How it works — and why it works.

You are the editor, not the audience. You’re not here to be handed perfect questions — you’re here to direct a competent-but-imperfect writer until the set is exactly what the SAT would throw at you.

Bad problems are the feature, not the bug. The model on purpose is a modest one. It hands you a rough draft: some questions are sharp, some are too easy, some are subtly broken or ambiguous. Spotting which is which — and prompting the bad ones away — is exactly the skill that moves a score.

Your goal is a perfect set. Chasing it — hunting down the buggy-looking items, tightening the distractors, fixing the difficulty — is what forces you to actually understand the test.

1. 1

   Commission a set

   Choose the SAT content domains you want, say how many problems, and (this is the good part) drop in a few real sample questions — paste them, or attach a screenshot or PDF. The writer imitates your samples’ style and difficulty. Finding good samples is itself part of the studying.

2. 2

   Read it like an editor

   The draft won’t be perfect. Good. Read every question and ask: is this the right difficulty? Are the wrong answers tempting, or obviously wrong? Is anything ambiguous or broken? You’re grading the problems, not answering them.

3. 3

   Fix the weak ones

   Each question gets its own page. Rate it — 🙁 😐 🙂 — and when one isn’t good, say what’s off (“the distractors are too obvious”) and hit Revise this one. The writer fixes that questionin place — keeping what works and changing only what you flagged — instead of throwing it away for a random new one. Deciding what’s wrong is the studying.

4. 4

   Reshape the whole set

   For set-wide changes — a formatting bug, or the overall mix (“fewer linear equations, more systems”) — use the Revise the whole set note. Each whole-set revision is a new version (v1, v2, v3…) you can flip back through to see how it evolved.

5. 5

   Use it

   Reveal answers and explanations, print or copy the set, or send it to a student as a link — they take it on their own device and you see exactly what they got right and wrong, broken down by domain. On Math, they get the same Desmos calculator as the real digital SAT.

This isn’t a gimmick — it’s built on how people actually learn. Four things are happening while you edit:

• You make it, so you own it.

  We remember what we build far better than what we’re handed. Producing and repairing a question demands deeper processing than passively answering one.

• Judging a problem forces you to know the standard.

  To decide a question is “too easy” or “a broken distractor,” you have to hold the real SAT’s standard in your head and measure against it. That act of judgment is where the understanding lives.

• Steering coverage exposes the test's real shape.

  When you say “fewer linear equations, more systems,” you’re confronting how the test is actually weighted — and, usually, where you’re weakest. The negotiation teaches the terrain.

• The friction is the point.

  Editing is harder than clicking “next.” That productive struggle — the desirable difficulty — is exactly what makes the learning stick past test day.

These questions run through Claude Haiku — a small, fast, inexpensive model — on purpose. A top-tier model would hand you near-perfect questions, and then there’d be nothing to edit. The editing is the studying, so we need a writer that leaves real work on the table.

A modest writer gives you a genuine rough draft: mostly serviceable, with a few items that are too easy, subtly off, or quietly wrong. Catching those is the whole exercise. (It’s also cheap and fast, so you can iterate as many times as you want.)

Stated plainly: it will sometimes be wrong — a distractor that’s actually correct, an off difficulty, phrasing the real test would never use. That isn’t a bug to report. It’s the raw material you’re here to fix.

SocratiQ is a tool for active review and targeted drilling, not first exposure. Reach for it when:

• you’ve studied a topic and want to pressure-test it,
• you want to drill a weak area (the “focus here” domains are built for exactly this),
• you want to force yourself to say why an answer is right or wrong,
• or a tutor wants to spin up a quick, targeted set in session.

It is not the right tool for learning a subject cold, and it shouldn’t be your only source of practice.

• Who gets the most from it

  Students who already know the test well enough to judge a question — motivated, active learners, and anyone working with a tutor. If you can look at an item and tell it’s off, the loop works for you.

• Who should be careful

  A true beginner who can’t yet tell a good question from a bad one. If you can’t judge the draft, the editing loop never fires — and you risk absorbing the model’s mistakes as if they were real. Newer or younger students should use it alongside a teacher or tutor.

You can just generate a set and take it — but you’d be leaving the whole point on the table, and taking on some risk: you’d be practicing on unvetted, AI-written questions. The value of this tool scales directly with how much you engage as the editor. If you use it passively, keep it to light drilling, always anchored on real samples — never a stand-in for official practice.

This is an imitation engine, not an authority. Every question it writes is a copy of the samples you feed it plus its best guess at the test’s style. That has real limits you need to respect:

• It can produce questions with wrong answers, ambiguous wording, or off-standard content — and if you aren’t anchored on the real thing, you may not catch them.
• It can drift from the test’s actual difficulty, format, and topic weighting.
• It has no official standing — it is not the College Board, ETS, or any exam board, and its questions are not real exam questions.

The rule is simple: keep official materials as your ground truth. Use real, released practice tests and question banks as your anchor; use SocratiQ to extend and drill around them. Paste real samples every time — the closer your anchor, the closer the imitation.

Used as a supplement to real prep, it’s powerful. Used as a standalone replacement for official practice, it’s a liability. It’s a study aid, not a substitute — and definitely not the exam.

Commission

Your order for a set of problems — the domains, count, format, and samples you ask for.

Editor / editorial note

You. And the plain-English direction you give the writer — one question at a time (Revise this one) or across the set (Revise the whole set). This is the studying.

Version (v1, v2…)

A snapshot after each whole-set revision. Flip between them anytime.

Rate & revise

Give each question a 🙁 😐 🙂 verdict. A low rating opens a box to say what’s off, and the writer revises that one question in place to fix it.

Reveal

Show a question’s answer and explanation.

Copy as text

Copy the whole set — questions, answers, and explanations — as plain text you can paste into a Google Doc, email, or notes app.

Send to a student

Get a link you hand to a student. They take the set on their own device; you get back a results link with every answer.

SPR

Student-produced response — a type-in answer with no choices. SAT Math only.

Domains

The SAT’s 8 official content areas, each with its real exam weight shown.