The zone where learning happens

You have had this experience in two flavours.

In one, you sat through a class that was completely beyond you. The teacher was speaking, and the words were words, but somewhere around the third sentence your brain quietly closed the shop and went home. You stared politely at the front of the room for forty minutes. Nothing went in.

In the other, you sat through a class that was completely beneath you. You knew the material. You did the exercises with your eyes half closed. Your phone was tempting. By the end you were a little annoyed at the time you'd given up. Nothing new went in there either.

These two failures share a common cause. Learning didn't happen because the difficulty was wrong.

A Russian developmental psychologist had a name for this

His name was Lev Vygotsky. The idea was the Zone of Proximal Development, which is a forbidding name for a very practical concept.

Vygotsky observed that for any learner, there are three zones of difficulty:

Things you can already do alone. Practicing in this zone is comfortable but doesn't grow you. You consolidate, you maintain, you don't advance.

Things you cannot do, even with help. Pushing into this zone is exhausting and demoralising. Nothing sticks because you have no scaffolding to attach the new material to.

Things you can almost do — that you can manage with help. This is the zone of proximal development. It is where learning actually lives. The task is just past your reach; the help is what makes the reach possible; the next time, you do it slightly closer to alone.

Eighty-seven studies, when later researchers synthesised the field, found near-uniform positive effects when teaching was deliberately calibrated to this zone. It is one of the most robust findings in education.

The trick — the genuinely hard part — is that the zone keeps moving. As you grow, what was just-past-reach becomes within reach, and what was beyond becomes just-past. A static lesson plan is, by definition, in the wrong zone for most learners for most of the time.

The classroom problem

A human teacher with twenty-five students in a room cannot stay in everyone's zone at once. The lesson is pitched at the middle of the class. For a third of the students, it's too easy and they tune out. For another third, it's too hard and they tune out. For the lucky third in the middle, it's roughly right and they learn.

This is not anybody's fault. It is a structural limit of one-to-many teaching.

The one-to-one tutor, historically the answer, is rare and expensive. Centuries of educational research come back to the same observation: the lone learner with a patient expert is a vastly more efficient learning machine than any classroom. We just couldn't afford it for most people.

What a tutor in your pocket can finally do

A speaking-first AI has, for the first time, a real chance to do this for everyone.

Jaimy is constantly listening for difficulty signals. Not in a creepy surveillance way — in the same way a good human tutor watches their student's face. How long are your pauses? Are you reaching for the easy words or risking the harder ones? Are your sentences getting shorter as the session goes on? Are you using the structures we've been working on, or backing away from them?

Each of those is a small signal about whether the current difficulty is right.

When the signals say too easy, Jaimy ratchets up. Slightly more complex prompts. Slightly less time waited before the next turn. Slightly more open-ended questions instead of closed yes-or-no ones. The bar moves.

When the signals say too hard, Jaimy quietly scaffolds. A simpler way of asking the same question. A choice between two options instead of an open prompt. The phrase you're reaching for, offered as a stepping stone. The bar comes down, just enough.

The aim, every session, is to keep you in that narrow band where you are stretching but not breaking.

Scaffolding that fades

The other half of the ZPD idea — the half that gets ignored in most discussions of it — is that scaffolding has to fade. The point of help is to make the unaided behaviour possible later, not to make the help permanent.

A classic example: a beginner says "I went to the cinema and watch a film." The tutor recasts: "ah, you went to the cinema and watched a film — what film was it?" The beginner now has the corrected form in their ear. Two sessions later, the beginner says the same thing again, slightly closer to right. Two sessions after that, the beginner says it right. The scaffold has done its job and is no longer needed.

This is why Jaimy is not the same tutor at session five as it was at session one. Help that was given freely at the start gets withdrawn as you grow into the structure. A learner who is past needing the scaffold doesn't get it — partly because they don't need it, and partly because giving it would prevent the very growth we're aiming for.

The opposite of a course

A course, by its nature, is the same for everyone who takes it. The first lesson is the first lesson. The exercises are the exercises. The pacing is the pacing. The instructor's job is to teach the course.

Jaimy is not a course. There is no syllabus that you and the person next to you are both moving through at the same rate. There are materials — chunks, scenarios, grammar points, vocabulary themes — but the order and emphasis are constructed for you, in real time, based on where your zone is today.

This is the part of building Jaimy we are most proud of, and most cautious about. Getting the zone right is hard. We won't always get it right. But the goal — the only goal that matters — is to keep you in the place where every session moves you forward a little. Not so far that you bounce off. Not so little that you doze through.

A patient, calibrated, slightly-too-hard conversation, every day, is what fluency is made of.