Skip to content

Free while SkyStudy is in beta

Interview Room: pilot interview practice, out loud

A technical oral and a standard airline interview, asked out loud and answered out loud, then marked against the points a complete answer contains. No score, because there is nothing honest to score you against.

Start with the 9 competenciesSee a marked answer

Open now and free while SkyStudy is in beta. The spoken rooms are not open yet, and this page says which is which.

12
questions in a session
117
questions written by a person, with model answers
0
scores, marks or predictions in a report

Open now, and free while SkyStudy is in beta

Three of the five things the Interview Room holds are open to everyone today, with no account and no payment. They are the material the two spoken rooms are built on, so reading them is preparation in its own right rather than a preview of something else.

The 9 EASA core competencies, and what to ask back

All 9 of them, each with the EASA description and, underneath it, what an interviewer is actually listening for when they ask about it. The angles are the useful half: leadership and teamwork is the hardest one to evidence honestly, because the tempting story is one where you were right and somebody else was wrong; workload management is really a question about what you dropped, so a candidate who says they did everything has not answered it.

The list is sourced rather than asserted, and the page says where from: EASA's own Flight Examiner Manual, Module 1 General, section 15.3.2, read on 2026-08-01. That matters because this site published a different set of ten for months, and the wrong version is still in print elsewhere.

The same page carries 10 questions worth asking the panel at the end, each with what the answer would tell you, including what a poor answer to it would mean. The questions you ask are scored in most airline interviews and a candidate with none reads as one who did not prepare.

Read it

Airline preparation, from profiles a human has verified

What a verified employer profile records about an operator: the hiring stages in order, what the operator says about its interview, what it says about its sim assessment, a typical end to end timeline in weeks, the fleet and the bases, each with the source it came from.

One rule governs the whole of it and it is enforced in the query rather than in an intention: an unverified profile produces nothing. Not a thinner page and not a page with a caveat. The lookup filters on a published profile with a verification timestamp, and a slug that misses renders not-found, because a generic filler page under an airline's name is worse than no page at all. And where a profile records nothing about a stage, the page says so rather than filling the gap, which is why these pages are short.

Open it

The sim assessment guide

What the day is, what is being looked at, and what preparation actually moves the needle. Building a browser simulator was considered and dropped, because a browser cannot reproduce the thing a sim assessment measures, which is how a pilot's capacity holds up with a control column in their hands and somebody watching. A toy version would teach a candidate the wrong lesson about their own readiness.

The guide is written operator-agnostic on purpose. It describes the general shape of the exercise and flags every part that varies, because sim details vary by operator, aircraft type and country. It names no airline and no screening product.

The most useful single idea in it is that the recoverable errors are not the ones that hurt you. A heading that wanders five degrees and comes back is a normal day. A heading that wanders five degrees while you say nothing, followed by a second thing you also do not mention, is the pattern that ends assessments.

Read it

Not open yet: the two spoken rooms

This part opens when SkyStudy leaves beta. It is the one piece that costs money to run every time somebody uses it, so it is not something the project can give away while it has no revenue. Everything else on this page is open now.

The spoken technical oral

EASA theory asked out loud and answered out loud, 12 questions in three timed parts, marked against a model answer written in advance by a person. You pick the subject for the middle part; the first and last parts range across the syllabus, because a candidate who could choose the whole interview would practise the subject they are already comfortable in.

Read what it covers

The spoken airline interview

The opening set every interview uses, then behavioural questions on the EASA core competencies answered with STAR, then the close and the questions you ask them. The report quotes your own words back at you, so a rehearsed paragraph is visible as one.

Read what it covers

What 12 questions in three parts actually looks like

Technical oral

12 questions, about 22 minutes of clock

Part 1: opening questions3 questions, 5 min
Three questions across the syllabus, of the kind a panel opens with. Answer in full sentences and say why, not only what.
Part 2: your chosen subject5 questions, 9 min
Five questions inside the subject you picked, going deeper as the part goes on. Expect to be asked why an answer is what it is.
Part 3: applying it4 questions, 8 min
Four questions that put the theory into an operational situation. There is rarely one right answer; say what you would do and what would change it.

