Skip to content
Maestro.
MaestroMaestro Overview — contents

The Concept

Adaptivity & Routing

What actually drives the adaptivity — a small, reviewed routing table rather than something open-ended. The result of the lesson's check, not a score, decides what is offered next, and the learner makes the choice.

The 30-second version

After every check, one routing table decides which options to offer and which one to mark Recommended. The learner chooses under "Where to next · you choose". Ideas that need another look gather on the course's Revision page; one comes back as a short "Welcome back" before the first lesson of a new day; and before each milestone, the required items need one honest try — passing isn't needed.

Participatory adaptivity & agency

The governing frame is transparent, by-choice adaptivity: read the result → say it plainly → offer options → the learner chooses. The system marks one option as Recommended, shown first with a deep border, and offers the rest beside it. The learner's choice is recorded as an agency signal — evidence of engagement, not noise. Because the recommendation is always visible, agency never becomes abandonment.

The routing table

The routing table is data, reviewed like content, and it is the only place the "what do we offer" rule lives. The routing decision is not pass/fail — each result opens a different set of options:

ResultOptions offeredRecommendedLeast to move on
SolidContinueContinueAlready met.
Nearly thereContinue · firm it up with one exampleContinueNothing extra; Next lesson is free.
Not yetSee it another way · walk through a worked example · continueSee it another wayNothing extra; Next lesson is free.
Not sureSee it another way · walk through a worked example · continueSee it another wayNothing extra, and no penalty for the honesty.
A mix-upSee it another way · rewatch the part · practise again · show your thinking againSee it another wayOne try at any option before Next lesson.

On Nearly there, Not yet and Not sure, a fresh question joins the options once the learner has taken one of the support steps; on a mix-up it is offered from the start. On Solid it is never offered — the idea is already met.

The options

OptionWhat it opens
See it another wayA second way in to the part the check tested, produced in Studio for that part. On a mix-up it is marked "made for this mix-up" only when it targets that misconception.
Firm it up with one exampleThe lesson's own firm-up prompt; the learner writes one example in their own words. It is kept, never marked.
Walk through a worked exampleA click-to-reveal walk through one approved variant of the check: the story, the right verdict, why (from the rubric), how a sound justification reads — then "Try one yourself".
Rewatch the part · Practise againThe lesson itself, opened at the tested part's video or reading, or at its Try it. No new content is needed.
Show your thinking againA fresh question: an unseen variant of the same check. At most two per lesson, enforced on the server.
ContinueThe next lesson.

Every option draws on pieces Studio produced and reviewed for that lesson — the mix-up message, the second way in, the firm-up prompt, the worked example and the look-again probe. Publishing refuses a lesson missing any required piece, so no learner page can arrive without its options.

A mix-up: one try first

When a result matches a named misconception, the card shows its approved label and the mix-up message written for it: what the learner said, why it misses, and the idea seen correctly. Then "Next lesson" appears after one try at any option — see it another way, rewatch, practise again, or a fresh question. Passing isn't needed; until then the page says so plainly. Leaving by the back link or the course page stays possible; only "Next lesson" waits. Every other result keeps Next lesson free.

For engineers — the table, the budget and the follow-ups

The routing table (engine adaptivity module) maps each result to offered moves, conditional moves, one recommendation, a least-to-advance rule, a follow-up schedule and a belief-write policy; Not sure defers the belief write. The fresh-question budget is two per lesson; it never refills and carries across result changes, and a worse re-reading is held as suspected rather than confirmed. The node view exposes next_available with a reason, so the mix-up rule is enforced server-side. Not yet, Not sure and a mix-up schedule a next-day look-back; a mix-up also schedules a later "Look again" item, and anything still open after its look-back waits on Revision.

The Revision page

Ideas that need another look don't vanish; they gather on one Revision page per course, open any time. It is a course tab (with a count) and also a Revision hub in the main menu that lists the learner's courses. Every item says where it came from — "From lesson 3 · … · open the lesson" — and the link opens that lesson. Items come in four kinds:

ItemWhat the learner does
Look again at your answerReads their own earlier answer, quoted as written, and answers one question: "What would you change in this answer, and why?"
Try the lesson's checkOpens the lesson straight at a check they skipped, then returns here.
One exampleWrites one example in their own words. Kept, never marked.
A fresh questionAnswers an unseen variant of the lesson's check.

Feedback says right or not yet in words — never a score, never a result name — and offers the lesson when it helps. "Short on time?" offers 5 minutes, 10 minutes or Everything: the server picks the items that fit, required first, then mix-ups, then the oldest; the rest stay for next time.

Before Milestone N

When a milestone is open or comes next, its required items show first on the Revision page as "Before Milestone N". The rule is deliberately light:

  • Each required item needs one honest try before the guide's Start assessment unlocks. Passing isn't needed.
  • At most five rows show at once, required first; "+N more after these" appears when there are more, and the next row appears as each is tried.
  • Skip moves to the next row but doesn't count as a try.
  • The Milestone Guide is readable at any time. Until the tries are done, its button reads "Start opens after Revision · N left" and links back here.

Every way out of the page goes to the guide, never straight into the assessment. The milestone itself is visible on the course road from the start; only Start waits.

"Welcome back"

Before the first lesson of a new day, when something is due, the learner meets one short look-back: one item, about a minute, never marked. Any honest answer moves them on to their lesson. It is required, but narrow:

  • Only starting the next lesson waits for it. It never blocks the course page, the guide, results or Revision.
  • Each idea gets the look-back once; if it is still open afterwards, it waits on the Revision page.
  • It never appears on the same day the item opened.
  • There is no deferral and no ignore counting.

Personalised assessment context

For milestone work, the rubric and standard stay identical for all learners, but each learner chooses a world — a context for the brief, with its own situation, audience and files. Same rigour, different context. This is an adaptive mechanism and an integrity mechanism at once, because a shared answer no longer fits another learner's world (see Lessons, Checks & Milestones).

The mechanisms, together

The routing table after every check · Studio-produced support for each result · the mix-up's one-try rule · bounded fresh questions · the Revision page with "Short on time?" · the required "Welcome back" look-back · "Before Milestone N" with one honest try · Atlas's quiet reminders (see Atlas) · personalised milestone worlds. Built incrementally, obvious cases first, with no point where the learner is left without a next step.