Continuous Measurement × Coe Learning A Joint Product

The diagnostic layer
between curriculum
and what happens next.

Resolve takes any lesson from a high-quality mathematics curriculum and produces a complete diagnostic package — clarified lesson, research-derived assessment instrument, and typed intervention modules — from a single cognitive-diagnostic model that makes all three coherent with each other.

The core idea

A sequence repeats.
A cycle builds.

Most instructional cycles are sequences in name only. Lessons are taught, assessments are given, interventions are assigned — but the three steps share no common model of what students know or where they are. The assessment was not derived from the instruction. The intervention was not derived from the assessment. The result is motion without inference.

Without a shared model

Teach. Assess. Intervene.
Repeat.

Each step is performed in good faith. Each step is, in isolation, defensible. But without a shared cognitive architecture underneath them, the data the assessment generates cannot be traced to what was taught, and the intervention cannot be matched to what the assessment found. The cycle runs — and nothing accumulates.

With Resolve

The same model drives
all three steps.

Resolve builds each output from a single psychometrician-authored cognitive-diagnostic model for the standard being taught. The clarified lesson surfaces the model's cognitive architecture for the teacher. The assessment measures against it. The intervention addresses the specific failure it identifies. Each cycle produces a sharper picture of where each student is — and why.

What Resolve produces

Three outputs.
One architecture.

Every lesson Resolve processes produces the same three artifacts — each derived from the same cognitive-diagnostic model, each designed to hand off cleanly to the next.

Output 01
Lesson Clarification
Precondition for the cycle

Resolve applies the cognitive-diagnostic model to the existing HQIM lesson — surfacing the mathematical intent embedded in its structure, adding scaffolds at the points where students are most likely to lose access to the concept, and clarifying the instructional path without reconstructing the lesson. The curriculum mostly stays. What changes is its legibility — for the teacher preparing it, and for the students receiving it.

The lesson is now coherent — which means the assessment can measure what was actually taught.
Output 02
Diagnostic Assessment
Where the model does its deepest work

The exit ticket is replaced with an instrument derived from a research-defined misconception framework — documented cognitive error types per standard, each representing a coherent, internally consistent alternative model of the mathematics that students actually hold. Every wrong-answer option is a hypothesis about a specific student's cognitive model — not a plausible distractor, a research-defined diagnosis.

The diagnosis is now valid — which means the intervention can address what the assessment actually found.
Output 03
Typed Intervention
Matched to the error

Because the misconception framework identifies not just what the error is but what kind of error it is, the intervention is matched to the cognitive failure type — not just the topic. Each type has a distinct, evidence-based instructional template. Resolve generates targeted 20-minute modules automatically from the diagnostic result. Teachers don't choose the intervention. The model does.

Students arrive at the next lesson from a real picture of where they are — and the cycle begins again from a better place.

The full introduction covers the cognitive-diagnostic model in depth — the pipeline architecture, the assessment design principles, the intervention framework, and what it means for a cycle to actually build rather than repeat. If you're evaluating Resolve for a curriculum organization, district, or research partnership, start here.

Read the Full Introduction →
A note on downstream applications

The data architecture
was built to travel.

Resolve's cognitive-diagnostic model was designed with a specific architectural property: every student response it generates is labeled against a research-defined misconception framework — latent states defined before data is collected, not inferred from behavioral proxies afterward. That structure makes Resolve's output useful well beyond the immediate instructional cycle.

For ML & ed-tech partners
The labeled data problem,
solved at the source.

Most adaptive systems are trained on behavioral proxies where latent student states are inferred post-hoc. Resolve inverts this — misconception codes and cognitive step sequences define the latent states before a single data point is collected.

Read the technical brief →
Without Resolve With Resolve
Right/wrong sequences, time-on-task, clickstreams Response sequences labeled against psychometrician-defined misconception codes from the empirical research literature
Latent states inferred post-hoc from behavioral patterns Latent states defined in advance by a cognitive architecture — before a single data point is collected
Black-box mastery probability Interpretable state: which misconception is active, at which construct level, at which cognitive step
Training corpus assembled from proxies Theoretically-labeled corpus generated automatically — data collection begins when the pipeline runs
Full technical brief for ML and ed-tech partners →
Next Step

See what the cycle looks like from a real lesson.

The fastest path to understanding Resolve is reviewing a complete lesson package — clarified lesson, diagnostic assessment, and typed intervention modules — built from a psychometrician-authored cognitive-diagnostic model for a Grades 3–8 mathematics standard.

Request a Sample Lesson Package