37 simulations. Zero output to fake.

A tutoring model without measurement is a black box. A measurement engine without a tutor is a test. QLM closes the loop: measure, teach, verify, improve. The assessment IS the learning and the learning IS the assessment — across 7 cognitive dimensions, on every interaction. Including 10 dedicated math modes.

Simulations

326+

Missions

124

NGSS Standards

Cognitive Dimensions

3D + 2D

Rendering

What Students Experience

Real simulations, not videos.

Predict-Observe-Explain — every mission follows the inquiry cycle that NGSS was built on
Adaptive difficulty — K-2 students get discovery missions; AP students get research challenges
Digital lab notebooks — built-in Claim-Evidence-Reasoning (CER) artifacts, exportable as PDF
Real-time annotations — scientific formulas and measurements displayed on the simulation canvas
Cross-discipline links — the system knows statistics is prerequisite to genetic drift
3D + 2D modes — Three.js 3D rendering with PBR materials; 2D fallback for older devices
Works everywhere — browser-based, no installs; Chromebooks, iPads, desktops

What Teachers Get

Standards reports, not grading.

Assign any simulation — students are running experiments in 60 seconds, no prep required
NGSS standards dashboard — see which students demonstrated which performance expectations
Grade export (CSV) — download mastery data for your gradebook
Standards coverage heatmap — rows = students, columns = NGSS standards, cells = mastery level
Skill radar per student — 7 cognitive dimensions measured continuously, not just at test windows
District dashboard — multi-school rollup with engagement metrics and at-risk flags
Clever SSO — single sign-on with your existing student accounts

New in TeacherOS

Three capabilities that turn cognitive data into teacher action — no AI prompt engineering required.

"What Should I Teach Tomorrow?"

Analyzes the class cognitive heatmap. Identifies the weakest dimension. Recommends 3 simulation modes targeting that gap — with an ASSIGN button. Lesson planning from data, not prompts.

Parent Report Generator

One-click plain-language report per student: D1-D7 levels, percentile, strengths, growth areas, specific home activities. Printable PDF. No consultant needed.

Standards Gap Analysis

Real NGSS + CCSS codes mapped to all 37 modes. Coverage heatmap shows which standards are at risk. Lesson suggestions per gap. Grade band filtering: K-2 through 9-12.

◆ Continuous Cognitive Measurement

Every simulation session measures 7 cognitive dimensions. Every interaction feeds the adaptive engine. The profile sharpens continuously — not once per semester.

D1

Analytical

D2

Quantitative

D3

Verbal

D4

Spatial

D5

Inference

D6

Collaboration

D7

Operational

Research shows AI tools without cognitive measurement cause "joint cognitive erosion" (67-study PRISMA review, 2026). QLM is the measurement layer.

What You Get — Real Platform

Not mockups. Live screenshots.

TeacherOS · Students Tab

5 students with XP, streaks, levels, engagement flags (Active/Slowing/Disengaged). Sort by recency, XP, streak, sessions. Search and filter.

Standards Browser · 93% NGSS Coverage

111 of 120 NGSS standards covered. Filter by grade band (K-2, 3-5, 6-8, 9-12) and discipline (Physical, Life, Earth/Space, Engineering). Each standard linked to simulation modes.

Student View · Ecosystem Simulation

Real student session in the spacecraft cockpit. Population dynamics, predator-prey, food web lenses. Every interaction scored across D1-D7. AI tutor available.

Mode Briefing · Force Field

Mission briefing with objectives, prediction prompts, and vocabulary. Students read the physics context, make a prediction, then run the simulation to test it.

How we compare to what you're using today.

Capability	QLM Crucible	Open-Source Sims	VR-Based Sims	Textbook Labs
Adapts per student	Yes — mission-level	No	Limited	No
Measures mastery (not completion)	Yes — per skill	No	Partial	No
NGSS standards report per student	124 PEs mapped	No	Manual	No
Cross-discipline prerequisites	37 modes linked	No	No	No
Built-in lab notebooks (CER)	Yes	No	No	Paper
K-2 through AP in one platform	326 missions	Partial	6-12 only	No
Works on Chromebooks	Yes	Yes	No (VR)	N/A
Cost	$3–5/student/yr	Free	$15–25/student	$50+ textbook

Designed for Durable Learning

Stanford's SCALE Initiative found that AI tools improve student grades while students have access to them, but gains disappear on independent assessment. QLM Crucible was designed from the ground up to protect the productive struggle that causes durable learning:

◆No AI generates answers — simulations ARE the first attempt. Students predict, observe, explain.

◆Mastery, not completion — system tracks durable understanding, not checkbox progress.

◆Memory decay detection — re-engages students before knowledge fades, preventing cognitive debt.

◆CER lab notebooks — forces metacognition through Claim-Evidence-Reasoning artifacts.

Full research position: quantumlearningmachines.com/research/cognitive-debt

Grounded in Research

Three research findings that shaped QLM's design:

◆ 67-study PRISMA review (2026): AI tools without cognitive measurement cause "joint cognitive erosion" — both creativity and critical thinking decline. QLM measures 7 cognitive dimensions continuously.

◆ SciVQR benchmark (CASIA, 2026): Even the best AI models score 20% lower on visual-spatial science questions than text-based ones. Simulation-based assessment is structurally harder to game.

◆ UC Berkeley A-grade surge (WSJ, 2026): A grades surged post-ChatGPT in homework-heavy courses. QLM's simulation-based assessment has no output to polish — the performance IS the measurement.

Research & Measurement

Every simulation IS the assessment. The system knows what students understand because it watches them think — not because it grades their output.

◆Productive struggle is measured, not assumed — hint independence rate, transfer success rate, and misconception detection are tracked per student, per session. The engine knows the difference between productive struggle and unproductive frustration.

◆Adaptive engine benchmarked against peer-reviewed methods — QLM’s adaptive algorithm is competitive with state-of-the-art item selection methods on established educational benchmarks (ASSISTments, EdNet, Junyi Academy).

◆10 dedicated math modes — from place value and fractions (K–2) through algebraic manipulation and geometric proofs (high school). Each mode detects specific math misconceptions: fraction-as-two-numbers, forgot-to-distribute, more-means-add.

◆Designed for interoperability — QLM’s measurement framework is designed to be interoperable with national tutoring research initiatives. We welcome partnerships with universities and research organizations to validate learning outcomes.

QLM launched April 2026. Actively seeking district partners to validate outcomes.

37 simulations. Zero output to fake.

Real simulations, not videos.

Standards reports, not grading.

Not mockups. Live screenshots.

How we compare to what you're using today.