Airline interview

12 questions, about 22 minutes of clock

Part 1: about you4 questions, 6 min
Motivation and self-awareness. These are the questions the panel has heard a thousand times, which is exactly why a vague answer costs so much.
Part 2: competency and CRM6 questions, 12 min
Behavioural questions on the EASA core competencies. Answer with STAR: the Situation, your Task, the Actions you took, and the Result. Use I rather than we when describing what you did.
Part 3: the close2 questions, 4 min
How you finish, and what you ask them. A candidate with no questions reads as one who did not prepare.

How an answer is marked, and why there is no score

There is no score, and that is the central decision in the product. A number would have to come from somewhere, and there is no external scale for how good an answer about V1 was, so any figure invented here would be a figure only this site believed, varying between runs, presented to a candidate as a measurement.

What replaces it is coverage against a written model answer. Every question in both banks carries its key points, written by a person before anybody was ever asked it. The report says which of those points your answer reached and which it did not, quotes you, and names anything you said that was actually wrong. That is a checklist rather than a verdict, so two runs agree far more often than a graded one would, and where they disagree you can read the key point and decide for yourself.

The distinction the report works hardest to keep is between a missed point and a correction, because a pilot needs to know which one they did. A missed point is something you did not say, and it goes in the coverage. A correction is something you said that is wrong, and it is stated in one sentence with what is actually the case. An empty corrections list is normal and common, and the grader is told not to manufacture one to look thorough.

The evidence is a verbatim quotation from you, one sentence or two, copied from the transcript. Never the examiner, never a paraphrase presented as a quotation, and never a quotation assembled out of separated fragments. Where an answer contains no sentence worth quoting, it quotes the most representative one anyway, because reading your own thin sentence back is itself the feedback.

Each question ends on one action line, addressed to you, and it has to be actionable in preparation rather than in an aircraft. Revise the relationship between load factor and stall speed until you can state it without thinking is an action. Be more confident is not. Know your systems is not.

What is enforced in code rather than asked for in words

All of that is instruction to a model, and an instruction is not a guarantee. So the guarantee is code. Every report passes a normaliser before anyone can see it, and it refuses rather than repairs, because a repair would leave the surrounding sentence saying something the model meant and this product did not.

Start with the part that is structural rather than watchful: there is no field for a score anywhere in the shape the model has to answer in. It carries coverage, corrections, a quotation and an action, and nothing else, so a mark is not something the report could render even if the model wanted one. What is left is prose, and the prose is refused outright if it says you will pass or fail, that anything is guaranteed or certified, that your score, mark or grade is anything, that you passed or failed this interview, or that you are ready for the real one. That second half is a list of phrasings and not a reading of meaning, which is the honest limit of it. It drops feedback on a question that was never asked, which is what a model does when it invents one. It ignores a key point index that does not exist, which would otherwise render as a blank bullet. And it recomputes the missed list from the covered list on the server rather than reading it back from the model, so the two cannot contradict each other and the count underneath is arithmetic this site did.

A skipped question gets no feedback at all. Not a gentle note, not a placeholder: nothing. A skip is a smaller sample, never a wrong answer, and inventing feedback for silence would be the worst thing the report could do.

One answer, marked

The fastest way to know whether this is worth anything to you is to read a marked answer rather than a description of one. Everything below is written for this page as an illustration. It is not a question from either bank and it is not anybody's real session: no candidate's answer is ever published anywhere, and there is no screen on this site that shows one person's interview to another.

Sample, written for this pageAirline interview, part 2, competency

The question

Tell me about a time you were under more pressure than you expected, and what you did about it.

The answer

We were on the third sector of four into a field I had not been to before, and the crosswind had gone from forecast twelve knots to a reported twenty five gusting thirty two. My task was the approach briefing and I was still working through the go-around gradient when we were given a shortcut that put us high. I said out loud that I was behind, asked for a delaying vector rather than trying to catch up on the descent, and told the captain I was going to stop working the performance figures and fly the aircraft while he read the plate. We took the vector, I re-briefed in about ninety seconds, and I made the approach normally. What I dropped was the cabin call, and I only remembered it once we were established.

Key points reached: 4 of 5

  • reachedSets the situation in two or three sentences and no more
  • reachedSays I rather than we when describing your own actions
  • reachedNames what you actually said or did at the moment it mattered
  • reachedSays what you dropped or deferred, not only what you completed
  • missedEnds on a result, and on what you do differently now
Evidence
I said out loud that I was behind, asked for a delaying vector rather than trying to catch up on the descent.
Corrections
None. Nothing in the answer is factually wrong, and an empty corrections list is the normal outcome on a competency question.
Next time
Finish the story. You stopped at the approach and never said what the outcome was or what you would do about the cabin call now. Add two sentences at the end and this answer reaches every point.

