How we create
our material.
Every note, question, flashcard and mock exam is drafted with a fine-tuned and law-specific AI model, reviewed three times by people, and monitored closely once it’s live. Here’s how it works, and how we keep it accurate as the law changes.
Anyone can generate an MCQ.
Trusting one is different.
It’s easy to get a chatbot to write a practice question. It’s much harder to write one you can trust your career to.
The law turns on small distinctions. A void contract never existed; a voidable one stands until someone sets it aside. On the SQE, several answers are often legally possible, and only one is the best. A question that gets that nuance slightly wrong can teach you the wrong instinct, and you may not find out until exam day.
- 9people built it
- 18months to develop
- 3human reviews for every piece
It took a team of nine 18 months to build the process below.
Made by people,
AI and data.
Smart Study Notes, MCQs, flashcards and mocks all go through the same five steps. AI helps us draft and check at scale. People decide what’s good enough. Then data from real learners tells us what we missed.
- Step 1Start with the source
The SRA’s SQE1 Assessment Specification and notes from top-performing candidates.
- Step 2AI expands it
In-house tooling drafts notes, MCQs, flashcards and mocks, using several models, not one.
- Step 3AI checks it
A separate multi-model evaluation layer tests every draft before a person sees it.
- Step 4Three human reviews123
Below the quality bar, it’s reworked or deleted. Above it, the next reviewer takes a look.
- Step 5Live, and still checked
Learner flags and performance data send anything suspect back for review.
Start with the source
We begin with the SRA’s SQE1 Assessment Specification, which sets out what the exam can test, alongside study notes from candidates who passed at the top of their cohort. Everything else is built on that foundation.
AI expands it
Our in-house AI tooling turns that starting point into full notes, questions, flashcards and mocks. It combines several models and techniques rather than relying on any single one.
AI checks it
A separate evaluation layer, also using several models, checks every draft against our quality bar before a person sees it.
Three rounds of human review
A reviewer then checks it by hand. If it falls below our quality threshold, it’s deleted or sent back for another round of AI revision. If it passes, a second reviewer checks it, then a third. Our reviewers are recent exam passers, tutors and legal experts.
Live, and still checked
Once it’s in the app, we keep checking it, using your feedback and how it performs.
Nothing goes live
on one opinion.
Three different people review every piece before it reaches you. At each stage, anything below our quality threshold stops there. It’s either deleted, or reworked by our AI tooling and sent back through review from the start.
Above the quality bar?
A second, separate check.
And a third.
Live doesn’t
mean finished.
Two things keep watching our material once it’s in your hands.
You. Flag anything that looks inaccurate, unclear or not explained in enough depth, and it goes back for review.
The data. We track how every question performs. If students with a strong Prep Score get a question wrong far more often than we’d expect, that usually points to a problem with the question, not the student. So it goes back for review too.
Spot something off? Tell us in the app.
AccuracyStyleDepthWrong answers, adjusted for Prep Score.
When the law changes,
so does Kinnu.
When new legislation comes in, a key case is decided or the SRA updates its specification, the material it affects goes back to step one. It’s redrafted, re-checked and reviewed three times, exactly like new material, before the update reaches you.
New legislation, a key case or an SRA spec update.
Affected material starts again from the updated law.
The same multi-model drafting and checks.
Then the update goes live in the app.
What it means for
each study tool.
MCQs
Questions that feel like the real exam.
Drafted and checked by our multi-model AI tooling, starting from the SRA specification and top candidates’ notes.
Then reviewed three times by recent exam passers, tutors and legal experts before it goes live.
Length and difficulty are calibrated to what candidates are likely to see on the actual exam, with an explanation for every answer option.
Tracked once live. If well-prepared students get a question wrong unusually often, we review it.
Smart Study Notes
Concise notes for every topic in FLK1 and FLK2.
Built from the SRA specification and top candidates’ notes, then reviewed three times. Concise and exam-calibrated in difficulty and coverage.
Interactive, with Lara alongside to unpack a tricky concept the moment it trips you up.
Flashcards
Pre-written cards that come back at the right time.
Build your own stack from 4,000+ pre-written flashcards.
Our spaced repetition algorithms calculate how often each card should come back, so you revisit what you’re about to forget.
Every card goes through the same three rounds of human review as our notes and MCQs.
Mocks
Exam day, before exam day.
Four mock exams, taken in strict exam conditions.
See your scaled score compared to your real cohort, then decide what needs your attention next.
Every mock question goes through the same three rounds of human review.
Lara AI
AI you can actually trust.
An SQE-specific AI model, directly trained on UK legislation, court procedure, SRA guidelines and top SQE revision notes.
Actively updated to reflect changes in law and regulation.
Answers are designed for the SQE context, with the level of detail and focus the exam calls for.
See it for
yourself.
Start my 7-day free trialNo credit card required · Then £48/month
Back to the toolkit