Multiple choice
Single correct option with distractors that actually test understanding, not luck.
AI question generation
Any topic. Any type. Twenty questions in sixty seconds.
Stop sourcing question banks. Generate the exact questions you need — for any technology, any difficulty, any depth — review them, regenerate the weak ones, and ship.
6 question types 10 coding languages Quality scored Human approves
Batch 1 of 45 / 5
Batch 2 of 45 / 5
Batch 3 of 43 / 5
Batch 4 of 4queued
Q · 13 · MCQ · Medium
Which of the following correctly describes the event loop's relationship with the microtask queue?
A · Microtasks run before macrotasks in the same tick B · Microtasks run only after I/O C · Microtasks block rendering until queue empty correct D · Microtasks run on a separate thread
Six types, one engine
Every question type your exam blueprint needs.
Multiple response
More than one correct option — for nuance that single-answer questions miss.
True / false
Quick recall checks with explanations attached, so a wrong answer becomes a teaching moment.
Fill in the blank
Tests precise vocabulary, syntax, and naming — answers auto-matched with synonyms allowed.
Short answer
Free-text answers with AI-assisted first-pass grading, then a human review.
Coding challenge
Full coding problems with starter code, function metadata, and hidden test cases — in 10 languages.
The review flow
AI generates. You approve. Only what you trust enters your bank.
No autopilot. Every question is yours to keep, regenerate, or discard before it's ever shown to a candidate.
Tell it what you need
Topic, subtopic, difficulty, marks, negative marking, hints, explanations. One form. One click.
Watch it generate in batches
Parallel batches of 5 questions, multiple concurrent. Twenty questions land in about a minute. Quality-scored on the way in.
Review every question
Keep, discard, or regenerate any single question without touching the others. AI works, you decide.
Save to your bank — only what passes
Approved questions land in your question bank with full metadata, ready for any exam, pool, or template.
Why the output is usable
It's not a single prompt. It's a pipeline.
Type-aware validation
Every generated question runs through type-specific checks — correct distractor count, valid test cases, parseable code — before it reaches your screen.
Quality score per question
A 0–100 score tells you which questions to trust, which to regenerate, and which to refine before saving.
Per-question regenerate
Don't like one out of twenty? Regenerate it alone, with the same context, while keeping the rest.
Top-up if short
If only 17 of 20 pass validation, the system tops up to 20 automatically — no silent shortfalls.
Provider-agnostic
Gemini 2.5 Flash or GPT-4o-mini, switchable at runtime. Your content, not theirs.
Transparent token billing
Pay only for what you save. Wallet shows reserved vs spent — no surprise bills, no subscription lock-in.
AI-generated coding challenges
Not just MCQs. Real coding problems — with hidden tests.
Generate full coding challenges in ten languages: starter code, sample tests for the candidate to run, and hidden tests for scoring. No more sourcing problems from a third-party library.
See the coding engine → Function metadata Name, params, return type
Starter code Per language, idiomatic
Sample test cases Visible to the candidate
Hidden test cases Used for scoring only
Edge case coverage Generated, not hand-coded
Languages supported 10 (JS, Py, Java, C++, C, C#, Go, Rust, TS, PHP)
Stop hand-writing questions
Bring a topic. We'll generate the bank live.
Pick any technology or topic. We'll generate twenty questions on the call, walk you through the review screen, and save the keepers to your bank. Eight minutes, real product.