Each activity leaves evidence. Reading the trajectory is under development.
A correct answer is no longer enough proof of learning. Each activity leaves traceable evidence of what the student did on their own: what they understood, where they were uncertain, what they revised and how much support they needed. Accumulating that evidence over time — reading the trajectory — is the direction this project is building.
A simulation of the mechanism, not real data: three unconnected activities become connected over time; the same error reappears in two moments; the help needed decreases; a review appears on the student's own initiative; and the line crosses from one context to another. In the end, technology may change: the accumulated evidence remains.
Grades show results. The trajectory makes evolution observable.
Each activity leaves a deliverable, but deliverables are not connected to one another. The institution knows what was answered at each point, not how the student changed between one point and the next.
Generative AI makes that gap urgent: when an answer looks correct, the result does not reveal how much of the reasoning is the student's and how much was assumed from the tool.
A grade records a result. A trajectory provides evidence of how reasoning changes.
One activity produces a point of evidence
Before the tool, during the contrast, when transferring: every moment leaves a record of what the student was able to do on their own.
Each activity teaches. Time will let us read the evolution.
-
Trace
The student begins with their own analysis, before consulting the tool. That first reasoning is kept as a record and a reference.
-
Contrast
The subsequent conversation puts hypotheses, variables, and decisions to the test. The student explains what they accept, what they revise, and what they still cannot affirm.
-
Transfer
With a new problem, the tool remains available but no longer guides. We observe whether the approach appears on the student's own initiative.
-
What the student can explain, justify, review, and transfer contributes evidence about their understanding.
Time turns isolated moments into signals of evolution.
Every moment is recorded. The goal is to link them and, over the years, distinguish what is consolidating from what is not. That longitudinal reading is under development, not operational.
Conceptual example — illustrative, not real results
- March Needs a lot of support
- April The same error reappears
- June Needs less help
- August Revises their reasoning independently
- November Transfers their reasoning to a new problem
An assessment observes each moment. A trajectory reveals the change between them.
One week → one month → one semester → one year
What schools begin to observe
Signals about real learning, beyond grades, without turning the classroom into surveillance.
- Comprehension & reasoning
-
- what they understand on their own, without support from the tool
- how they structure causes, variables, and conditions
- Doubt & error
-
- the limit of what they can affirm
- the errors that reappear between activities
- Support & revision
-
- how much help they need to move forward
- whether they correct and review on their own
- Transfer & evolution
-
- whether the approach appears with new problems
- whether it consolidates over time
CognitiveTraceability does not seek to monitor students or judge them as people. It records pedagogically relevant evidence of what they do during activities designed to observe learning.
The evidence is reviewable, its interpretation rests on an explicit methodological framework, and decisions remain with teachers and institutions.
The trajectory informs. The institution decides.
From recorded evidence to a pedagogical decision, the path is deliberate.
-
Recorded evidence
what the student did on their own, preserved and linked to its source
-
Interpretation
a shared methodological framework turns it into observations
-
Longitudinal pattern
observations accumulate into signals about evolution
-
Human pedagogical decision
teachers and school teams decide what to do: support, reinforce, talk
The platform does not decide for the institution: the decision is always pedagogical and human.
Methodological grounding
The framework observes four dimensions of reasoning: causal complexity · technical specificity · epistemic awareness · decision under uncertainty. It is developed within the research line at Universidad de Santiago de Chile; its methodological validation is ongoing.
Operating today. The full trajectory, in development.
The state, without adornment: what already works and what is still being built.
Working school journey
The student signs in with a course code, works first, contrasts with the socratic tutor, and the activity is recorded. The journey is verified with demo data and the first school in onboarding.
Longitudinal MVP and validation in development
Per-student patterns and trajectories, and the methodological validation of the framework, are still being built. The academic research line advances separately and is never mixed with school data.
Can your institution show how a student's reasoning evolves over a year?
Let's talk about connecting every activity into a longitudinal learning trajectory — without mistaking a correct answer for demonstrated learning.