Homework – Fractions sheet 2
AssignmentPhotographed handwritten work
Name (name masked)Student A
[1]Q4: 1/3 + 1/4 = 2/7
[2]Q5: 2/5 + 1/2 = 3/7
Launching soonBy ClassHeart
Know what every student truly understands.
Teachers add the evidence they already collect: assignments, assessments and their own observations. Socrates drafts a diagnosis for each student, citing the work it came from. Nothing becomes a conclusion until a teacher accepts it.
Draft diagnosis Synthetic example
DraftAdds numerators and denominators separately instead of finding a common denominator[1][2]
The loop
One objective, followed from first evidence to a recorded outcome.
Every cycle starts from one curriculum objective: what the student should be able to do.
Teachers add the evidence they already collect: assignments, assessments and their own observations, including photographed handwritten work.
Socrates drafts a diagnosis from that student’s own evidence, naming a weakness or misconception and citing the passages it rests on.
The teacher accepts, edits and accepts, or rejects the draft; only an approved diagnosis becomes a conclusion.
An intervention names a teaching strategy, such as a worked example or retrieval practice, together with a plan for reassessment.
An accepted intervention can generate a targeted activity aimed at the specific weakness.
The student is reassessed against a baseline captured beforehand, and the teacher confirms, overrides or rejects the result.
The outcome is recorded as improved, unchanged, deteriorated or uncertain, without claiming the intervention caused it.
Then back to 01 for the next objective
Recorded as learning history
Every step is recorded in the student’s learning history: the Student Intelligence Card. It is append-only, so new entries are added and nothing is overwritten.
The four promises
The four promises Socrates makes about every piece of a teacher’s evidence.
Only teacher-approved diagnoses become conclusions. AI output is always a draft, and drafts and rejected suggestions never change a student’s record.
Each draft is accepted, edited and accepted, or rejected. A teacher’s edit is stored alongside the original draft.
Each approved claim links to a short, redacted citation and its source record.
No diagnosis reaches teacher review without at least one valid citation, and every citation is checked against its source.
Student names are masked before evidence is analysed by AI. Raw evidence and prompts stay out of operational logs.
Names from the class list become a stable pseudonym; other personal details become [REDACTED].
Class-level priorities, plus year and weekly plans, sit alongside each student’s history.
Class Intelligence shows who needs attention, which reassessments are due and where misconceptions cluster, built only from approved data.
Walkthrough
Evidence first, then the draft, then the decision and what follows from it.
Synthetic example: no real students, names or schools.
Step 1 · Objective
Step 2 · Evidence first
The cited passages come first, so the teacher reads the work before the draft.
Homework – Fractions sheet 2
AssignmentPhotographed handwritten work
Name (name masked)Student A
[1]Q4: 1/3 + 1/4 = 2/7
[2]Q5: 2/5 + 1/2 = 3/7
Typed note
Teacher observation[3]Student A explained Q4 as “one plus one on top, three plus four on the bottom”.
The student’s name is replaced with a stable pseudonym before any AI analysis or staff display.
Step 3 · The draft diagnosis
Written by AI, marked as a draft, and tied to the evidence by citation markers.
Fractions: unlike denominators
DraftConfidence is the model’s own estimate, shown as High, Medium or Low.
Step 4 · The teacher’s decision
Three choices, and only the first two change the record.
Demo: choose a decision
What each decision does
The diagnosis becomes the teacher’s conclusion and joins the Student Intelligence Card with its citations. Attainment does not change until a teacher confirms a reassessment.
The teacher’s wording becomes the conclusion, and the AI original is stored separately alongside it. Attainment does not change until a teacher confirms a reassessment.
The student’s learning record does not change. The draft stays in the audit trail.
Student Intelligence Card Preview
Student A Year 8 Maths, Set 2
Audit trail
1 draft awaiting review
Draft accepted by teacher
Draft edited and accepted; AI original kept
Draft rejected and kept here, outside the record
Step 5 · The follow-through
Intervention
Model finding a common denominator with fraction bars, then 6 guided items. Includes a reassessment plan.
Accepted by teacher
Targeted activity
8 items on adding fractions with unlike denominators.
Reassessment
Measured against the baseline captured before the intervention.
Before1 of 6
After5 of 6
Outcome: improvedConfirmed by teacher
“Improved” describes the result, not its cause. The teacher could also have overridden or rejected the reassessment.
Step 6
Student A Year 8 Maths, Set 2
Append-only. Each entry links back to its redacted citation and source record. In Class Intelligence, the reassessment is no longer due, and the approved misconception can join a class cluster.
Privacy
Masked names, short citations, separate schools.
Before any AI analysis or staff display
Class list name(name masked)Student A
Other personal details(detail masked)[REDACTED]
Names are masked before evidence is analysed by AI. Names from the class list become a stable pseudonym and other personal details become [REDACTED], after text extraction and before any AI analysis or staff display.
Raw evidence and prompts stay out of operational logs.
Citations are short and redacted. A conclusion points to the passage it rests on, not to a whole document.
Each school’s data is isolated. Teachers see only their assigned classes.
A diagnosis draws only on that student’s own evidence. Another student’s history is never retrieved.
Student work is treated as data, never as instructions.
School admins stay in control. They can export data, delete a student’s records and set retention periods. Deleted data may remain in backups until those backups expire.
No surveillance. No camera, audio, biometric or emotion monitoring.
FAQ
Something else? Write to contact-us@classheart.com.
No. Socrates does not grade or mark students, and it does not handle discipline or parent messages. It drafts diagnoses from the evidence teachers add, and a teacher decides what becomes a conclusion.
School staff: teachers, academic reviewers and school admins. There are no student or parent accounts.
The student’s learning record does not change. The rejected draft stays in the audit trail, so there is a record of what was suggested and decided.
Student names from the class list are replaced with a stable pseudonym, and other personal details with [REDACTED]. This happens after text is extracted from the evidence and before any AI analysis or staff display.
Yes. Each approved claim links to a short, redacted citation and its source record. A diagnosis cannot reach teacher review without at least one valid citation.
No. Year and weekly plans sit alongside each student’s history and stay drafts until a teacher or school admin approves them. AI is not involved in creating plans.
Assignments, assessments and teacher observations, as PDF, Word, plain text, spreadsheets or photos, including photos of handwritten work. Socrates works in English.
Socrates is launching soon. To request early access, email contact-us@classheart.com.
By ClassHeart
Socrates is the first product from ClassHeart, a company that creates tools to empower teachers and students to get the best results. ClassHeart is an Archon Studios company.
Launching soon
Tell us about your school and how you would like to use Socrates.
or write to contact-us@classheart.com