The seven-layer defense stack
Rather than policing cheating, Maestro is engineered so honest evidence is the rational choice — and so faked evidence is expensive, self-defeating, and impossible to sustain.
The problem we design against
Any on-screen task whose complete answer is visible can be outsourced to an AI tool in seconds. This matters twice over: it corrupts grades, and — more deeply — it poisons adaptation. A faked check writes a false “secure” belief into the learner model, so the system routes the learner past the very support they needed.
The strategy
We deliberately do not build the system around catching cheaters. Lockdown browsers and AI-detection tools are an arms race current research shows is being lost. Instead, we design so that real mastery is the only thing cheaper to have than to fake: honest evidence is made easy and rational, dishonest evidence expensive and self-defeating, and evidence is only believed when it stays coherent over time.
Seven reinforcing layers
Layers 1–4 are built into the content before any learner sees it. Layers 5–7 run while the learner works.
1 · Incentive design
Checks aren't graded directly and give no in-flow feedback, so outsourcing one buys nothing.
2 · Task design
Every check demands something no outside tool has: the learner's own reasoning, confidence, or history.
3 · Per-learner variation
Each check is drawn from a pre-generated, SME-reviewed variant pool, so answers don't travel between learners.
4 · Locked, kept work
Milestone work autosaves with its full version history, and is locked before it is defended.
5 · Behavioural signals
How an answer was produced only ever weights evidence and triggers support — never accusations.
6 · Longitudinal triangulation
Mastery is believed only when independent witnesses agree across time.
7 · Spoken defense
Every milestone ends with a live spoken defense of the learner's own locked work.
Signals support, never accuse
The same signal that might suggest outsourcing also describes an anxious learner — and the response to both is the same supportive check-in.