Build AI-native lessons

Faculty design the lesson, not a prompt box

You decide what the AI knows, how hard it pushes, and what counts as done. The Level is bounded, instrumented, and authored by the educator from the first block.

Product, Levels

Faculty design the Level, not a prompt box.

A Mizain lesson or "Level" is the atomic unit of learning progression: Faculty-authored, bounded, and instrumented. Faculty build a governed environment for one concept, in a workspace called the Level Composer.

The anatomy of a Level

Intro Content Activity Checkpoint Activity Testimonial Sign-off

Blocks in lilac are instrumented: every action inside them becomes a measurement event.

Multiple activity types

  • AI debate: the Student argues a position, the AI attacks the thesis
  • Reading, with a hint ladder the Faculty member wrote behind it
  • Lecture video with pause and answer checkpoints inside the video
  • AI-tool project challenge, with every exchange logged
  • Game-template challenges in 2D and 3D, populated by Faculty

Level 04, Thermodynamics

Faculty-authored objective

Student must defend entropy's directionality against three AI-posed counterexamples before advancing, then explain the result in their own words.

Why it holds up

Three properties a skeptic can test.

01

Deterministic

Every number recomputes from the raw event log. If the recomputation does not match, the record fails. A summary is frozen at the version that produced it, so history is never silently re-scored.

02

Transparent

Every AI exchange is captured with the model identifier and the system-prompt hash. Assistance transparency travels with every artifact, so no reader can mistake assisted output for unaided capability.

03

Attested

A named Faculty member reviews the summary and signs it. No consequential decision is algorithm-only, which is also the mitigation UAE data protection law requires.

The chain

Event log Evidence Summary Faculty attestation Signed receipt

We do not claim to have proven learning. We claim a chain of evidence that another person can inspect, recompute, and challenge.

AI debate

A sparring partner, not a co-writer

You set the motion, the sides, the turn limit, the LAM stance, the refusal boundaries, and the corpus in scope. The Student takes the stage turn by turn, and the full transcript stays with the Level.

The hint ladder works the same way. Three levels of help. Each one gives more. The last one still leaves the work to the Student.

A product interface render of the AI debate stage, showing the motion, the turn by turn exchange, and the running transcript.

Pedagogical sovereignty

Pedagogical sovereignty over AI, Level authoring with a measurement loop, and evidence of what works. The AI is a sparring partner for the Student, and it answers to you.

Design your first Level with us

Bring a course you already teach. We will walk through what a Level looks like inside it.