That is the shape a report entry takes, and it is worth reading twice for what is not in it. There is no mark, and no comparison with anybody else. And there is no opinion about the story itself: the room has no way of knowing what happened on that approach, so it marks whether the story was told in a way an interviewer could follow and could score.

Notice which point was missed and why it is the one that costs. Four of the five were reached inside the first two thirds of the answer, and the answer then stopped at the decision. A panel scoring this hears a candidate who can describe a problem and cannot describe an outcome, which is the commonest way a genuinely good story loses marks. That is a fixable habit, and it is fixable only if somebody names the missing sentence rather than telling you to be more structured.

What you can be asked

117 questions are written down: 81 technical oral prompts and 36 airline interview questions, each with its model answer and its key points. On top of those sit 10 questions to ask the panel and 18 written lines the examiner falls back on: depth probes for when a question's own follow ups are exhausted, STAR probes for a story that stops before its result, repair lines for a candidate who has dried up, and one last line that exists so no turn can ever arrive silent.

The technical bank covers 9 subjects rather than all fourteen ATPL ones, and that is a decision rather than a gap. It leaves out general navigation, radio navigation and the two communications papers, which almost never appear in an airline technical oral, and human performance, which this product asks about in the airline interview as behavioural questions on the 9 core competencies rather than as theory. Padding the technical bank with any of them would make the room's own subject picker offer practice nobody needs.

The airline bank carries exactly two questions per competency across the 9, each written against that competency's own interview angle so a candidate who reads the reference and then sits the room does not meet two different stories. It contains no technical question, no scenario, and no named operator: everything in it is written to work for any airline.

Nothing in either bank was written by a model, and nothing at run time asks a model to write a question. That is not a stylistic preference. Every line the examiner can speak has to exist before the session starts so it can carry pre-rendered audio, and a line invented at run time arrives as silence on a device with no speech voice. The follow up you get is chosen by rule, from the ones already written on the question you just answered or from those fallback lines, and the one it picks is the one your answer has covered least.

The guidance on each question describes what a strong answer does, and never supplies sentences to memorise. That is the product rather than a nicety: a candidate who recites a supplied answer sounds like somebody reciting, which is a worse position than sounding unprepared.

The 9 subjects the technical oral draws from

010 Air Law
9 questions
021 Airframe, Systems and Powerplant
10 questions
022 Instrumentation
9 questions
031 Mass and Balance
8 questions
032 Performance
9 questions
033 Flight Planning
8 questions
050 Meteorology
10 questions
071 Operational Procedures
8 questions
081 Principles of Flight
10 questions

Your voice, your words, and where they go

You answer out loud and your own browser turns your speech into text, so SkyStudy never receives audio of you, could not produce a recording of you if it were asked for one, and stores none. What reaches this site is the text your browser produced, and you can see it and correct it on screen before you submit it.

One thing about that is not this site's to control and is worth saying plainly: most browsers do that recognition on their maker's servers rather than on your own device, so the audio can leave your device and go to your browser's maker under that maker's terms. It does not pass through SkyStudy and SkyStudy never sees it. If you would rather no audio left your device at all, type instead. Typing is always available, it is marked identically, and does not use the microphone. Submitted text is sent to SkyStudy and its AI service for feedback.

Typing is not a lesser path in any other sense either. Firefox has no built-in speech recognition, on desktop and on Android, so a Firefox user types, and the room says so before the first question rather than at question one. Your own device will still dictate straight into the box if you want it to.

Your answers are stored on your account so you can read the session back, and they are deleted automatically after twelve months. The two tables holding them are restricted to their owner for ordinary authenticated database access. Privileged backend and database access remains possible for necessary operations or support; there is no staff screen that displays another person's interview. Deleting your account deletes every transcript and report in the same request, and your transcripts are in the data export if you want your own copy first.

The report is written by an AI, and here is exactly where it sits

The written report is produced by a large language model, and this page says so here rather than leaving you to work it out. The room says it too, on the screen where the report appears, and the report carries the statement in its own text: this is AI-generated practice feedback on one practice session, it is not an assessment, it is not a mark, and it cannot predict how any real interview or selection process will go.

