Tests are the thermometer of your preparation. If you do few, you don't know what you know or what you ignore. If you do many but always the same closed PDFs, you reinforce wrong patterns and waste time. The solution that arrived in 2026 is AI-powered adaptive tests: new questions every time, adjusted to your real level, with explanation and connected to your spaced repetition.
In this guide you'll see how they work, what differentiates them from classic tests, when each is best and how to combine them to maximise learning.
What an adaptive test is and why it changes the rules#
A classic test is a fixed list of questions. You take them, get corrected, and if you redo them you get the same. Useful but limited: once you know them, the value drops.
An AI adaptive test works differently:
- Generates new questions every time on the topic you choose.
- Adjusts difficulty according to your performance: get them right, it climbs; miss, it drops a bit.
- Reinforces your weak points automatically: failed questions return in future tests until you master them.
- Explains the correct answer with BOE or other official source.
- Measures your real progress in each block, not by number of tests taken but by retention %.
This turns the test from a static exam into a personalised training tool.
How it works technically (without overwhelming you)#
Three AI pieces work simultaneously behind an adaptive test:
-
Question generation: AI takes the official syllabus (updated BOE) and creates multiple-choice questions with plausible distractors. It doesn't copy existing questions, it generates new ones each time. This avoids "I already know this answer by heart".
-
Difficulty calibration: each question has an estimated level. If you get several mid-level ones right, the next rises to hard. If you fail, it drops to easy. Same principle used by exams like GRE or GMAT.
-
Error memory (SM-2 or similar): each missed question enters a queue. It returns in the next test of the same topic, and again in the next — with growing intervals as you master it. The spaced repetition algorithm (explained here).
Adaptive vs classic tests: comparison#
| Aspect | Classic tests (PDFs) | Adaptive tests (AI) |
|---|---|---|
| Question variety | Limited to the PDF | Unlimited |
| Question repetition | Yes (when re-doing) | No, new each time |
| Level adaptation | No | Yes, automatic |
| Mistake reinforcement | Manual (note and redo) | Automatic |
| Correction with explanation | Only some | Yes, with official citation |
| Progress analysis | Manual | Automatic dashboard |
| Cost | Variable (books, packs) | 0-19€/month |
| For pure volume | Good | Good |
| For real retention | Limited | Excellent |
Honest verdict: classic tests are useful to get familiar with the format, but adaptive ones win from month two of preparation because they help you progress faster in less time.
When to use adaptive tests and when not#
When YES they're the best option#
- Day-to-day of preparation: 20-40 daily questions on the topic you're studying.
- Reinforcing weak blocks: when your dashboard shows poor results in an area, generate targeted tests.
- Short sessions: 15-20 minutes before sleep, on transport, on work breaks.
- To avoid memorising answers: if you do the same PDF 5 times, you end up memorising "answer C on question 12" without knowing why. With adaptive this doesn't happen.
When they aren't the only thing you need#
- To simulate the real exam: you need complete mock exams with timer and mix of all blocks. See mock exams guide.
- To get familiar with paper: on exam day many oposiciones are still on paper. Also practise with paper tests in the final weeks.
- For open practical cases: practical-case supuestos aren't closed tests, they require a different type of practice.
The 5 best practices for using adaptive tests#
1. Start easy, don't show off#
When you begin a new topic, let the difficulty start low. If you request hard tests before internalising the syllabus, you get frustrated and quit. Better: start with basic questions, gain confidence and let AI raise the level when ready.
2. Accept mistakes as information#
Each missed question is value. It tells you exactly where the hole is. Candidates who progress best are those who obsess over understanding why they failed, not how much they fail.
3. Don't do tests "for the sake of it"#
If you do 100 questions and then don't review any, you've wasted 1h. Reviewing mistakes is where learning happens, not in volume.
4. Alternate test types#
Each week mix:
- Topic tests (reinforce specific blocks).
- Cross-topic tests (block mix, simulates real exam).
- Spaced repetition tests (questions you've already missed).
This trains your brain to retrieve information in different contexts.
5. Measure retention, not volume#
The important metric isn't "I've done 5,000 questions". It's "I get 80% right on mocks of questions I've never seen". The second number predicts exam success; the first doesn't.
How many questions you should do#
A rough estimate for a 9-12 month oposición:
- Month 1: 200-400 questions (format familiarisation + initial topics).
- Months 2-4: 500-800 questions/month (covering all blocks).
- Months 5-7: 800-1,200/month (high volume, full syllabus in play).
- Last 2-3 months: 1,000-1,500/month (intensive + massive spaced repetition).
Total estimated: 6,000-10,000 questions in a full preparation. Sounds like a lot, but it's 22-37 questions/day distributed. Perfectly feasible if you make it routine.
Mistakes that ruin your tests#
1. Doing tests passively#
You read the question, mark the answer without thinking much, continue. This isn't practising, it's killing time. Each question must be an active retrieval effort.
2. Not reading explanations when you get it right#
If you got it right but it was "by intuition" or because other options sounded wrong, you got it right for the wrong reason. Read the explanation anyway: you might learn something you didn't know.
3. Comparing your score with others#
Forums full of people saying "I'm scoring 85%" without context. Maybe their tests are easier, maybe they use different tools. The only useful comparison is with yourself 2 weeks ago.
4. Doing all tests from the same block in a row#
If you only do Constitutional Law tests one week, then Administrative the next, etc., you don't train context-switching that you'll have in the exam. Mix blocks daily.
5. Not using negative tests#
Some cuerpos penalise mistakes in the exam (deductions of 0.25, 0.33...). If yours does, practise with that penalty enabled from the start. It teaches you when to risk and when to leave blank.
How adaptive tests look in Oposilab#
So you know what to expect:
- You pick a topic block (e.g. "Law 39/2015 - Administrative Procedure") and number of questions (20 by default).
- AI generates questions adapted to your current level, with calibrated difficulty.
- You answer. After each answer you can see the explanation or postpone to the end.
- When done: score, hits/misses, block retention %, most failed questions.
- Mistakes enter your spaced repetition queue automatically.
No configuration, no card maintenance, no hand-kept Excel. Just "generate test" → study → review.
Frequently asked questions#
Do adaptive tests replace the syllabus?#
No. The syllabus is the base, tests are training on that base. You need to read the syllabus first to understand concepts, then tests reinforce long-term memory.
Are adaptive test questions the same as the real exam?#
Not exactly — they're of the same format and difficulty, but generated in real time. For the real exam, official questions will be similar but not identical. That's good: if they were identical, you'd be memorising, not learning.
Can AI generate badly-formulated questions?#
A generalist AI (ChatGPT) can. A specialised AI like Oposilab is anchored to the official syllabus through RAG and filters, drastically reducing badly-formulated questions. Still, if you find one, you can report it for review.
How many adaptive tests in Oposilab's free plan?#
The free plan allows several daily tests, enough for an initial month of preparation. The Pro plan (19€/month) unlocks unlimited.
Better to do 100 questions in a row or 25 each day?#
25 each day, no doubt. Time distribution (the spacing effect) consolidates memory much better than marathons. Plus your head performs much better in short, consistent sessions.
How do I know I'm improving if questions change each time?#
By two metrics the dashboard shows:
- Hit % on cross-topic tests (block mix). Should rise over weeks.
- Retention % on spaced repetition (you get right what you'd previously failed). If it rises, you're consolidating.
Start with one test#
Don't wait until you have "the syllabus ready" to do tests. Study a topic, do its test, review mistakes. That loop applied for 6-9 months separates the candidate who passes from the one who falls short.
To go deeper, read the complete AI preparation guide or check the list of best mock exams to combine with your tests.
