Why Jaimy speaks first, and only first

There is a familiar shape to a language-learning regret. People describe it the same way, in any country, in any age group: "I studied this in school for years and I still can't say anything." The grammar is in there somewhere. The vocabulary is in there. The certificates are framed on someone's wall. And yet, in front of a real person who actually speaks the language, the words refuse to come.

This isn't a moral failing. It's a perfectly predictable outcome of how the language was practiced. If you spend a decade reading and writing and almost never speaking out loud, you end up extremely good at reading and writing, and not very good at speaking. Speaking is its own skill. It does not arrive as a bonus track on the album of "knowing a language."

The thing that builds the thing

There's a finding in second-language research that sounds almost too simple to be useful. It's called the output hypothesis, and the punchline is this: input — listening, reading, watching — is not enough. To get good at producing language, you have to be pushed to produce it.

When you speak, three things happen that simply do not happen when you listen.

You notice your gaps. You go to say a thing, and the word isn't there. You feel the shape of where it should be. That awareness is what makes you absorb the word the next time you hear it.

You test a hypothesis. Every sentence you produce is a tiny experiment: I think this is how you say it. Then you find out. The feedback loop is how the language gets corrected and consolidated in your head.

You think about how the language works. The moment you have to assemble a sentence, you're choosing tenses, picking word order, deciding which little particle goes where. Reading doesn't make you do that. Speaking does.

Researchers who recorded learners during speaking tasks found roughly ten of these micro-learning moments per session — moments that simply do not show up in passive comprehension. Speaking is not just practice for fluency. Speaking is what builds it.

So why doesn't everyone just do it?

Because it's terrifying.

Speaking a language you don't fully command, in front of another human, is one of the most anxiety-spiking situations adults willingly put themselves in. Your accent feels wrong in your mouth. The words come out slow. You can hear yourself making mistakes, and you imagine the other person hearing them too. Most people, given the choice, find a way to study without speaking. They watch a series with subtitles. They do another grammar exercise. They tell themselves they're not ready yet.

The result is the regret at the top of this post.

The whole product, in one sentence

Jaimy exists to remove every reason you might have to avoid speaking.

There is no typing in Jaimy. No flashcards. No multiple-choice. There is a face on the screen, a microphone, and a conversation. You open the app, and within seconds you are speaking the language. You stay speaking it for fifteen minutes. Then you go about your day.

It sounds simple. It is the entire pitch. Every other decision in the product — how patient the tutor is, how it corrects you, how it picks topics, how it remembers what you said last week — is downstream of the choice to make speaking the only thing you do.

What "speaking first" actually changes

When you commit to speaking-first, a few things follow.

Lessons get shorter. You can't have a productive thirty-minute spoken session every day; your voice tires, your attention drifts. Fifteen minutes is closer to the right number. Which, conveniently, is also the dose the spacing research recommends.

The AI has to be patient. A speaking-first product cannot rush you. If the silence stretches, the AI waits. If you need three tries to get the word out, the AI takes three tries with you, warmly.

Correction has to be gentle. The fastest way to make a learner stop talking is to flood them with red marks. Jaimy's correction style is built around keeping you in conversation, not pulling you out of it for a grammar lesson. (We'll write more about that one another time.)

The affective filter — the technical term for "anxiety blocks acquisition" — becomes the central design problem, not an afterthought. The AI tutor's biggest advantage over a human teacher is not its grammar. It is that no one is judging you. Lean into that hard.

What we're not pretending

We're not pretending that speaking is the only thing that matters in language learning. Reading is wonderful. Watching films in the language is wonderful. Travelling and being thrown into the deep end is the best teacher there is.

What we are saying is this: most learners already have those other tools. What they are missing — what they have been missing the whole time — is a low-stakes way to actually open their mouth every day. Not when they feel ready. Not when they find a willing native speaker. Today, on the walk to lunch, with a tutor who has nothing better to do than listen to them try.

That's Jaimy. That's the whole idea. The rest of the product is built around protecting it.