Interactive Demo

See what's hidden
inside your assessment data.

Real student data. Real PCA. Hover any dot to reveal the learner behind the score.

↓ Live PCA Plot ↓ Product Walkthrough ↓ The Gradient Difference

Every class. Every level. One diagnostic truth: scores don't tell the whole story.

Hover any student dot to reveal the diagnostic profile behind the score.

Coordinate Geometry — 27 Students × 13 Items

Hover any student dot to reveal their diagnostic profile — the information your gradebook has never shown you.

Awaiting selection

Variance Explained

Intervention Tier

High Performance
Developing
Needs Support
Intensive Support

Class Snapshot

27
Students
13
Items
72%
Class Avg
3
Dimensions

From materials to diagnostic clarity. In one teaching day. And then every day after. The loop at your disposal from August to June.

Explore Path E (Formative Loop) and Path B (Diagnostic Engine) — the two teacher tools that form the core Gradient workflow.

01
Send in Materials
30 sec
02
Check for Understanding Generated
Auto
03
Students Respond
5 min
04
Gradient Analyzes
Auto
05
Tomorrow's Plan — In Teacher's Language
Ready

Send in your lesson materials.

After class, share your lesson content with Grλdient. What you taught becomes the foundation for a targeted assessment built around your actual lesson.

30 seconds No lesson planning required Any subject

Path E — Lesson Intake
📋
Analyzing lesson content…
λ
Lesson content received
λ
Learning objectives identified
λ
Assessment being generated…

A targeted assessment, auto-generated from your lesson.

Gradient creates a targeted exit ticket based on what you actually taught — not a generic question bank. Items are targeted to your lesson objectives.

A Google Form is created instantly and sent to your class roster. Students see it on their devices before they leave the room.

Google Classroom sync Tiered difficulty Standards-aligned
Generated Assessment — Coordinate Distance
1
Find the distance between (2, 3) and (5, 7).
2
A segment has endpoints (−1, 4) and (3, −2). Find its length.
3
Point P is equidistant from A(0,0) and B(6,8). Describe the set of all such points.
Sent to 27 students via Google Classroom

Students respond in the last five minutes of class.

Students complete the assessment on their phones or Chromebooks as a seamless close to the period. Response rates are high because it's short, targeted, and directly connected to the day's lesson.

Responses flow into Gradient in real time. You don't touch a spreadsheet.

5 minutes Any device Auto-scored
Live Response Feed
Responses
23/27
Student_013/3Proficient
Student_042/3Developing
Student_111/3Foundational
Student_173/3Proficient
4 students pending…

Gradient analyzes response patterns automatically.

While you're wrapping up class, Gradient is detecting item-level patterns: which questions broke down, which students cluster together, where the class misconceptions live.

This is not a gradebook. It's a pattern detection engine applied to every response in every item position simultaneously.

Item analysis Misconception detection Group clustering
Item Analysis — Coordinate Distance
Q1
92%
Q2
63%
Q3
31%
⚠ Q3 breakdown: 14 students applied distance formula procedurally but failed to generalize to locus concept.

Tomorrow's teaching plan, waiting when you wake up.

By the time you check your phone that evening, Gradient has generated a specific, actionable plan for next class: which students need re-teaching, which concept to address, which students are ready to move forward.

Not a list of scores. Not "review with struggling students." A precise instructional sequence with the students named and the concept specified.

Actionable Student-specific Ready overnight
Tomorrow's Teaching Plan

Open with (8 min)

Warm-up: Locus definition — what does "equidistant" mean geometrically? Target: Student_07, 11, 14, 19, 22.

Small group (12 min)

Group A (proficient, 9 students): Extension — perpendicular bisector as locus proof.

Group B (developing, 11 students): Q2 reteach with signed coordinate scaffolding.

Group C (foundational, 7 students): Distance formula fluency with Tom.

01
Upload Assessment
2 min
02
PCA Runs
Auto
03
Dimensions Named
Auto
04
Student Profiles
Instant
05
Intervention Groups
Export

Upload your scored assessment CSV.