Nothing else in a session involves a model. There is exactly one model call and it happens once, at the end, on the finished transcript. Every question, every follow up and every repair line is written in advance and selected by rule, so the interview you sit is not improvised and the examiner's words are the same words a person wrote.

What is sent, at that one moment, is the text of the questions you were asked and your answers to them. Your name and your email address are not sent with it. It goes to Anthropic's Claude API, which the privacy policy names, and to nowhere else. Nothing is ever sent to an airline, a recruiter, a training organisation or an examiner.

And where it says you missed a point, check that against what you actually said: the same answer read twice can be judged differently on a borderline point, which is why the report shows you the key point and your own sentence instead of a number.

Privacy policyTerms

The limits, before you ask

Every number below is enforced on the server, so a client that has been edited, a replayed request or a tab left open overnight cannot move any of them.

40

sessions per account, for life

Across both rooms combined, not 40 in each. The cap bounds a runaway rather than rationing the product: a session's model call costs a few euro cents, so the whole allowance is a small fraction of the price.

12

exchanges in a session

Three parts, each with its own clock, and the clocks are enforced on the server too. A part ends when its time is up: an answer that arrives just after the bell is still recorded and still marked, but the part does not get to ask one more question.

3

answered questions before a report exists

Below that the honest outcome is the session handed back rather than feedback, and that is what happens. Three answered questions is three real pieces of feedback; two is somebody whose laptop died being charged a session for almost nothing.

0

sessions lost to a failed report

If the model call fails, is refused, or the process dies between the answer and the report, the session is returned to your allowance and the room tells you so. That is a database write and not an intention: the same column the allowance counts is cleared, and a crash is covered by the same path as a polite failure.

The planner looks at your last 8 sessions and prefers questions you have not seen. Against 81 oral prompts and 36 interview questions that is a preference rather than a filter, so it degrades gracefully instead of running out. An answer is capped at 6,000 characters, which is far longer than anybody speaks in one turn.

Who it is for

  • You have an interview or an assessment day in the diary, or you expect one this season, and the thing you cannot rehearse alone is answering out loud while somebody waits.
  • You passed the ATPL exams as multiple choice and have never had to say any of it in your own words. That is the gap the technical oral exists for, and it is the one most pilots say nothing prepared them for.
  • You have written your stories down and want to know whether they land: whether you say I rather than we, whether you reach a result, whether you name what you dropped.

Who it is not for

  • You want to know what a named airline asks. Nothing here will tell you, on purpose.
  • You want a score, a percentile or a readiness verdict. There is none, and the report is built so it cannot produce one.
  • You are earlier in the path and have no CV yet. The free pilot CV maker and the aptitude practice come first, and both are linked at the foot of this page.
  • You want the free reading rather than the practice. The pilot interview guide covers the STAR method and the competency framework end to end, free and with no account, and it is the better place to start.

The price

EUR 79

Planned. Not available yet. Paid once, not a subscription.

That is the price this will carry when it opens, stated now so it is not a surprise later. There is no date, because nothing here knows when the trading registration will be approved, and there is no discount, no countdown and no early-bird rate. There is no card field anywhere on this site today and no payment step exists yet.

Free while SkyStudy is in beta, with no card and no trial. As individual tools mature some of them move to paid plans, and nothing that is free today starts charging without notice being published on this page first. Use while it was free is never billed retroactively.

Planned prices are shown in EUR. Final pricing, applicable VAT, and billing terms are confirmed when paid access launches; see terms.

Refunds and your 14-day withdrawal right

Nothing on this page is for sale today, so there is nothing to refund today. The terms are published in advance anyway, so they are readable before any of this goes live rather than after.

We treat the Interview Room as a hosted online service supplied over time, including generated sessions and account features. If you are a consumer in the EU you have 14 days from the contract date to withdraw without giving a reason. The payment step asks you, in an unticked box, to start the service during that period. The request does not give up the right. If you withdraw after service starts, applicable law may allow a charge proportionate to what was supplied.

There is also a voluntary money-back guarantee for a first purchase inside 14 days, offered once per customer. On that first purchase the refund is normally self-serve and immediate, with no email to write. Other requests go to a person under any statutory right and the published guarantee. The full wording, deadlines, model withdrawal form and complaints route are on the refunds page.

