Human Intelligence, First.
Human Intelligence, First.
A measurement system built on Evidence-Centered Design — not a test, not a grade, not a black box.
Most assessment tools capture what a student answered. Logemics captures how they reasoned. The difference is architectural — built into every layer of the system, from task design to measurement model.
The foundation is a low-stakes measurement philosophy: no pressure, no ranking, no punitive grading. Students engage with reasoning scenarios rather than knowledge recall tests. The system observes patterns across multiple sessions — not a single performance on a single day.
Traditional assessment
One score
Logemics
Six independent dimensions
Traditional assessment
Single snapshot
Logemics
Longitudinal tracking
Traditional assessment
Knowledge recall
Logemics
Reasoning process
Traditional assessment
High-stakes pressure
Logemics
Low-stakes scenarios
Traditional assessment
Opaque AI grading
Logemics
Transparent, auditable rubrics
Traditional assessment
Certainty without evidence
Logemics
Explicit uncertainty bands
Every score Logemics produces is traceable back to a student response, through five explicit steps. Nothing is hidden.
Students engage with short interactive scenarios — evaluating an argument, identifying a logical flaw, interpreting conflicting evidence. These tasks are designed to isolate specific cognitive skills, not test subject-matter knowledge. A student who has never studied WWI can still reason about a primary source from that period.
For each response, the system captures three signals — not one:
These three signals together are more informative than any single answer.
The system uses AI strictly to extract features from student responses and apply predefined rubrics — evaluation grids designed by pedagogical experts. AI has no decision-making authority. Every inference is bounded by the expert-designed framework. Every score is auditable back to the evidence that produced it.
No black box. No opaque grading. Full transparency.
Rather than collapsing all evidence into a single score, the system generates an independent estimate for each of the 6 cognitive dimensions. Each dimension has its own trajectory. A student can be strong in logical coherence and still developing in metacognitive calibration — and both signals are visible and actionable.
The real value is not a single session result. It is the trajectory across sessions — weeks, months, a full school year. As more evidence accumulates, estimates become more precise and growth becomes visible.
The teacher dashboard shows each student's reasoning profile and flags where the gap between confidence and performance is widening. The system explicitly displays its level of certainty — uncertainty is shown, not hidden. Teachers retain full control and access subject-specific activity suggestions to act on what they see.
ECD is the gold standard in educational psychometrics — the framework used by the world's leading assessment bodies. It structures measurement as an explicit chain: from what we want to measure (claims), to the evidence we collect (tasks), to the inferences we draw (dimension estimates).
Every element of Logemics is designed within this chain. Nothing is added that cannot be justified. Nothing is inferred beyond what the evidence supports.