Export your scored assessment from Google Sheets, your SIS, or any gradebook tool. Gradient accepts any CSV where rows are students and columns are items.

Before any data reaches the analysis engine, student names are automatically replaced with anonymous tokens. FERPA compliance is architectural — not a policy, a code path.

Any gradebook export FERPA-compliant Student_01 anonymization
File Upload — Assessment Data
📊
coord_geo_unit3.csv
27 students · 13 items · Uploaded
Names replaced with Student_01…27
Item labels mapped to learning targets
Data validated — ready for PCA

NumPy SVD decomposes your response matrix.

Gradient applies Singular Value Decomposition to your item response matrix, extracting the principal components that explain the most variance in student performance.

This is the same mathematics behind every major psychometric instrument — now available to any classroom teacher in seconds, with no statistical training required.

SVD-based PCA No statistics knowledge needed Runs in seconds
PCA Engine — Running
Decomposing matrix…
λ₁
42%
λ₂
28%
λ₃
14%
84% of class performance variance explained by 3 dimensions

Dimensions are named from your actual data.

Gradient's AI interpreter reads the item loadings from each principal component and generates assessment-specific dimension names — based on which items load heavily, what those items measure, and what those items measure.

These are not predetermined labels. λ₁ might be "Conceptual Understanding" today and "Spatial Reasoning" in a geometry unit next month. The labels emerge from the data, every time.

Auto-interpreted Assessment-specific No predetermined categories
Dimension Interpreter — Output
λ₁ · 42% variance
Conceptual Distance Understanding
Items 4, 7, 11 load heavily — locus, generalization tasks
λ₂ · 28% variance
Procedural Formula Execution
Items 1, 2, 3, 6 load heavily — direct distance calculation
λ₃ · 14% variance
Coordinate System Navigation
Items 5, 9, 12 — signed number fluency, quadrant awareness

Every student gets a multidimensional profile.

Each student is placed in PCA space — not just ranked by score. Their position reveals their cognitive signature: strong on λ₁ (conceptual) but weak on λ₂ (procedural)? That's a completely different intervention than the student with the opposite profile.

Two students at 72% are often in completely different positions in this space. Gradient shows you exactly where each one is.

Individual positioning Cognitive profile Intervention routing
Student Profile — Student_07 vs Student_14
Both scored 72% — different learners
STUDENT_07
λ₁ Conceptual
+1.42
λ₂ Procedural
−0.89
Needs: procedural fluency
STUDENT_14
λ₁ Conceptual
−1.18
λ₂ Procedural
+1.05
Needs: conceptual grounding

Export intervention groups ready for Monday.

Gradient generates a five-tab XLSX report: class scatter, individual profiles, item analysis, intervention groupings, and parent communication messages — all pre-written and ready to send.

The intervention groupings are based on PCA position, not score rank. Students are grouped by what they actually need, not by how much they got right.

5-tab XLSX export Pre-written parent messages Intervention groups by PCA cluster
Export — Diagnostic Report
Tab 1: Class PCA Scatter
Tab 2: Individual Student Profiles
Tab 3: Item-Level Analysis
Tab 4: Intervention Groupings
Tab 5: Parent Communication Drafts
↓ Download Report

Same score. Completely different learners. Now you can tell them apart.

This is the core insight Gradient was built around. Two students at 72% are often invisible to each other in a traditional gradebook — and receive the same intervention that helps only one of them.

72%
Traditional Gradebook
A single composite number. No diagnostic information. No intervention guidance.
Student_0772%
Student_1472%
Student_2172%
"Review with students who scored below 75%."
λ₁ λ₂ λ₃
Gradient Diagnostic
Three latent dimensions. Three different learners. Three different interventions.
Student_07Needs: conceptual grounding
Student_14Needs: procedural fluency
Student_21Needs: notation/language
"Three students. Three paths. Zero guesswork."
λ   λ   λ

Ready to see your students this clearly?

Request a live demo with your own assessment data, or go straight to the toolkit.