Refunds and withdrawal rights

What the Interview Room is not

It does not predict anything. Not whether you will pass, not whether you will be invited, not whether you are ready. The grader is told it has no basis for such a statement and that it is not lawful for this product to make one, and the normaliser rejects the report outright if one appears anyway.

It does not tell you what a specific airline asks. The airline pages state what a verified employer profile records and nothing else, and both question banks are written to work for any operator. A page claiming to hold one company's actual interview questions is either wrong or is repeating something a candidate should not have shared.

It does not assess how you sound. The rubric tells the grader to mark a typed answer and a spoken one identically, and not to remark on which it was. Where a word is obviously a mis-transcription it marks what you plainly meant; where it is genuinely ambiguous it neither credits the point nor corrects you for it.

It is not a simulator, and the sim assessment guide is a guide. It is not a coach either: it will not write your answer, rehearse a script with you, or tell you which of your stories to use. And it is not an assessment of anything. It is practice, and the report says so in its own last paragraph rather than in a footnote on a policy page.

Frequently asked questions

Do I need this to prepare for a pilot interview?

No. The pilot interview guide is free, needs no account, and covers the STAR method and the EASA competency framework end to end. This is for the part reading cannot do, which is answering out loud, under a clock, and then being shown which of the points a complete answer contains you actually reached.

What can I use right now?

The competency reference, the verified employer profiles and the sim assessment guide. All three are open to everyone while SkyStudy is in beta, with no account and nothing to pay. The two spoken rooms are not open: each finished session ends in a call to a paid model, so it is the one part the project cannot give to every account that exists while it has no revenue.

What will it cost when it opens?

The planned price is EUR 79, paid once, not a subscription. It is not available yet and there is no date, because nothing here knows when the trading registration will be approved. Nothing that is free today starts charging without notice being published first.

Is the feedback written by an AI?

Yes, and the report says so on its own face. One model call is made per session, once, at the end, on the finished transcript. The questions, the follow ups and the repair lines are all written by a person in advance and chosen by rule, so nothing you are asked during a session is generated.

Does it give me a score or tell me if I am ready?

No, and it is built so that it cannot. The report lists which of the key points a complete answer contains your answer reached, quotes you, names anything you got factually wrong, and gives one thing to do differently. There is no field for a mark in the shape it comes back in, and on top of that a report whose words predict an outcome, award a grade or say you are ready is rejected before you would see it.

Is my voice recorded or uploaded?

SkyStudy never receives audio of you and stores none: your browser does the transcribing and only the text reaches this site. Most browsers do that recognition on their maker's servers, so the audio can leave your device on a path SkyStudy is not part of and never sees. Typing is always available, is marked identically, and does not use the microphone. Submitted text is sent to SkyStudy and its AI service for feedback.

What happens if the feedback fails to arrive?

The session is given back to your allowance and the room says so. The same applies if you answer fewer than three questions: below that the honest outcome is the session returned rather than feedback drawn from almost nothing.

Will it ask me the same questions twice?

It prefers questions you have not seen in your last 8 sessions. With 117 written questions behind the two rooms that is a preference rather than a hard filter, so it degrades gracefully rather than running out of things to ask.

Does it tell me what a particular airline asks?

No. The airline pages state what a verified employer profile records about how that operator hires, with the source for each, and nothing else. Both question banks are written to work for any operator, and no page here names a screening or aptitude product.

How long does a session take?

12 questions in three parts. The technical oral runs 22 minutes of clock and the airline interview 22, and each part's clock is enforced on the server, so finishing early is possible and running over is not.

Next steps

Pilot interview guideFree, no account: the STAR method and the competency framework end to end.Free pilot CV makerThe document the interview is about, built in your browser with nothing uploaded.Pilot jobs boardThe vacancies open now, which is where a prepared interview is aimed.

Start with the 9 competencies. They are what the room marks against.

Every behavioural question in the airline interview is written against one of them, and the report marks your answer against what a strong answer to that competency does. Reading them is free, needs nothing from you, and is the preparation most candidates skip.

Read the 9 competencies

Optional analytics measures site use and campaign visits, with Vercel, Plausible when configured, and Sentry browser error monitoring. Login and security remain available if you reject. Your choice is saved on this device for up to 180 days. Cookie policy.