Free · runs on your machine · you bring an API key
You don't need 300 more problems. You need to know what breaks.
A voice tutor for coding, system design and behavioral interviews. You talk out loud. He writes down what you said, tells you the one thing that went wrong, and plans tomorrow around it.A voice tutor for coding, system design and behavioral interviews. You talk. He finds the one thing that went wrong, and plans tomorrow around it.
Four minutes from here to your first session. One command to install, one to start. Create an account — only if this machine is not only yours.
No card, no trial clock
Your voice never leaves your machine
About $0.09 a session, on your own API key
Session 89 · with Theowrap-up
3. Longest Substring Without Repeating Characters · 41 minLongest Substring · 41 min
Diagnosis
You named the invariant only after the second hint.
“so like… we move left until, um, there’s no repeat?”
your words · 12:04 · replayyour words · 12:04
3 sessions in a rowsame slip, three different problems
Next session starts here
Before you write a line, say the invariant out loud — the one thing that stays true at every step. Theo will not take your code until you do.Say the invariant — what stays true at every step — before you write a line.
Drill this tomorrowcloses after 3 clean solves
Computed by rules over the session journal — not written by a model.
What you will never see here:What you will never see:“You are 78% ready”
The part that hurts
You already did the hours. That was never the problem.
Two engineers who did everything the internet told them to do. Nothing they used could tell them what exactly was wrong.
“the feedback primarily boiled down to a very repetitive you need to go deeper”
Blind · on a $100 mock with a Google L7
“I knew the pieces of what makes a system but didn't have the intuition to put it all together”
Blind · failed Amazon L5 after Grokking and Alex Xu
Volume is not the missing ingredient — a diagnosis is.Volume is not the missing ingredient.
Every product on this market sells you another run at another problem. None of them remembers the last one.Every product here sells another run at another problem. None of them remembers the last one.
A session, start to finish
Three pictures. That is the whole product.
1 You talkHe asks once and goes quiet. The pause after your sentence is deliberate — it is where you think.He asks once and goes quiet. The pause is where you think.
2 He listens for the shapeNot for keywords. For the parts an interviewer scores: the situation, your decision, the number at the end.For the parts an interviewer scores: situation, your decision, the number.
3 A cell turns solid — or does notSay “we decided” to a direct question and nothing is credited. On purpose: everyone is sure their story is about ownership.Say “we decided” to a direct question and nothing is credited. On purpose.
No screenshot on this page was drawn. They are taken from the running app.
Three rounds, three ways to fail
What goes wrong — and what he does about it.
Each round breaks you differently. Here is the break, and here is the fix.
Algorithms
You solve it. They still say no.
You get the right answer, but you go quiet while you think. Nobody sees your reasoning. You start typing before you can say why the code is correct.You get the right answer but go quiet. Nobody sees your reasoning.
What he does
He will not take your code until you say the invariant in one sentence. Help comes in six small steps, and the step is picked by rules, not by the model — so you never get the answer by accident.No code until you say the invariant. Help comes one small step at a time.
System design
You know the parts. You cannot put them together.
You can name a cache, a queue, a shard. In the room you jump to boxes and arrows, skip the requirements, and cannot say what you traded away.You can name the parts. In the room you skip requirements and jump to boxes.
What he does
The model answer stays locked until you say yours out loud and commit to it. You draw on a board; he draws back what he heard, so you can see how you were understood.The model answer stays locked until you commit to yours out loud.
Behavioral
Your story is good. It is scored as weak.
You tell a real story from your work. They hear “we decided” where they needed “I decided”, and a result with no number in it.They hear “we decided” where they needed “I decided”, and a result with no number.
What he does
One hard stop: he does not move on until you name a decision as yours. A cell on your map turns solid only when a session actually heard it — tagging it yourself is not enough.He does not move on until you name a decision as yours.
Every fix above is in the app today. Nothing here is a plan.
Where this fits
AI mocks are free now. That's the wrong thing to be selling.
So we do not sell it. Run as many mocks as you want. The difference is what is still there tomorrow morning.
Gadflyfree · your key
interviewing.io$0 (AI)
Hello Interview$79 / year
mockingly.ai$0 · $14 / mo
LeetCode Premium$159 / year
Human mock$100–350 / session
The interviewer speaks back
yes, and you can cut in
listens only, replies in text
no
yes
no
yes
Shared whiteboard
both of you draw
no
yes
no
no
yes
Remembers your last session
a journal per session
no
no
no
no
only if you rebook the same person
Feedback points at your own words
quote + timestamp
rubric score
rubric score
no
no
“go deeper”
Help rationed in steps, by rules
six-step ladders
no
no
no
no
depends on the person
Repeats scheduled between sessions
spaced, and on new problems
no
no
no
no
no
Your voice leaves your machine
never — speech is local
uploaded
n/a
uploaded
n/a
recorded
Checked against public pricing pages and shipped product code in August 2026. Something wrong or out of date? Tell us and we'll correct it.
The product is free. Here is what it actually costs to run.The product is free. Here is what it costs to run.
It runs on your machine and calls the model with your own API key, so the bill is yours and it is tiny. Nobody pays for someone else’s GPU, so nobody has to charge for one.
free
The app
forever, all three tracks
$0.09
What a session costs you
you pay the model, not us
9 GB
What you need
of disk, plus an API key
Listening and speaking cost nothing — both run on your machine. Set a daily limit, say $2, and it stops there. What it needs, in detail →
What eight weeks look like
It does not get easier. It stops being wasted.
The hours are still yours to put in. What changes is where they go.
Week 1
You find out what actually breaks.
It is usually not what you thought. Most people guess “I need more problems”. The journal says something narrower, and quotes you saying it.Usually not what you thought — and the journal quotes you saying it.
Week 4
The same slip stops coming back.
Three clean sessions in a row and it clears from your plan. Then the next one surfaces, and it is a smaller one.Three clean sessions and it clears. The next one is smaller.
Week 8
The room is no longer a surprise.
You have said your reasoning out loud forty times. The interview is the forty-first — that is the whole trick, and there is no shortcut in it.You have said your reasoning out loud forty times. The interview is the forty-first.
No score, no percentage, no “you are 78% ready”. Those are guesses. This is a journal.
The work is still yours. It just stops being wasted.
Pick a goal. Pick a track. Do one session tonight. Tomorrow your plan is different, and you will know exactly why it changed.Pick a goal, pick a track, do one session tonight. Tomorrow your plan is different.
