Start learning

For parents and educators

Using Toorova in a classroom

You don't need to be an AI specialist to teach this well. You need a sequence you trust, a way to see who's stuck, and a plan for the fact that every answer is a search away.

Setting up, in about ten minutes

  1. Create a class. It generates a join code.
  2. Give students the code. They see a screen naming the class and you, then join. Regenerating the code kills every older copy of it immediately.
  3. Restrict it if you need to. A class can be limited to your school's email domains, and students are told that before they tap Join. You can also close enrollment or archive the class.
  4. Assign work. Directly, or by building a learning track — an ordered sequence of lessons you write once and reuse with any class. Assigning takes a snapshot, so editing the track later never moves the syllabus under a class that's partway through it.

Step-by-step detail, and how sponsored access for schools works, is in the help center.

Three patterns that work

Flipped: lesson at home, argument in class

Assign the lesson as homework and spend the period on the part software can't do — the disagreement. Open with a question the lesson left contestable and let them fight about it. Because the course puts the question before the teaching, students arrive having already committed to a position, which starts a discussion far better than a chapter does.

Stations: the dashboard as a grouping tool

Group by who has passed what. Students who passed explain it to those who didn't, while you take the group that's stuck. Explaining is the hardest test of understanding there is, so the helpers are getting the toughest exercise in the room.

The prediction ladder: fifteen minutes, no prep

Put a situation on the board. Everyone privately writes a prediction. Show hands, then reveal the answer. Spend the rest of the time on the wrong predictions, finding the belief underneath each one rather than correcting it.

How the fifteen minutes divides: two minutes for everyone to write a guess, one minute for a show of hands, two minutes to reveal the answer, and ten minutes arguing about the wrong ones. Fifteen minutes, no prep Everyone writes a guess 2 min Show of hands 1 min Reveal the answer 2 min Argue about the wrong ones 10 min
The reveal is the short part. The lesson is in the ten minutes spent on why a wrong answer looked right.

It works in any subject, and it's the fastest way to find out that half the class is carrying a misconception you didn't know existed.

Assessing when the answer is a search away

Recall questions have become typing tests. Assess the things a model can't hand over:

  • Prediction. "What will this system do with an input it has never seen, and why?" Needs a mechanism in their head.
  • Error-finding. Give them a confident, plausible, wrong answer and ask them to find the mistake. It's the most useful exercise in the subject, and it's what the job actually is.
  • Explanation without jargon. Terminology is compression, and compression hides gaps. Ban the vocabulary and the gaps show up immediately.
  • "Why not the simpler thing?" Understanding a method means understanding what breaks without it.
  • The process, not the artifact. Drafts, revisions, and what they tried and abandoned. Harder to outsource than a final answer, and more informative.
A rule of thumb

If a chatbot could produce a full-mark answer to your question in ten seconds, the question isn't measuring your students anymore. It's measuring their typing.

What the dashboard tells you

You can see activity and how consistent it is, XP, level, streak, badges, daily goal, which lessons each student has finished and passed, and the titles of what they've been working on. It answers one question: is this student learning, and are they stuck?

You can't see individual answers, which option a student picked, private journal reflections, or AI tutor conversations. That line is enforced in our database rather than on the screen, for the same reason the method works at all: a student who thinks every wrong answer is reported to their teacher stops guessing. You lose item-level analytics and keep students who answer honestly.

A bar chart with one pair of bars per student, showing lessons finished beside lessons passed. Most pairs are nearly equal. Two students have finished the most lessons of anyone and passed the fewest, and those are the two to talk to. finished, not passed finished passed
One pair of bars per student. The useful signal isn't the XP, it's the gap between finished and passed.

A student with many lessons finished and few passed is sitting through the course without it landing, and that's the conversation to have.

Practical notes

  • Ages. Toorova is for 13 and up, and the AI features are never available to an under-13 account. For younger students an approved educator can invite a parent to link: the parent still has to receive the email, sign in and accept, so the consent belongs to an adult throughout.
  • Access for your students. Schools can apply for sponsored Premium; once approved, a class draws capped, dated seats from the license. Get in touch to start.
  • Bilingual classrooms. The whole course exists in Spanish at parity with English, written rather than machine-translated, and two students in the same class can work through the same lesson in different languages.
  • No AI background needed. Every technical term is defined in plain language before it's used to explain anything else. If you can run the prediction ladder above, you can teach this.

Next