monitor and act: cohort monitoring, individual learner status, submission and grading queues, integrity-flag surfacing, and a review queue for provisional or flagged formative diagnoses. These are release-one requirements, not later polish — an instructor must be able to run a live course from them.

The end-to-end learner experience

For the learner, the runtime is a single coherent experience: a student dashboard, node navigation, the Companion present throughout, progress against milestones, and a personal learning record. On first entry, onboarding is minimal — the learner lands on the dashboard, the Companion introduces itself and orients them to the journey (milestones, how the adaptive flow works, how it supports them), and they start the first node. The intent is that the learner always has guidance alongside them, with no gaps.

Pathway recommendation

Maestro Journey also carries the AI-powered pathway recommendation engine. Before and during a course it helps learners discover and design a pathway based on their level, history, goals, gaps, and the institution's rules — answering "where should I start?", "which certificate fits my goal?", "what should I study next?". Its strength depends on a well-structured academic library (programs → courses → certificates → badges → nodes → prerequisites), which is exactly what Studio produces. For release one the goal is a basic course-level recommendation capability; the fuller engine matures in later phases.

Academic marketplace

The Academic Support Marketplace gives learners governed access to human support — approved professors, tutors, mentors, or experts — for paid sessions, with the university managing quality assurance, payment, and revenue share. It balances the self-paced model: AI for scale, human experts for depth, university governance for trust. Like recommendation, it is part of Journey's longer arc rather than the core January build.