They said no. You still don’t know why.

Gadfly is a voice tutor for technical interviews. You think out loud; it asks the next question instead of handing you the answer — and it remembers what broke last time.

Free while it is early · a practice room, not an earpiece · your voice never leaves your machine

ACT I · THE SILENCE

You solved it. In silence.

what you did✓ all tests pass

      
what the room heard 00:00 spoken

The room scores what it hears.
It heard nothing.

1 in 5
That is how many engineers perform the same way twice. The same person can pass one interview and fail another on the same day. interviewing.io, across thousands of interviews
“the feedback primarily boiled down to a very repetitive you need to go deeperBlind · on a $100 mock with a Google L7
“My brain is straight mush. Concepts blending together, learn 1 new thing, push out anotherBlind · thread “Anyone else’s brain fried?”

ACT II · 399 BC · ATHENS

Socrates never answered a question. He asked better ones until the answer was yours.

Athens called him the gadfly — small, persistent, impossible to ignore, stinging a drowsy city awake. The city put him to death for it. The method survived twenty-four centuries.

Now the method has a microphone.

GADFLY

/ˈɡad·flʌɪ/  n.  a persistent, annoying questioner — one that provokes others into thought.

ACT III · 11:42 PM · TONIGHT

Nothing moves until the plan is said out loud.

That is the first rule of the room, and it does most of the work: a plan that sounds fine in your head rarely survives being spoken. The gaps come out where an interviewer would find them — before the interviewer does.

Keep scrolling — the session plays as you go ↓

M Session with Mia mic on
Well, night owl or not — glad you’re wide awake for this. Open the statement on the left, take your time, and tell me if anything’s unclear.
i think i move two pointers from both ends and compare the sum to the target
Good — we’re ahead of ourselves a little, but let’s run with it. Suppose the sum at the two ends is smaller than target. What do you do, and why would that help?
move the left pointer right?? because the array is sorted… that’s the only move that can make the sum bigger
That sentence is the invariant — the thing that stays true at every step. Now check it still holds after the pointer moves.
Mic is on — speak or type… I’m stuck (Ctrl+Space) Tell me the answer

THE BUTTON · TRY IT

There is a button that says Tell me the answer.”

Go ahead. Press it.

The answer does not come. The question you were dodging does.

Answers given: 0 · Questions you got instead: 0

A tutor who caves is a search engine with manners — and you already have one of those.

ACT IV · THE MEMORY

It remembers what broke.

It listens while you think, then names the one thing that went wrong — in your own words, with the minute it was said. Free AI mocks forget you the moment you close the tab. The memory is the product.

what you got wrong comes back on
day 1day 3day 7day 14

Spaced repetition — applied to reasoning instead of vocabulary.

nothing closes on one good day
clean runclean runclean run

A mistake clears after three clean runs in a row, not one. One good evening is luck; three is a change.

real sessionapp.usegadfly.com
The end of a session: the one thing that went wrong, quoted in your own words
A real wrap-up, from a real session — nothing on it is drawn. The badge means the same slip came back twice in a row; that is why it is on tomorrow’s list.

8:00 AM

By morning, your plan is different.

Not because you filed anything. The wrap-up wrote it when the session closed: which pattern slipped, which drill it spawned, which day it comes back. Session twelve starts where session eleven found the crack.

real sessionapp.usegadfly.com
Tomorrow’s queue, with every item traced to the session it came from
A real morning queue: every line says why it is there — the pattern, the drill, the day it returns: 1 · 3 · 7 · 14.

ACT V · THREE ROUNDS

Each round breaks you a different way.

Not one of them is testing what you know.

ALGORITHMS

You solve it. They still say no.

You get the answer right, but you think in silence. The room scores what it hears — and it heard nothing.

WHAT HAPPENS INSTEADIt will not take your code until you name the invariant — the one thing that stays true at every step.

SYSTEM DESIGN

You know the parts. You cannot put them together.

You can name a cache, a queue, a shard. In the room you draw boxes before you ask what the system must do.

WHAT HAPPENS INSTEADThe model answer stays locked until you say yours out loud. You draw on a board; it draws back what it heard.

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.

WHAT HAPPENS INSTEADOne hard stop: it does not move on until you name a decision that was yours. It asks for a number in the result too — and would rather hear “we never measured it” than a number you made up.
real sessionapp.usegadfly.com
Six axes scored against your own rating of the session
The end of an algorithms session, for real: six axes anchored by deterministic rules over the session journal — a model may nudge a score only when it cites your own recorded words, and uncited claims are thrown out. The gap is between your number and its.

Every screenshot on this page came from a real session. Everything above is in the app today. Nothing here is a plan.

ACT VI · THE ARC

Eight weeks. Same hours. Different engineer.

The hours are still yours to put in. What changes is where they go.

TONIGHT

A sound check and a first named slip.

One session in, the wrap-up quotes you saying the thing that broke. It has a name now.

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.

WEEK 4

The same slip stops coming back.

Three clean runs in a row and it clears from your plan. Then the next one surfaces, and it is a smaller one.

WEEK 8

The room is no longer a surprise.

By then you have said your reasoning out loud more times than you can count. The interview is one more of those.

No score, no percentage, no “you are 78% ready”. Those are guesses. This is a journal.

ACT VII · THE MATH

A mock with a real engineer runs $100–$300 an hour.

$0 one hour, one human mock — then it forgets you
vs
a whole evening here — and it remembers everything

66¢ · one evening

The whole bill for an evening on our key: the model doing the thinking. Speech in and out add nothing — they run on your machine and never leave it.

$0 · today

No card and no meter: while the product is early, sessions run free on our key. That is the entire pricing page right now.

one ceiling · out loud

Live sessions have a daily cap. Past it the tutor stops and says so — it does not quietly turn into a template machine.

000 bytes of audio leave your machine. Speech recognition and the speaking voice both run locally. There is nowhere for the audio to go.

The loop is a performance. A “no” freezes you out for six to twelve months.

The work is still yours. Start tonight.

No card — about 2 minutes and a microphone. Your voice stays on your machine.