The short answer

Russian is a listening-and-speaking problem long before it is a grammar problem. Stress is unwritten and mobile, and because unstressed vowels reduce, putting the stress in the wrong place distorts the entire word rather than just the emphasis. That is not something a transcript can show you, and it is exactly what a tutor that hears your audio can. Cyrillic is shown with a Latin reading underneath by default, so a beginner is never locked out.

Russian has a reputation for being hard that is only half deserved. The alphabet takes a weekend. The cases take longer but they are systematic. The genuinely difficult part is that spoken Russian and written Russian relate to each other through a set of rules nobody writes down.

Stress is the whole problem

Russian stress is not marked, it is not predictable, it moves between the forms of a word, and it occasionally distinguishes two different words.

That would be a manageable memorisation problem on its own. What makes it structural is that unstressed vowels reduce heavily, so the stress determines which sounds you actually produce. Put it in the wrong place and you have not merely emphasised the wrong syllable, you have changed most of the vowels in the word.

This is why learners who can read Russian comfortably often cannot say a sentence aloud without it coming out unrecognisable, and why the fix has to happen in speech rather than on paper.

What hearing you actually buys

Your audio goes to the model directly rather than being transcribed into text first, so pronunciation, hesitation and pace are part of what the tutor responds to.

Stress errors and softness errors are precisely the class of thing a transcript destroys. Say молоко with the stress on the first syllable and a transcriber will give the tutor молоко, correctly spelled, with no indication that you mangled it. Say a hard л where a soft ль belongs and the text is identical. The tutor never finds out, so it never helps.

Cyrillic, and the reading underneath

The tutor writes in Cyrillic, because learning to read it is part of learning the language and it is genuinely quick.

A Latin reading of your own speech is shown underneath by default. Somebody who can already read the alphabet turns it off in a moment; somebody who cannot has no way to discover the setting exists, so it starts on and gets out of the way when you no longer need it.

What a session is

Unscripted conversation with something you can interrupt, backed by an authored CEFR syllabus of 271 items from A1.1 to C1, of which 96 are capabilities: concrete things you can do rather than grammar points. What you keep getting wrong shapes later sessions with nothing to configure.

A note on cases, since it is the thing people dread: the tutor does not stop to mark a wrong ending. It keeps the conversation going and feeds the right form back inside its own reply, the way a person would. Being frozen mid-sentence computing an instrumental is the failure mode to avoid, and it is a habit built by being corrected too eagerly.

Travel Mode

If you are going somewhere Russian is spoken, the app changes shape for the trip: rehearsal before you leave, one real-world mission a day while you are there set in the city you are in, and a debrief with your tutor afterwards about how the exchange actually went.

The five Russian tutors

Every language has five, each from a different place and each sounding like it. You can switch whenever you like, and the app remembers where you are regardless of who you talk to.

AnastasiaMoscow
The clear, standard Russian of the capital, with crisp consonants and balanced vowel reduction.
DmitriSaint Petersburg
Standard Russian with the slightly softer, more even delivery the city is known for.
SvetlanaYekaterinburg
Clear, neutral Russian with the easy, unfussy warmth of the Urals.
MaratKazan
Clear standard Russian with the calm, even cadence of the multicultural Volga capital.
IgorNovosibirsk
Clean, neutral Siberian Russian with a grounded, no-nonsense directness.

What English speakers usually find hard

These are drawn from the pronunciation patterns the tutor listens for in Russian. Because it hears your voice rather than a transcript of it, they are things it can actually notice and work on with you.

Stress is unwritten, mobile, and it changes the vowels: Nothing on the page tells you where the stress falls, it moves between forms of the same word, and it occasionally distinguishes two words entirely. Because unstressed vowels reduce heavily, getting it wrong does not just misplace the emphasis, it changes which sounds come out of your mouth.
Unstressed o becomes a, and then becomes nothing much: Молоко is malako, not moloko. Reduction is graded: the syllable just before the stress keeps a fairly full a, and the ones further away collapse toward a schwa. Хорошо runs schwa, a, then a full stressed o.
Almost every consonant has a soft twin: Softness is a simultaneous raising of the tongue toward the palate, not a separate y sound after the consonant, and it is phonemic. English speakers either ignore it or insert an audible y, and both are immediately wrong. The soft sign and the vowels я е ё ю и are what trigger it.
The ы vowel, which English has nothing like: High, central, unrounded, lips spread, tongue pulled back. Мыло and мило are different words. It only follows hard consonants, which is a useful cue, and English speakers usually substitute the vowel of bit.
Clusters get articulated, not broken up: Здравствуйте is real, and inserting a helping vowel to survive it is the most common English-speaker strategy. Also worth knowing: что is shto and конечно is kanyeshna, a small frozen set where ч is pronounced sh.